An Expert System for Automatic Relay Performance Analysis

(1)

Degree project in

Performance Analysis

SHISONG, GAO

Stockholm, Sweden 2012

XR-EE-ICS 2013:006 ICS Masterthesis,

(2)

(3)

Acknowledgments

First of all I want to thanks all the people who make this project possible,specially Prof. Lars Nordstrm and Dr. Murari Mohan Saha,who give me this precious oppor- tunity. Then I want to express my deep and sincere gratitude to all my supervisors.

In the ABB Hans Jernberg, Stig Lidstrm,Torbjrn Einarsson. In ICS department Ar- shad Saleem,Nicholas Honeth. You help me with all your heart. Thanks you for your knowledges and endless patiences.Nothing would accomplished without your help. I also want to thanks my programme mates also close friends Davood babazadeh, Zhao pengcheng, Chen zhenwei. Thanks for you accompany and encourage when i need them the most. Finally, I wish to thank my parents for all their love and support.

(4)

Table of Contents

Abstract i

Dedication ii

Acknowledgments i

List of Acronyms iv

List of Tables v

List of Figures vi

1 Introduction 1

1.1 Introduction . . . 1

1.2 Background . . . 1

1.3 Project Objectives . . . 2

1.4 Initial Motivation of the project . . . 2

1.5 Project Plan . . . 3

1.6 ESARP System . . . 5

2 Theoretical and background knowledge survey 6 2.1 Literature review . . . 6

2.2 RTDS relay testing environment . . . 6

2.3 RTDS relay testing system model . . . 7

2.4 Relay model study . . . 8

2.4.1 Full scheme measurement . . . 8

2.4.2 Impedance characteristic . . . 9

2.5 COMTRADE Files . . . 11

2.5.1 Header Files . . . 12

2.5.2 Conﬁguration File . . . 12

2.5.3 Data File . . . 12

3 Concept of model development 13 3.1 Previews expert system concept . . . 13

3.2 General Structure for the system development . . . 14

3.3 Functional Requirement of the system . . . 15

3.3.1 Multi-ﬁle reading . . . 16

3.3.2 Signal decomposition . . . 17

3.3.3 Dynamic performance of relay calculation . . . 17

(5)

3.3.4 Identiﬁcation of diﬀerent protection zone . . . 18

4 System design 20 4.1 Detail steps of the system . . . 20

4.2 Develop environment and component selection . . . 22

4.3 Multi Files Reading . . . 22

4.4 Relay Model transfer . . . 26

4.5 Signal Decomposition . . . 27

4.6 Matlab Java Builder . . . 28

4.7 Logic steps in Expert environment . . . 30

5 System function result 31 5.1 Analog signals channel plotted . . . 31

5.2 Impedance dynamic performance with corresponding time scale . . . 31

5.3 Automatic saved value of resistance and reactants . . . 36

5.4 Evaluate from Expert advice . . . 36

6 Future work 37 6.1 User friendly interface . . . 37

6.2 Robust rule based system . . . 37

6.3 GUI function for adoptive domains deﬁne . . . 37

6.4 Optimized parameter resetting for executing analysis . . . 37

7 Conclusions 38

8 Discussion 39

Bibliography 40

(6)

List of Acronyms

IED Intelligent Electronic Device CT Current Transformer

VT Voltage Transformer

SAS Substation Automation Systems

ESARP An Expert System for Automatic Relay Performance Analysis RTDS Real Time Digital Simulator

IEC International Electrotechnical Commission GOOSE General Object Oriented Substation Event

COMTRADE Common format for Transient Data Exchange for power systems PLL Phase-locked loops

(7)

List of Tables

2.1 Parameters in zone 1 setting . . . 10 4.1 tasks for diﬀerent roles within java builder . . . 29

(8)

List of Figures

1.1 General structure of expert system . . . 2

1.2 Diﬀerent zone indication for the fault analysis . . . 3

1.3 Project model . . . 4

2.1 RTDS testing environment . . . 7

2.4 Relay protection domain deﬁne . . . 9

2.5 Fault loop model . . . 11

3.1 Concept of structure of expert system . . . 14

3.2 Closed loop testing of a physical relay . . . 15

4.1 Detail steps of the build up expert system . . . 21

4.2 Test circuit of relay model in RTDS . . . 26

4.3 Signal process in the relay model . . . 27

5.1 Analog signals channel plotted . . . 31

5.2 Impedance dynamic performance with corresponding time scale . . . 32

5.3 0 percent of line value . . . 33

5.6 Corresponding time scale . . . 33

(9)

Chapter 1 Introduction

1.1 Introduction

Modern electrical power system is a very large scale of complex and interconnected system that the small disturbance easily would create havoc in power system stability and sustainability without operation of power system protection.[1] In the analysis of power system protection the protection relay IED is the crucial part in order to prevent the unexpected event and maintain the integrity and security of the system.

It is very useful at identifying problems with the fault relay settings or the algorithm of protection which could cause undesired operation and system disturbances. [2]By replaying the recorded fault waveform which generated from the testing devices, the unexpected performance of the relay can be changed through the resetting of the parameters.[3] Such analysis is complicated and expertise knowledge is usually required. Through such test running time of testing devices would usually demand very high cost. At the same time analysis of recognize and categories the faults to determine fault location and record data about fault level and fault resistance take lots of human resources in order to evaluate the correctness of the relay operation for a fault. The objective of this paper is to verify the performance of the relay IED through post-mortem analysis, the completion of design such system will help with operational decision making while saving working hour as well as the cost of running time of testing devices.

1.2 Background

The project is the collaboration between KTH ICS department and company ABB Substation Automation. The project starts from March 15 2012 officially. The requirements of the project are elicited from the senior experts from aspect of Simulation Engineer, Line protection and Application. The material related with the testing relay and background knowledge is provided from ABB. The following figure shows the general structure of starting point of the project. The domain knowledge including Literature, document and manual will be provided from ABB. The result of testing will be generated from ABB relay. The knowledge and testing result being put into the system after data categorization and calculation should evaluate the relay performance and the expert advises as a result should generated automatically. Such developed system should greatly benefit from enhance work efficiency since it takes long time to analysis all the generated results after relay testing. Also save time from

(10)

the running time of simulator in order to reduce testing cost base on existing ﬁles.

