大电力系统可靠性评估的软计算模型

Research on Soft Computing Models for Reliability Assessment of Bulk Power Systems

为减少大电力系统可靠性评估中偶发事件的潮流计算次数,提高故障模式识别率,将粗糙集与神经网络、遗传算法有机融合,提出了电力系统偶发事件模式识别的软计算模型——粗神经网络融合模型(RNNIM)和基于RNNIM的电力系统可靠性评估算法。首先利用粗糙集方法约简RNNIM的输入变量,提炼样本,提取事件类与系统状态关系的概略化规则集;其次利用概约化规则集建立RNNIM,用遗传学习算法训练网络。以RBTS和IEEE-RTS79测试系统为例,说明所提模型是正确、可行和有效的。

In order to decrease the times of power flow calculation of stochastic events in bulk power systems reliability evaluation and identify whether a stochastic event belongs to contingency pattern quickly and effectively, a soft computing model of contingency pattern identification-Rough Neural Network Integrated Model (RNNIM), is presented in this paper by means of integrating rough set, artificial neural network and genetic algorithm. Furthermore, a power system reliability evaluation algorithm based on RNNIM is put forward. Firstly, the input variables of RNNIM are reduced; learning samples are extracted; a probable rule set about the relation between stochastic event classes and system states are draw out by means of rough set methods. Secondly, RNNIM is decided on the basis of the probable rule set. Finally, the network is exercised with a genetic learning algorithm for improving the accuracy identifying stochastic events modes. The numerical experiments for RBTS6 and IEEE-RTS79 show the correctness, feasibility and availability of RNNIM.