进化神经网络在柴油机故障诊断中的应用

Fault Diagnosis for Diesel Engines Using Evolutionary Neural Networks

瞬时转速波形诊断法是柴油机故障诊断中的一种较为方便实用的诊断方法,本文较为系统地介绍了其测量、故障参数提取及分析方法,并在诊断分析中使用了进化RBF神经网络,此网络的训练采用遗传算法,优化了径向基神经网络的结构和参数,加强了网络的实用性。最后用神经网络对所模拟的故障进行辨识,证实了瞬时转速法与进化神经网络的结合在柴油机故障诊断中的可行性。

The fault diagnosis about diesel engines with instantaneous speed is a convenient and practicable method, which is systematically introduced about its measuring,extracting fault parameters and analysing method.Evolutionary radial basis function neural networks(RBFNN) is used in the analysis. This network trained with genetic algorithm optimizes the RBF neural network's configuration and parameters, making the network more useful. After some faults are identified by using RBF, the results show the effectiveness of the combination between the RBFNN and the fault diagnosis using instantaneous speed.