The completion of the system should be highly replicable since the diﬃcult of knowledge heritage from the expertise. Also such system should be easy to further develop into more robust system in order for meeting higher requirement of the automation analysis.

Figure 1.1: General structure of expert system

1.3 Project Objectives

After testing through diﬀerent scenarios the performance of the relay should be evaluated. The correctness of the relay will be judged through comparison of what the relay seen from the fault and what the actual fault is applied in the system. Domain parameter should be reset from certain behavior of the relay. The result generated from such system should indicate such behavior in the format of report through running all the testing ﬁles.

1.4 Initial Motivation of the project

The envisage system should have the characteristic of automatic result report generated and saved through Multi-ﬁles processing of the Comtrade Files. The dynamic relay performance should be calculated and compare with the domain setting to get

(11)

the different expert advises. The following figure shows the initial project functionality: 1. when impedance is within a defined area it is considered an easy shot 2.when it is near another border even if it operates correctly a warning could be given 3.if it is outside a border instruction to increase the border should be given Since the

Figure 1.2: Diﬀerent zone indication for the fault analysis

relay model will be built and the testing result will be calculated in multi-ﬁles way.

Evaluate if manually changed protection parameters will improve the result by using already saved Comtrade ﬁles become possible, so there will be no need to rerun the simulator batch. The cost of the running testing devices could be saved with such manner is another important motivation.

1.5 Project Plan

To illustrate the project model a ﬁgure was drawn with the tollgate:

Project model

As shown in the chart below, it illustrates that the whole project is distributed into 7 phases. Milestone represents the work should be completed in certain time stage, toll-

(12)

Figure 1.3: Project model

gate stands for one correction completed work is accepted by the responsible person.

Next it explains the documents should be delivered in the project.

Knowledge engineering

To develop an expert system, the ﬁrst step is to begin collect the knowledge from e.g., human experts, manuals and documentation. Rules will be derived from the collected knowledge.

Deﬁning a common data structure

After all the knowledge has been collected, the knowledge needs to be examined.

Then data structures will be designed: first, the major concepts are identified; second all the variable characteristics will be listed. Investigation will be made in order to design a common data structure that can accommodate concepts from different data sources such as DFR and IEDs etc. Different approaches will be investigated including Comtrade, XML, Ontologies.

(13)

Testing and Veriﬁcation

For making the system more robust, more modular and better understood. Rigorous test need to apply into every step in system development.

Interface building

To categorize fault types interface building like database access and diﬀerent component integration would be investigated.

Writing the rules

Once the former steps all accomplished the process of writing rules shall begin. It will be easier for programming and testing if the rules divided into separate modules.

Iterative development

During the developing of the system, more information back to the source always needs to be reviews to increase the quality of the work.

Final report

The report which covers all stages of the project, as well the detail functions will be presented.

1.6 ESARP System

An Expert System for Automatic Relay Performance Analysis (ESARP) is designed in the project according to the requirements provided above. ESARP system should perform post-mortem analysis of the fault incidents and relay operation. During such process the raw data should be collecting, grouping or clustering automatically. Then relay operation correctness will be veriﬁed in order to minimize time spend for manual detailed analysis of simulation shots.

(14)

Chapter 2

Theoretical and background knowledge survey

2.1 Literature review

Distance relay have changed dramatically in functionality and operating principles in the last two decades. Microprocessor protective relays nowadays become more and more complex with the pre-programmed control logic and additional functions.[4][5]

In the paper of Testing of advanced distance protection relays typical schemes and functional hierarchy are illustrated then the methodology and tools available for the relay testing are further discussed.[3]In the paper of automated test solution for multifunctional protection relays the automatic validation test set-up are being discussed.

[5]Such test environment is built up under the test component Agile which will generate the test files. Then the test results are extracted and stored in the defined database. Then the test report is generated. While some related model building are done in the paper Modeling and Testing of Protection Relay lED , a matlab based power system model and fault are simulated. Then the matlab based relay program is tested with satisfactory results[6]. Then a PSCAD based faults are simulated with the real relay tested on line. Different test scenario is applied and the type and location of fault detected by the relay IED was compared with the simulated fault. [5]Similarly in the paper Modeling and Testing of a Digital Distance Relay Using MATLAB/SIMULINK a digital distance relay for transmission line protection is modeled and tested under power system block-set and the implement program of distance relaying algorithms are served as the main environment. Difference in testing environment and way of analysis of this paper will be explain into details with following sectors[7].

2.2 RTDS relay testing environment

The relay testing is performing under the real time digital simulator (RTDS) in this paper[8]. Such advanced and effective system can be connected in the closed loop with power system model. Under the controlled and flexible environment the protection component to be subjected to all possible faults and operating condition within the digital simulation. The insight of performance of relay as well as the interactive with the power system will be provided. Proportional secondary voltage and current signal which the protection equipment would see are used through detailed models of transformers instrument such as (CT,PT and CVT).So the effect on the performance of the relay can be evaluated. The primary voltage and current signals

(15)

Figure 2.1: RTDS testing environment

also can be sent directly to the protection component by using appropriate scaling factor. With secondary voltage and current provide to the protection equipment via power ampliﬁers, the protection will respond as connected with actual network. The protection component should react with corresponding fault with trip providing and subsequent reclose signal. Such signal should respond back to the simulated network model since the real time operation. The operation of the mechanical reaction time and the breaker status also should provide through such system. If the protection equipment is IEC 61850 compliant the commands of the breaker can be imported into the simulation by using GTNET-GSE which supports both GOOSE and GSSE messaging.[] The power system is simulated in a real time with a time step of 50100 s.

The appropriate signals are sent from RTDS to the relays and also the output of the relays can be fed back to the simulator. To test the IED stability and investigate the mal-operations, 100-2000 faults and operating scenarios would apply within closed loop protection equipment. Faults and operating scenarios can be run manually from the Run Time console or using the automated batch mode facility. The results can be saved in the RSCAD MultiPlot, COMTRADE, jpg or pdf format. Key data from results (trip times, max/min values, etc.) can be stored in ASCII or Word report ﬁles.

2.3 RTDS relay testing system model

As the ﬁgure shown above, a model of one line model is implemented in the RTDS simulator that includes the high voltage components like transformers CT, VT, lines, breaker and meters. Then the control function and required protection will be applied through such model under testing.

