暂态稳定仿真的综合人工智能方法

TRANSIENT STABILITY SIMULATION BASED ON ARTIFICIAL INTELLIGENC

研究了利用综合人工智能进行暂态稳定仿真的方法:利用神经网络训练动态元件的输入输出关系,同时,利用遗传算法优化神经网络的结构和加快神经网络的训练速度。算例结果表明,利用遗传算法和神经网络相结合的综合人工智能方法暂态稳定仿真比传统的时域仿真计算速度快,并能保持较高的精度,可以有效地提高暂态稳定仿真的实时性。

This paper presents an integration method to simulate transient stability based on artificial intelligence which integrates artificial neural network (ANN) and genetic algorithm. The method uses ANN instead of traditional differential equations to describe the input and output relationship of dynamic elements in power system. This paper also build a time delay artificial neural network in order to identify dynamic characteristics of pattern samples and uses Karhunen-Loeve transformation to reduce redundancy of input vector of ANN. In this paper, genetic algorithm is used to optimize the structure and train the weights of ANN for speeding up the simulation. The simulation result for IEEE 39 system illustrates the validity of the method.