Transient Stability Assessment Using Fuzzy Combined Support Vector Machines

Abstract

In the present day scenario, the electric power demand is increasing, so power systems have become larger and more complex. In order to keep reliability, fast and reliable transient stability assessment schemes are desired in power system operations. In this article, a quick and unfailing transient stability assessment algorithm is proposed, where support vector machines are employed as pattern classifiers so as to build fast relation mappings between the stability results and selected input attributes. Support vector machines with different kernel functions and kernel parameters are constructed and trained to calculate hyperplanes that separate the stable and unstable states of power system for .n 1/ faults. The outputs of three randomly selected support vector machines are combined by fuzzifying the distance of a given operating state from their optimal hyperplanes and their validation accuracies. By combining the outputs of a team of classifiers, a more accurate decision than that of a single best member has been achieved. Each fuzzy combined support vector machine is designed to deal with a single contingency scenario. Feature extraction through discrete wavelet transform alleviates the problem of high input dimensionality. The simulation results obtained using three bench mark systems demonstrate the good potential for fuzzy combined support vector machine classifiers in transient stability assessment.

Prediction of Probabilistic Transient Stability Using Support Vector Machine

“…In [68], a unique TSA algorithm was presented, where SVMs were employed as pattern classifiers. SVMs with different kernel functions and kernel parameters were constructed and trained to compute hyperplanes that split the stable and unstable states of power system for (n − 1) faults.…”

A Hierarchical Method for Transient Stability Prediction of Power Systems Using the Confidence of a SVM-Based Ensemble Classifier

“…Description Feature Description f 1 Average{P mi (t b )} f 18 Max{ ω i (t k )} + Min{ ω i (t k )} f 2 Average{P ei (t FOT )/P mi (t FOT )} f 19 Coefficient of variation{ ω i (t k )} f 3 Max{P ei (t FOT )/P mi (t FOT )} f 20 α COI (t k ) f 4 Min{P…”

“…The simulation results show that the proposed hierarchical method can balance the accuracy and rapidity of the transient stability prediction. Moreover, the hierarchical method can reduce the misjudgments of unstable instances and cooperate with the time domain simulation to insure the security and stability of power systems.Energies 2016, 9, 778 2 of 20 from among massive sets of data, have been used to predict the transient stability [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19]. The transient stability prediction can be treated as a two-class classification (stable and unstable) problem and solved by machine learning methods.…”

“…Recently, the construction of input data and the improvement of machine learning methods to increase the accuracy have been the main research focus. Methods such as neural networks [6-8], decision trees (DTs) [9][10][11], support vector machines (SVMs) [12][13][14][15], fuzzy theory [16,17] and some ensemble classifiers [18,19] had been used for post-fault transient stability analysis.In fact, a classifier should possess high prediction accuracy, but it is also important to estimate the credibility of each classification result, or the confidence level. The existing transient stability prediction only provides a prediction result but seldom considers the confidence level.…”