(16)

Figure 2.2: RTDS relay testing system model

2.4 Relay model study

2.4.1 Full scheme measurement

The online testing model of relay which used in the system is ABB REL670. The line distance protection is up to five zone full scheme protection. The fault loops include three phase-to earth faults and three phase-to-phase faults for each of independent zone. Also the individual setting of resistive and reactive of each independent zone is flexible for the user as the backup of transformer connected to the overhead line and different kinds of cables. The suitable application with auto-reclosing is based on independent each fault loop measurement of impedance and sensitive and reliable built in phase selection function. Also the adaptive load compensation algorithm for the quadrilateral function could prevents overreaching of zone1 at load exporting end at phase to earth faults. Such relay can operate independently in directional (forward or reverse) or non-directional mode. The following figure shows the different measuring loops for basic five protection zones. The use of full scheme technique

Figure 2.3: RTDS relay testing system model

compared to switched schemes gives faster operation time[9]. Depending on the fault type a start element is selected with correct voltages and current. With those selected value of measuring current and voltage each distance protection zone performs like on independent distance protection IED.

(17)

2.4.2 Impedance characteristic

Some basic parameters regarding zone1 setting

The distance measuring zone includes three intended phase-to-earth fault loops and three phase-to-phase fault loops. According to the parameter listed above the distance measuring zone will essentially operate. The full loop reach is presented as following ﬁgure shown. The fault loop model with respect to each fault type is

Figure 2.4: Relay protection domain deﬁne

illustrated as following for further calculation. The R1 and X1 stands for positive sequence impedance from measuring pint to the fault location. The actual fault resistances are represented as parameters RFPE and RFPP. The calculation of fault loop use complex values of voltage, current and current changes. The earth return compensation applies in a conventional manner.

(18)

Name Values Values Unit Step Default Description

Operation Oﬀ On On Operation Oﬀ / On

IBase 1 - 99999 A 1 3000 Base current, i.e. rated current

OperationDir

Oﬀ Non- directional Forward Reverse

Forward

Operation mode of di- rectionality NonDir / Forw / Rev

X1FwPP 0.10 - 3000.00 ohm/p 0.01 30.00

Positive sequence reactance reach, Ph-Ph, forward

R1PP 0.01 - 1000.00 ohm/p 0.01 5.00

Positive seq. resistance for characteristic angle, Ph-Ph RFFwPP 0.10 - 3000.00 ohm/l 0.01 30.00 Fault resistance reach,

Ph-Ph, forward X1RvPP 0.10 - 3000.00 ohm/p 0.01 30.00

Positive sequence reactance reach, Ph-Ph, reverse

RFRvPP 0.10 - 3000.00 ohm/l 0.01 30.00 Fault resistance reach, Ph-Ph, reverse

X1FwPE 0.10 - 3000.00 ohm/p 0.01 30.00

Positive sequence reactance reach, Ph-E, forward

R1PE 0.01 - 1000.00 ohm/p 0.01 5.00

Positive seq. resistance for characteristic angle, Ph-E

X0PE 0.10 - 9000.00 ohm/p 0.01 100.00 Zero sequence reactance reach, Ph-E R0PE 0.01 - 3000.00 ohm/p 0.01 47.00

Zero seq. resistance for zone characteristic angle, Ph-E

RFFwPE 0.10 - 9000.00 ohm/l 0.01 100.00 Fault resistance reach, Ph-E, forward

X1RvPE 0.10 - 3000.00 ohm/p 0.01 30.00

Positive sequence reactance reach, Ph- E, reverse

RFRvPE 0.10 - 9000.00 ohm/l 0.01 100.00 Fault resistance reach, Ph-E, reverse

Table 2.1: Parameters in zone 1 setting

(19)

Figure 2.5: Fault loop model

2.5 COMTRADE Files

COMTRADE stands for Common Format for Transient Data Exchange. [10]In this IEEE standard file format contained exchanged medium used which defines fault types, test, or simulation data of electrical power systems. Currently the relay designed using microprocessors are being developed. The relays have the ability to capture and store input signals in the digital form and transmit this data to another device. Each COMTRADE record has a set of up to four files which carries a different class of information. The four files are as following:

* Header

* Conﬁguration

* Data

* Information

All files in the set must have the same filename; the different is the extension that indicates different file types. .HDR for the header file, .CFG is the configuration file, .DAT for the data file and .INF for the information file. The names of the file shall follow the IBM compatible DOS conventions for legal characters within the file name. The file names are limited to eight characters and extension is limited to three characters.

(20)

2.5.1 Header Files

Header file is an ASCII text file created by the originator of the COMTRADE data, usually generated through the word process program. Such data are intended to be read by the users. The file format is ASCII.

2.5.2 Conﬁguration File

The configuration file is an ASCII text file which is intended read by the computer program so such file must be saved in a specific format. The information contained in the configuration file can be used by the computer program to interpret the data in the .DAT file. This information includes items related with sample rates, line frequency, number of channels, channel information, etc. The configuration file can be created with a word processing program or by a computer program that creates the configuration file from the source of the transient recorder. If a word processor is used to generate the file. Such data must be in ASSCII text file format.

2.5.3 Data File

Each input channel for each sample in the record will be saved in the .DAT ﬁle. The number stored for a sample corresponding to a scaled version of value presented to the device that sampled the input waveform. The stored data maybe zero based or zero oﬀset. The zero-based data spans from a negative number to a positive number.

While zero oﬀ-set number are all positive with a positive number stands for zero. Also each sample is corresponded with a sequence number and time stamp. In addition to data represent analog input that indicate on/oﬀ signals are frequently recorded.

These are referred to as digital inputs, digital channels, event inputs, or status inputs.

Usually such status is represented by the number 1 or 0 in the DAT. File.

(21)

Chapter 3

Concept of model development

3.1 Previews expert system concept

An expert system which performs detailed diagnosis of the relay operation is presented from the paper An Expert System for Diagnosis of Digital Relay Operation.

