基于两级BP网络伺服阀故障诊断专家系统的研究与应用

Research on Servovalve Fault Diagnosis Expert System Based on Two-stage BP Network and Its Application

运用人工智能提高电液伺服间的故障诊断水平。研究伺服阀静态特性曲线和伺服阀状态的对应关系，在特性曲线上提取状态特征参数作为人工神经网络样本，把训练好的神经网络作为专家系统的知识库。状态特征参数提取方法能提高训练样本的质量。利用两级BP网络建立的伺服阀故障诊断专家系统已成功应用于液压AGC测控系统，并具有推广价值。

Artificial intelligence is used to impr0vethe fault diagnosis of electro-hydraulic servovalve. Theexperimental method t0 construct fault diagnosis expertsystem knowledge base 0f servovalve is given by practica1examples. The c0rresponding relationships between theservovalve static characteristic curve and its status arestudied, the status characteristic parameters are takenfrom servovalve static characteristic curve and used as thespecimen for the training and learning of neuraI netw0rk,and the trained netw0rk is used as the kn0wledge base ofexpert system. The meth0d taking status characteristicparameters used can improve the quality of neural net-work specimen. The servovalve fault diagn0sis expertsystem constructed on the basis of two - stage BP net-work has been successfully used in the measuring andcontrol 0f hydraulic AGC system,with a potentia1 valueof wider application.