基于自适应MPCA的翻车机液压系统状态监测与故障诊断

Car dumper hydraulic system state monitoring and fault diagnosis based on adaptive MPCA

由于翻车机液压系统故障频发,严重影响生产,对其进行状态监测与故障诊断具有重要的实际意义。根据翻车机工作过程具有间歇性和时变性的特点,采用自适应多向主成分分析(MPCA)对其进行状态监测与故障诊断。应用加权递归算法自适应更新协方差矩阵,并通过权值调整新数据对最终模型参数的影响。针对Q贡献率不能反映各过程变量偏离正常范围的程度,导致故障定位不准。采用主成分t贡献率进行故障定位,其直接揭示了过程变量的变化程度。根据翻车机在线监测系统的运行情况证明自适应MPCA方法可以准确及时的发现故障,基于t贡献率的故障定位方法准确率可以达到90%。

Faults of the hydraulic system of car dumper occurred frequently and the faults bave a serious impact on production.Its condition monitoring and fault diagnosis has important practical significance.According to characteristics of working process of car dumper is intermittent and time-varying,adaptive multi-way principal component analysis(MPCA)was used for condition monitoring and fault diagnosis.Covariance matrix was updated adaptively using the weighted recursive algorithm.Impact of new data on final model was adjusted by weight.Contribution rate of Q can not reflect the degree of deviation from normal range of each process variable that leads to inaccurate results of fault diagnosis.Parameter of principal component contribution rate t was used for fault location which directly reveals the change degree of process variables.The operation of the online monitoring system over car dumper proved that the adaptive MPCA method can accurately detect fault in time.The accuracy of fault diagnosis is up to 90%based on the t contribution rate.