The system use the relay generated ﬁle and report as resource for analyzing. The tragedy is forward chaining reasoning, logic reasoning and backward chaining. The implementation of knowledge through software CLIPS is used for the prediction of performance of relay. Commercial software called RELAY ASSISTANT and a relay test set are used to generate the simulated voltage and current to trigger the relay operation. The actual behavior of the relay and circuit breaker in terms of status and timing of logic operands are recorded which has been deliberately manipulated to introduce the abnormities. As the result a diagnosis report is generated. Both validation information and diagnosis information are listed also the correctness of the analysis is further proved. In the paper Enhancement of Expert System Philosophy for Automatic Fault Analysis ideas for automatic fault analysis system based on oscil- lographic fault recorders are discussed. The purpose of the system is to recognize and categories the fault. Other purpose is to determine fault locations and record data about fault level and fault resistance. The correctness of the fault should be analyzed.

Such system has not been developed yet because of diversity of protection devices and enormous amount of binary indication possible. And due to the suppliers desire to maintain the secrecy of their programming code in case of losing intellectual property.

Such developed system is intended to provide the system operator control personnel with information which helps them with operational decision making. While at the same time some other automatic fault analysis system also explored in this paper.

Yuehai et al. (2004) demonstrate a method of using threshold pickups for several variables for fault type identification. The type of fault identified by a logical table which requires a great deal of predetermined values based on human expertise. How- ever from all recorders need to be synchronized, clustered and categorized in terms of relevance to a particular fault which required exact fault inception time and fault duration need to be resolved. In a paper by Barros, Perez and Pigazo (2003), the authors claim that the Short Time Fourier Transform (STFT) and use Kalman filters to identify disturbance of power quality .Bollen et al. (2007) compare abilities of Ex- pert Systems (ES) and Support Vector Machines (SVM) to classify PQDs. A survey by Ibrahim and Morcos (2002) reveal that Artificial Intelligence methods like Fuzzy Logic, Expert Systems, Artificial Neural Networks (ANN), Genetic Algorithms (GAs) and Adaptive Fuzzy Logic can be experimented with analyses PQDs. A method was

(22)

developed by Kezunovic, Luo (Shanshan) and Sevcik (2002) uses Genetic Algorithms (GAs) to perform fault location in conditions of sparse field recording availability. Ex- amples regarding fault automatic analysis. For instance, Sun, Jiang and Wang (1998) show how Artificial Neural Networks (ANNs) can be used for different time-spaces on the fault event sequence. Fukuyama and Ueki (1991) showed that ANNs can be successfully used to perform waveform recognition and fault identification in order to cluster relevant recording; as long as they are verified by means of a rule based Ex- pert System (ES). Zhang and Kezunovic, 2006 seem to prefer rigorous mathematical solutions for detection, classification and location through synchronized sampling by comparing modal components of currents against thresholds. Unfortunately due to fears of sabotage and direct access to information from protection relays is not usually positive. For that reason, the AFA Systems illustrated above most are incorporate with recorder report instead of relay record [11][12].

3.2 General Structure for the system development

The concept of such server system is that it shall recognize the fault record from the Fault Recorder Master station. This Master station shall operate separately from such expert system. Also master station shall access fault recorders of various types through different appropriate protocol. The fault record obtained from master station shall export and saved in the entry folder of expert system in ASCII COMTRADE format. The various sequences of analogue and digital channel sample data should be recorded. All of recorded files shall be imported and processed automatically by means of signal decomposition, algorithm calculation and Fault classification. All of the data and results of analysis also need to be stored automatically. The following is a general structure of expert system. The further requirement of the system and detailed structure will be illustrate in the following chapters[12].

Figure 3.1: Concept of structure of expert system

(23)

3.3 Functional Requirement of the system

The project is initiated under the scenario of relay testing system execute with RTDS as simulation devices. Through user interface the simulated power system will be set for the testing scenario which is the one line system illustrated above. Then those testing scenario of different line value will be applied through a voltage and current amplifier to the relay. The analog channel value will be recorded and generated as the IEEE standard comtrade files which need further analysis of the human expert. The knowledge of human expert will be used as the reasoning to analysis the correctness of the relay perform. As the result related evaluation decision come out from the expert will be compare with simulated scenario applied from RTDS. If the analysis result is different from what is expected from the simulation system. The parameters of the relay will need further adjust to adopt with the line.

For the testing of the relay usually 100-2000 shots would be applied which mean high cost of testing device if such test need to rerun for checking reset of parameters, also to analysis batch of shots will demand long working hours of human resource. To save the running time of the testing device as well as the working hours of human expertise an automatic system for relay testing under RTDS environment is urgently needed. The ESARP (An Expert System for Automatic Relay Performance Analysis)

Figure 3.2: Closed loop testing of a physical relay

is developed as a post-mortem analysis for the fault which is recorded as ASCII comtrade format. The server system shall perform multi-files analysis of the comtrade files and generate the report files into the final work space. In such system the analysis of newly imported record by means of multi-files reading, signal decomposition, impedance dynamic performance calculation, identification of difference protection

(24)

zone. Also the automatic plotted impedance value and binary results are stored in the same work place after the automatic analysis is performed. After automatic plotted result and recorded ﬁles were saved in the relevant destination, the access of those data is allowed by the testing controller. The users and experts can amend data and analysis formulas through speciﬁc requirements shall be permit.

The requirement of the system will be elicited from the human expert and the document provided from ABB. According to the requirement the system will be designed into separate process. The priority and the importance of the detailed design also will be presented to fulﬁll the requirement of the ESARP system. As the following table shown is the elicited requirement of the system.

3.3.1 Multi-ﬁle reading

ID 3.3.1.1

Requirement The Comtrade ﬁle address shall be au-

tomatically identiﬁed

Priority High

Importance Must

Process Multi-ﬁles reading

ID 3.3.1.2

Requirement Record DAT. File information into

Matlab workspace

Priority High

Importance Must

ID 3.3.1.3

Requirement Calculate the analog voltage and cur-

rent signals through IEEE standard

Priority High

Importance Must

ID 3.3.1.4

Requirement Plotted analog channel information

Priority Medium

Importance Shall

(25)

3.3.2 Signal decomposition

ID 3.3.2.1

Requirement Identify Trigger signal

Priority High

Importance Must

Process Signal decomposition

ID 3.3.2.2

Requirement Flexible time window for decomposed

result

Priority High

Importance Must

ID 3.3.2.3

Requirement Comply with Multi-ﬁles reading func-

tion

Priority High

Importance Must

ID 3.3.2.4

Requirement All three phases analog signals shall be calculated

Priority High

Importance Must

3.3.3 Dynamic performance of relay calculation

ID 3.3.3.1

Requirement Obey ABB algorithm

Priority High

Importance Must

Process Dynamic performance of relay calcula-

tion

(26)

