电液伺服系统泄漏故障识别及分类方法研究

电液伺服系统的泄漏故障具有普遍性、缓变性和多样性,此类故障的诊断意义十分重大,但诊断过程极为困难。针对泄漏故障的识别与分类问题,提出一种基于鲁棒残差的支持向量机诊断方法。在考虑系统未知输入的前提下,设计系统的线性鲁棒观测器,计算系统的鲁棒残差,发现残差可以有效识别泄漏故障,但却不能正确区分泄漏形式。由于位移、压力也含有丰富的故障信息,于是提取残差、位移、压力的波形指标和压力的峭度指标,作为支持向量机的输入,用于系统的泄漏故障分类。分类结果表明:该方法不仅可以确定系统是否发生了泄漏,而且可以判定系统发生了哪种泄漏。

Leakage Faults Detection and Isolation for Electro-hydraulic Servo System

Leakage faults of electro-hydraulic servo system are of catholicity,gradual change and variety.Faults diagnosis of leakage is significant and difficult.In order to detect and isolate leakage faults,a diagnosis method named support vector machine was presented based on robust residual.Considering unknown inputs,a robust observer was designed.Residual curves show that they are effective to detect leakage faults,but non-effective to isolate these faults.Because displacement and pressure contained abundant fault information as well as residual,a support vector machine was used to isolate leakage faults,whose inputs were waveform of residual,displacement and pressure,and kurtosis of pressure.Classified result reveals that this method can be used to successfully detect and isolate leakage faults of electro-hydraulic servo system.