Artificial Neural Network Based Transient Stability Assessment Of Power Systems

Transient stability is one of the most important problems in modern power systems, which have complicated characteristics. Traditional transient stability assessment (TSA) methods are time-domain numerical simulation and direct method. Artificial neural network (ANN) is a new technique for TSA of power systems. In this paper, a novel compound neural network for TSA is proposed. The compound neural network is composed of a Kohonen network (KN) and several radial basis function networks (RBFN). The feasibility of the presented approach is validated by the simulation results of Central China Power Grid (CCPG).Furthermore, the TSA module based on ANN is embedded the transient stability on-line pre-decision system. The system is implemented by integrating the ANN based TSA and time-domain numerical simulation based strategy generation. In CCPG's transient stability control (TSC) system, the control strategy mode can be called "centralized management and hierarchical control". The pre-decision system is located in the center of the whole TSC system. When the system operation conditions change and are censored, the system will automatically generate the control strategy for the foreseen contingency set using the real-time data. When faults occur, the sub-stations will immediately match the faults with some proper strategies in the strategy table transferred before from the host-station. If needed, the control strategy will be executed by terminal device. The test results indicate that the new pre-decision system, which is embedded ANN based TSA module, calculates faster than the old one, therefore, is more propitious to online application.This paper includes six chapters. In Chapter 1, basic ideas of power system transient stability are introduced. The application of ANN in electric power systems is also included. In Chapter 2. the principia of the self-organization feature mapping (SOFM) neural network and radial basis function neural network are introduced. Chapter 3 is mainly devoted to the novel compound ANN's construction, principle and performance for TSA. The basic idea of numerical based strategy generation combined with ANN based TSA is expatiated in Chapter 4. Some test samples are presented at the end of this chapter. In Chapter 5, an optimizing method based on genetic algorithm for enhancing the performance of the compound ANN is discussed. Chapter 6, the last chapter, concludes the whole paper with the author's expectation of good future for the on-line practical ANN based pre-decision stability control systems.