ID 3.3.3.2

Requirement Automatic saved R and X result value

in txt ﬁle

Priority High

Importance Must

tion

ID 3.3.3.3

Requirement The dynamic of impedance value being

plotted

Priority High

Importance Must

tion

ID 3.3.3.4

Requirement Plotted result of impedance value au-

tomatically saved

Priority High

Importance Must

tion

3.3.4 Identiﬁcation of diﬀerent protection zone

ID 3.3.4.1

Requirement Initial Jar component

Priority High

Importance Must

Process Identiﬁcation of diﬀerent protection

zone

ID 3.3.4.2

Requirement Read txt ﬁle with R and X value

Priority High

Importance Must

zone

(27)

ID 3.3.4.3

Requirement Save each 60 sample value into separate array

Priority High

Importance Must

zone

ID 3.3.4.4

Requirement Deﬁne protection domain

Priority High

Importance Must

zone

ID 3.3.4.5

Requirement Compare generate array with deﬁned

domain

Priority High

Importance Must

zone

(28)

Chapter 4 System design

4.1 Detail steps of the system

After elicit the requirement, detail steps of the system is further designed to fulfill the functionality. Each process is designed as the above flow chart. From automatic perspective the close loop of the system guaranteed the whole reading, calculation and analysis could be realized by just click one button. In the ESARP system the multi- file of Comtrade will be listed and identified first, then analog channel of current and voltage will be extracted according to the IEEE standard. In the process of extracting the analog channel the multiplier value should be add with data in the DAT file. The CT, VT value of the .DAT file need to be further considered from different views of the system. Since the project will be evaluated the performance of the relay. So the transformer value should be seen from the secondary side. The analog channels should be saved in the matlab work space which is needed to be decomposed and further calculated for the dynamic performance of the relay. The fundamental value will be calculated through the DFT filter from the saved value in the matlab workspace. Such filter should be set with the variable time value so the flexible time window could be prepared for different processes of fault behavior. The critical part of relay performance calculation will be executed according to ABB algorithms after all the extracted and filtered values are ready. The samples of calculated value should be plotted from selected time window and saved in the text file automatically as the data media transfer between the matlab and java/jess environment. Steps illustrated above are performed in the matlab environment considering the complex filter value calculation. To integrate matlab environment together with logic steps in the Java environment. The compiler of matlab java builder is chosen as the tool for the construction of ESARP system. In this builder matlab files will be wrapped up separately in the compiler as class and function, after compiling Jar component will be generated and used for the built up the system in the expert environment. Expert system then will initial the jar component and deploy the functions of the multi-file reading, signal decomposition, dynamic relay performance calculation. And the result of the impedance value will be generated in the work space of expert system. Logic steps for the analysis first will pick up the files generated from the jar component.

Then each 60 samples will be saved in the separate array. Such array will be projected into the domain plane for the further comparison. The last step is expert advices will be generating after comparison being done and advices will be saved in the document as the ﬁnal outcome of the system.

(29)

Figure 4.1: Detail steps of the build up expert system

(30)

4.2 Develop environment and component selection

To build up such system is to verify the relay performance through the extracted the analog channel from multi-comtrade ﬁles to calculate the critical value by means of signal ﬁltering and relay algorithm. That functionality is suitable to perform in the matlab which is numerical computing environment and fourth-generation programming language. Matlab allows matrix manipulation, plotting of function and data, implementation of algorithm, creation of user interfaces, and interfacing with programs written in the other languages, including C, C++, Java and FORTRAN.

For the further develop of the system as the robust rule based system. Rule engine based platform JESS is considered for the developing speciﬁc logic for reasoning like relay malfunction. When running the rule engine, rules are carried out also new data will be created. Usually Jess library is embedding in the java code so Eclipse platform is chosen as the development environment to explore the features.

Eclipse is the platform chosen for the system development considering is an integrated development environment and an extensible plug-in system. It can be develop application in JAVA by mean of various plug-INS such as the Matlab Jar component. The Eclipse SDK includes the Java development tools, offering an IDE with incremental Java complier and Java Source files. The IDE also make use of the workspace which allowing external file modification.

Java installation needs to download Java JDK from web java.sun.com. After the JDK being appropriate installed system need to conﬁgure the environmental variable as the same path with java bin setting. Then matlab should verify the java environment by type version number and javac version, if the javac is installed, the version number will be appeared from the command window. Also the work space of the matlab should be the same as Java IED. So the comtrade ﬁles just need to be put into the workspace then automatic analysis can be performed.

4.3 Multi Files Reading

The IEEE COMTRADE standard is a file format designed for time series data that is established worldwide and is supported by standards making bodies. The set of comtrade files of a given event usually consist three types of files: HDR, CFG and DAT as the obligatory extension. It has a significant number of recording parame- ters that can be adapted for phasor data.[?] In this project the comtrade files are all generated from testing relay which means the specific form of Header, configuration and data files. The information contained in the file will be explained as following.

Header ﬁles:

The header ﬁle is an ASCII ﬁle contains textual information which the originator wants to convey to the end user. The information related with simulation system, the power system location, power system impedances, and transformer ratio is conveying

(31)

through generals, analog channel, binary channels, fault locations, and event channels. In the generals channel, recorded ID, trigger information, sampling frequency, Line value, IED information are defined with specific data. The channel index number, id, transformer rating value, trigger value and fault angle are all defined in the analog channel. Also information regarding channeled, trigger level setting, trigger enable, value of trigger are set in the binary channels. Fault location loop, direction value can be check through fault locations channels and in the event channel mainly contain the status change of the channel value.

Conﬁguration ﬁles.

The CFG files is a pre-fixed ASCII format which contains information include the station name, record for the simulation also defines the format of the DAT file. The crucial data as date and time, sampling rate, frequency, number and different channels are saved. The CFG. File used in our project has the information of channels and since the format is prefix, the possibility of multi-files reading for automated processing is prepared. Since analog channel including crucial information regarding voltage and current, the format as following shall be used.

