为了提高对复杂系统所产生故障的诊断效率,提出一种基于遗传算法(GA)和人工神经网络(NN)的综合诊断模型。该诊断模型的运行分为两步:1)利用NN对故障诊断数据进行预处理;2)运用GA进行诊断。将嵌套神经网络与遗传算法有机结合,嵌套神经网络的功能相当于一种预处理器,可以减少由遗传算法处理的故障类型的数量。弥补了遗传算法不能诊断出可信度和神经网络难以鉴别出最可能故障点的缺点。提高了诊断结果的精确性、可靠性和一致性,而且在一定程度上减少了总运行时间。

In order to improve the fault diagnosis efficient of complex systems,a hybrid diagnostic model is put forward based on Genetic Algorithm(GA) and Artificial Neural Network(ANN). The diagnostics model runs through the following two steps:1) using NN to preprocess the fault diagnosis data;and 2) using GA to diagnose. The model combines Nested Neural Network (NNN) and GA together,where NNN is used as a pre-processor which can reduce the quantity of fault types to be explored by the GA. The model overcomes the shortcoming of GA which can not judge reliabilities and that of NN which is difficult in identifying the most possible fault spot. It improves the accuracy,reliability,and consistency of the diagnosis result,and the total running time is also reduced to some extent.