Performance evaluation of measurement algorithms used in IEDs.

Many Intelligent Electronic Devices (IEDs) are available for the protection of power systems. These IEDs use a series of mathematical algorithms for fault detection and execute various protection functions. The first and essential mathematical algorithm of any IED is the measurement algorithm. The aim of the measurement algorithm is to estimate the fundamental frequency component (phasor) of input current and voltage signals. Most protection algorithms use the estimated phasor for their executions. The most important factors for the successful use of the protection algorithms in IEDs are accuracy and speed of the phasor estimation by the measurement algorithms. A fault in a power system produces step changes in the current and voltage phasors recorded by IEDs as well as a variety of nuisance signals. The nuisance signals introduce significant input distortions to measurement algorithms. Measurement algorithms that estimate the fundamental frequency phasor component from the distorted input signals produce some errors. Different measurement algorithms produce different amounts of error. This is because their design is based on different approaches with different assumptions that result in different performance in the presence of nuisance signals. It is important to evaluate the performance of measurement algorithms in the presence of nuisance signals. The evaluation is to ensure that measurement algorithms estimate the fundamental frequency component at the required design accuracy and speed. The result of the performance evaluation can be used to select appropriate measurement algorithms for specific protection applications. However, the parameters of nuisance signals are uncertain due to their dependence on unpredictable factors such as fault location and fault impedance. Thus, a methodology for the evaluation of measurement algorithm performance should take into account the uncertainty of the parameters of nuisance signals. The traditional method of evaluating the performance of measurement algorithms is based on the local sensitivity method using a linear function approximation at a nominal point. The local sensitivity method varies only a single nuisance parameter (factor) while other factors are fixed at their nominal values. The studied factor is varied to observe errors in the output of the measurement algorithm. Such an approach, however, does not provide the overall performance of measurement algorithms. Besides, varying the single factor does not represent realistic scenarios. This thesis proposes a new methodology to evaluate the performance of measurement algorithms implemented in IEDs. The proposed methodology uses the global uncertainty and sensitivity analysis method. In this method, all factors representing nuisance components are varied simultaneously. Uncertainty analysis measures the uncertainty in output of the measurement algorithm due to the uncertainty of input factors. Sensitivity analysis measures the contribution of all factors and their interactions to output uncertainty. In general, the global uncertainty and sensitivity method that is based on the Monte Carlo approach requires extensive evaluations. Its implementation can be prohibitive, particularly in practical testing, because the number of factors is large. Thus, a two-stage methodology with a significantly smaller number of evaluations is used. The first-stage is the use of the Morris method as a preliminary (screening of factors) sensitivity analysis and the second-stage is the implementation of the Extended Fourier Amplitude Sensitivity Test (EFAST) technique for comprehensive global uncertainty and sensitivity analysis. A single evaluation involves one run of the IED injection test which can take a few minutes. Thus, it is justifiable to search for the methodology that is uses the smaller number of evaluations. The proposed methodology contributes to an automated testing method integrating ATP/EMTP, MATLAB and SIMLAB programs as well as the injection test facility. The ATP/EMTP program is used to generate fault test scenarios. The MATLAB program is used to model elements of the IED to calculate performance indices on the output of measurement algorithms and automatically control the process of extensive evaluations (simulations). The main role of the SIMLAB is to analyze the uncertainty and sensitivity of the measurement algorithms outputs. The proposed methodology has been demonstrated by evaluating the performance of a known measurement algorithm in simulation and an unknown measurement algorithm of a commercial IED (SEL-421). The methodology has been successfully performed in the simulation as well as in practical testing. The results of the analysis indicate that the performance is typically most sensitive to a few parameters out of many possible factors. These important parameters should then be the focus of research for the optimization of measurement algorithms.Thesis (Ph.D.) -- University of Adelaide, School of Electrical and Electronic Engineering, 201