An, chid, ph, ccbm, uu, a, b, skew, min, max, primary, secondary, P S

Where An

is analog channel index number. Critical, numeric, integer, minimum length = 1 character, maximum length = 6 characters, minimum value = 1, maximum value = 999999. Leading zeroes or spaces are not required. Sequential counter from 1 to total number of analog channels without regard to recording device channel number.

ch_id

is the channel identiﬁer. Non-critical, alphanumeric, minimum length = 0 characters, maximum length = 64 characters.

ph

is the channel phase identiﬁcation. Non-critical, alphanumeric, minimum length = 0 characters, maximum length = 2 characters.

ccbm

is the circuit component being monitored. Non-critical, alphanumeric, minimum length = 0 characters, maximum length = 64 characters.

uu

are the channel units (e.g., kV, V, kA, A). Critical, alphabetical, minimum length = 1 character, maximum length = 32 characters. Units of physical quantities shall use the standard nomenclature or abbreviations speciﬁed in IEEE Std 260.1-1993 [B4] or IEEE Std 280-1985 (R1996) [B5], if such standard nomenclature exists. Numerical multipliers shall not be included. Standard multiples such as k (thousands), m (one

(32)

thousandth), M (millions), etc. may be used.

a

is the channel multiplier. Critical, real, numeric, minimum length = 1 character, maximum Length = 32 characters. Standard ﬂoating point notation may be used (Kreyszig [B7]).

b

is the channel oﬀset adder. Critical, real, numeric, minimum length = 1 character, maximum length = 32 characters. Standard ﬂoating point notation may be used (Kreyszig [B7]).

The channel conversion factor is ax+b. The stored data value of x, in the data (.DAT) file, corresponds to a sampled value of (ax+b) in units (uu) specified above. The rules of mathematical parsing are followed such that the data sample x is multiplied by the gain factor a and then the offset factor b is added. Manipulation of the data value by the conversion factor restores the original sampled values.

skew

is the channel time skew (in s) from start of sample period. Non-critical, real number, minimum length = 1 character, maximum length = 32 characters. Standard floating point notation may be used (Kreyszig [B7]). The field provides information on time differences between sampling of channels within the sample period of a record. For example, in an eight-channel device with one A/D converter without synchronized sample and hold running at a 1 ms sample rate, the first sample will be, at the time, represented by the timestamp; the sample times for successive channels within each sample period could be up to 125s behind each other. In such cases the skew for successive channels will be 0; 125; 250; 375...; etc.

min

is the range minimum data value (lower limit of possible data value range) for data values of this channel. Critical, integer, numeric, minimum length = 1 character, maximum length = 6 characters, minimum value = -99999, maximum value = 99999 (in binary data ﬁles the range of data values is limited to -32767 to 32767).

max

is the range maximum data value (upper limit of possible data value range) for data values of this channel. Critical, integer, numeric, minimum length = 1 character, maximum length = 6 characters, minimum value = -99999, maximum value = 99999 (in binary data ﬁles the range of data values is limited to -32767 to 32767).

primary

is the channel voltage or current transformer ratio primary factor. Critical, real, numeric, minimum length = 1 character, maximum length = 32 characters.

(33)

secondary

is the channel voltage or current transformer ratio secondary factor. Critical, real, numeric, minimum length = 1 character, maximum length = 32 characters.

PS

is the primary or secondary data scaling identiﬁer. The character speciﬁes whether the value received from the channel conversion factor equation ax+b will represent a primary (P) or secondary (S) value. Critical, alphabetical, minimum length = 1 character, maximum length= 1 character. The only valid characters are: p,P,s,S.

The data in the data ﬁle, the channel conversion factors, and the channel units can refer to either primary or secondary units. So, a 345 kV to 120 V transformer for a channel in which the units are kV will have the primary factor of 345 and a secondary factor of 0.12 (345, 0.12). The primary or secondary variable (PS) is provided as a means to calculate the equivalent primary or secondary values in applications where the primary or secondary value is desired and the alternate value is provided. If the data originate in an environment that has no primary/secondary relationship such as an analog power system simulator, the primary-secondary ratio shall be set to 1:1.

With the determination of the primary (P) or secondary (S) values from the ax+b equation, the user can then determine the values required for analysis or playback.

Data ﬁles.

The DAT ﬁles contains the digital time tagged actual data samples which is recorded or simulated event. The discrete analog or logical sources recorded are saved in different channels. Analog channels are analysis for the dynamic impedance calculation.

The voltage and current value are presented in columns which corresponding to the time rows. The data is stored in ASCII format as defined in the CFG file. For the automated files reading the DAT files must conform exactly to the format of CFG files.

After we obtain all of the information and format contains in the comtrade file, we can start to decode the files. First we need to import data from the .cfg and .dat file.

In this step, the files addresses are listing by using DIR command. After that through derive the .dat file string, file name and path shall store into the matlab workspace.

The text information can be scan into the local cells by opening the .cfg and .dat ﬁles.

Detailed data will be extracting such as timestamps, analog channels information and digital channel information after scaling. The final waveforms will be produced as the post-process through data configuration according to IEEE standard. The saved channels will be then used by the transferred relay model which will be specified next chapter. By defining a globe variable the relay model and data extracting could be integrated. And Through loop reading the function of data extracting the existing comtrade file could be seen as testing batch injected into a relay model which in the format of algorithm which reflect the behavior of the relay. The RTDS relay model shows how the signals of analog channels are processed in the relay. Before the analog channel signal entering into relay model the signal shall be filtered usually through low pass filter to remove the DC noise. Then the sampler will decide the sample

(34)

rate since the input signal usually is at a much higher sampler rate than the actual rate for protection algorithm. And a PLL in the ﬁgure is used for performing the frequency tracking to adjust the sampling rate. So the protection algorithm could be synchronous with the system frequency. The signals then will enter into the full cycles discrete Fourier transform ﬁlter to extract the fundamental values. A full cycle discrete Fourier transform can be used to create the real and imaginary values for the phasor representation of values that will be used for the further calculation of the impedance magnitudes. The single phase quantities can be used to create the phase to phase values for the necessary calculation of the phase to phase impedance values.

For the calculation of phase-to-ground impedance of fault include the ground the single phase current quantities are selected to create the zero sequence current then multiplied by the K0 factor. The value of zero sequence compensation is necessary to calculate the correct magnitude of phase to ground impedance for faults which is determined by the positive sequence impedance and zero sequence impedance of the transmission line. The last step is the magnitude quantities calculated are used for the phase to phase and phase to ground fault according to the vendors algorithms.

4.4 Relay Model transfer

Figure 4.2: Test circuit of relay model in RTDS

The analogy current and voltage signals ﬁrst will enter into the signal adaption block, and then through the Fourier ﬁlter signal will be decomposed into complex value. Phase selection is calculated in order to further calculate the quantities of distance or phase to phase fault. The data will be recorded in the memory and through

(35)

Figure 4.3: Signal process in the relay model

MMI as the man machine interface. The detailed description technical solution for each block will be concentrated in the following section.

4.5 Signal Decomposition

UL11it = _m²

t+^m₂

n=t−^m₂ UL1ncos^{1∗2∗pi∗n}_m UL11rt = _m²

t+^m₂

n=t−^m₂ UL1nsin^{1∗2∗pi∗n}_m UL21it = _m²

t+^m₂

n=t−^m₂ UL2nsin^{1∗2∗pi∗n}_m UL13it = _m²

t+^m₂

n=t−^m₂ UL3nsin^{1∗2∗pi∗n}_m

The signals are decomposed by using above equations. For guarantee a flexible time window, variable n is applied as time window variable so the time scale could be chosen as the certain time cycle base on the time variable t value. In this project the trigger value are identified and used as the different behavior observation windows which will show in the result section. The analog signals will be filter and transfer

(36)

into complex value through such equations and further calculated for the dynamic performance of relay by using ABB algorithms.

4.6 Matlab Java Builder

Matlab builder JA enables to create a Java classes within MATLAB programs. The jar component can be integrated into Java programming as well as deployed with desk- top computer or web servers which do not have MATLAB installed. Such component makes MATLAB based computation, visualization and graphical user interface de- ployable and accessible to the end users of java program. The component behaves just like Java class. So it can be run on the all the platform support MATLAB.[13]

The jar ﬁles depend on javabuilder.jar which ship with toolbox.java builder.jar when is instantiated as series events occur.

a) Load dependent class in javabuilder.jar.

b) The initialization of such classes will trigger the loading of the shared libraries which will implement a number of native methods which form the bridge from generated MATLAB builder JA to the Compiler runtime implementation.

c) After the shared libraries are loaded the MATLAB runtime is initialized by creating an instance of C++ class.

d) Such MCR (Matlab Compiler Runtime) instance triggers the initialization of many subsystems. One of the subsystems is the MATLAB-Java interface which allows MATALB to call JAVA directly. When the MCR is loaded into a running JVM, the matlab-Java interface established a connection to the running JVM by calling Attach Current Thread.

e) Attach Current Thread is a class that loads all classes needed by MATLAB utiliz- ing the interface of Matlab-java.

Notice that deploying support for the component shall under the condition of access with run time on end user machine. So the created installer for the application shall support Java on target machine which means MCR must be installed.

Such project demanding different roles in charge of different task as following table shown, at the beginning of the project end user requirement need to be elicited and satisfied by matlab coding and Java programming. Then for the MATLAB builder and Java developer tasks are divided into details. The final solution that created and tested shall be executable for the end user after the whole project developed.

(37)

Role Tasks Details Matlab

Program- mer

• Understand relay testing re- quirements and the mathematical models of the relay

• Write MATLAB code for ﬁle reading and calculation

• Build an executable jar component with MATLAB compiler (usually with support from a Java developer)

• Package the component for distribution to end-user

• pass the packaged compo- nent to the Java developer for rollout and further integration into the expert system environment. . .

• Copying the matlab Files

• Testing the MATLAB File You Want to Deploy

• Creating the Java Compo- nent

• Packaging the Java Compo- nent

• Copy the Package You Cre- ated . . .

Java De-

veloper • Write Java code to execute the Java package built by the MATLAB programmer

• Roll out the packaged com- ponent and integrate it into the expert system environment

• Use the component in Java applications, adding and modifying code as needed

• Address data conversion is- sues that may be encoun- tered, according to the end user’s speciﬁcations

• Ensure the ﬁnal Java ap- plication executes reliably in the end user’s environment. . .

• Gathering Files Needed for Deployment

• Testing the Java Compo- nent in a Java Application

• Distribute MATLAB Code Using the MATLAB Com- piler Runtime (MCR)

• Calling Class Methods from Java

• Handle Data Conversion as Needed

• Build and Test . . .

End user Execute the solution created by MATLAB and Java developers

Run the deployed application (outside the scope of this document)

Table 4.1: tasks for diﬀerent roles within java builder

(38)

4.7 Logic steps in Expert environment

For the automatic system matlab workspace should be set as the same workspace of expert system environment. So the ﬁle identify, data media generated and result saved can be easily accessed.

The data media generated in our case is the value of impedance and reactance. After the value being generated the expert system will read from the data media and save each 60 samples into the each array in the work space. Such array will stand for the dynamic relay value and will project into the relay plane for comparison. Different array project into different relay plane will generate different expert advice and the advice will be saved automatically. Such file will be used as the further comparison of RTDS simulation condition. Then if the relay has faulted perform. Parameters which shown in the above table need to be reset, then we shall test the relay again to guarantee the relay performance.

(39)

Chapter 5

System function result

5.1 Analog signals channel plotted

The precondition of the test system is P=381MW Q=-36.9 MVAr export from left side. Pre-fault current 0.56kA.90 deg inception, CT, VT value are 1000/1 and 400kV/100kV. L1 to ground fault and diﬀerent indication should be given as 0

Figure 5.1: Analog signals channel plotted

5.2 Impedance dynamic performance with corresponding time scale

As shown in the plotted figure the analog signal extract from the Comtrade files will be used as the source for the relay performance calculation. The different analog channels are indicated by using different color function. From the figure we can easily see that from 0.1-0.2 time scale the disturbance happen in the relay. Such trigger signal is very important for diagnose of the dynamic performance of relay. We will

(40)

Figure 5.2: Impedance dynamic performance with corresponding time scale

further observe the diﬀerent relay behaviour through such trigger signals. Form the table shown above. The diﬀerent activities of relay corresponding to the time line are illustrated. Before the trigger signal 35ms is the measuring time for the pre-fault value so we could see the impedance entering into the fault zone from that time period.

Then after the trigger signals for 4-10ms is the start function of measurement fault then 5ms is taken for the synchronization of the measurement. The fault value then will be identiﬁed and measure for 35ms. After that the values will be transport and recorded in the data media Comtrade in this case. So the diﬀerent time scale chosen base on the trigger signal here is crucial for the analysis of the relay performance. The performance of the relay would be the dynamic process of entering into the fault zone when we chose the time scale corresponding to the red color indicate time window.

Different line value of fault has different behaviour which is plotted as figures shown.

Then the 3 cycle’s samples chosen behind the trigger signals time window are plotted.

The fault behaviour based on the fault line value can be seen as the trajectory of rotating in the fault zone. Except that in the 0 percent of the line value the fault is set into 0 of reactants and resistance without further perform. The 1 cycle and

(41)

Figure 5.3: 0 percent of line value Figure 5.4: 50 percent of line value

Figure 5.5: 90 percent of line value Figure 5.6: Corresponding time scale

before fault and 2 cycles after fault is shown above which is the scenario we choose for the relay performance criteria. Not only has the trajectory showed the impedance behaviour entering into fault zone, but also the fault zone dynamic rotation. Such impedance will be saved into the data media and further compare with the impedance plane for the expert advice generated.

(42)

(43)

(44)

5.3 Automatic saved value of resistance and reactants

After all the ﬁles reading and data extraction the samples will be calculated according to the ABB algorithm. The result of 60 samples of resistance and reactants will be automatically saved into a txt ﬁle. Such data could be further used in other program for analysis. Also saved data can be the resource of research related whether algorithm of optimal parameters resetting could be developed.

5.4 Evaluate from Expert advice

The expert advice will be the ﬁnally result as the system being saved in the txt ﬁle.

Such result can be compare with the simulated scenario applied on the relay. If the diﬀerence between what relay seen from the fault and what fault applied appear. The analysis related with relay parameters will be interfere with the relay.

(45)

Chapter 6 Future work

6.1 User friendly interface

A user interface needs to develop with the chosen parameters and time window which the test people want to observe.

6.2 Robust rule based system

The mal-operation of the relay could have diﬀerent reasoning:

1. relay is aﬀected by the remote end in feed.

2. when the 3I on parallel line have the same direction as 3I on protected line it will cause the distance protection to under reach.

3. No start of PHS (phase selection) . . .

Due to the time issue, the system is not fully developed as the phase selection and directional element need to be added. After the system have fully functions, the system could further evolved into the robust rule based system based on artiﬁcial intelligent software JESS which has the pattern match reasoning function. The rule base function system would be more intelligent and more resourced will be saved by such system.

6.3 GUI function for adoptive domains deﬁne

The zone 3 value usually shows the relay has the domain which needs to be reset. If a GUI function has the dynamic domain reset will be more human interactive. The user could use such function get more straight forward information of relay performance.

6.4 Optimized parameter resetting for executing analysis

Since the whole system is about to adjust the relay parameters. The saved value of the impedance could be used as the future analysis. The optimization of the parameter resetting for executing analysis may develop through saved values.

(46)

Chapter 7 Conclusions

In this project. The ESARP system which has the functionality of multi Comtrade files reading,signal decomposition, impedance dynamic calculation and indication of difference protection zone is achieved. Such system shall benefit the relay testing process with time saving of both human analysis and testing devices. Such system is highly replicable and can be modular into a highly reusable tool which assist the human expert to enhance the working efficiency.

(47)

Chapter 8 Discussion

The next step of the system development will be implement with other functionality of fault phase selection, direction element of protection and load encroachment. After the fully function developed the system shall testing and integrate with the testing process. The possibility of system generated alarm signal and stop testing process shall be discussed. Also to much the into a more intellectual level shall be the future work to assist the human expert.

(48)

Bibliography

[1] A. Apostolov and B. Vandiver, “Ensuring the correct operation of distance relays under dynamic system conditions,” in Protective Relay Engineers, 2008 61st Annual Conference for, pp. 72–77, IEEE, 2008.

[2] A. Apostolov, “Eﬀective and eﬃcient testing of multifunctional protection ieds,”

in Protective Relay Engineers, 2012 65th Annual Conference for, pp. 382–388, IEEE, 2012.

[3] A. Apostolov and B. Vandiver, “Maintenance testing of multifunctional distance protection ieds,” in Transmission and Distribution Conference and Exposition, 2010 IEEE PES, pp. 1–6, IEEE, 2010.

[4] W. Zhang, S. Lee, M. Choi, and S. Oda, “Considerations on distance relay setting for transmission line with statcom,” in Power and Energy Society General Meeting, 2010 IEEE, pp. 1–5, IEEE, 2010.

[5] Z. Li, S. Sheng, S. Richards, and P. Sankarakumar, “Automated test solutions for multifunctional protection relays,” in Developments in Power Systems Pro- tection, 2012. DPSP 2012. 11th International Conference on, pp. 1–5, IET, 2012.

[6] M. Kezunovic, Y. Xia, Y. Guo, C. Fromen, and D. Sevcik, “Distance relay ap- plication testing using a digital simulator,” Power Delivery, IEEE Transactions on, vol. 12, no. 1, pp. 72–82, 1997.

[7] A. Abdlrahem and H. Sherwali, “Modelling of numerical distance relays using matlab,” in Industrial Electronics & Applications, 2009. ISIEA 2009. IEEE Sym- posium on, vol. 1, pp. 389–393, IEEE, 2009.

[8] R. technologies, “Rtds@ONLINE,” Nov. 2012.

[9] A. manual, “distance protection,” in Section 6, pp. 116–297, IEEE.

[10] A. Phadke, J. Jodice, M. Adamiak, J. Brandt, J. Bright, R. Burnett Jr, C. Cas- tro, T. Cease, D. Clark, and G. Clough, “Comtrade; a new standard for common format for transient data exchange,” IEEE Transactions on Power Delivery (In- stitute of Electrical and Electronics Engineers);(United States), vol. 7, no. 4, 1992.

[11] X. Luo and M. Kezunovic, “An expert system for diagnosis of digital relay oper- ation,” in Intelligent Systems Application to Power Systems, 2005. Proceedings of the 13th International Conference on, pp. 6–pp, IEEE, 2005.

(49)

[12] J. Bekker and P. Keller, “Enhancement of an expert system philosophy for au- tomatic fault analysis,” in Proceedings of 11th Annual Georgia Tech Fault and Disturbance Analysis Conference, 2008.

[13] D. Oswill, “Matlab javabuilder@ONLINE,” June 1994-2012.