论文部分内容阅读
由于水压试验机机理建模的复杂性,很难从机理方面对其生产过程进行故障诊断,基于多向Fisher判别分析的故障诊断方法同时利用正常工况和故障条件下的数据建模,具有很强的诊断能力.为了解决应用此类方法遇到的不等长周期问题,根据水压试验过程的数据特性提出了子时段首部数据等长法将所有批次调整成平均长度.在此基础上对每个子时段分别建立MFDA模型,并结合递推思想将三个时段的诊断结果密切结合以提高诊断精度.采用水压试验机实际生产过程的4个锁阀故障数据对该方法进行测试,结果验证了该方法的有效性.
Due to the complexity of hydrostatic testing machine mechanism modeling, it is very difficult to diagnose its production process from mechanistic aspect. The fault diagnosis method based on multi-directional Fisher discriminant analysis can make use of data modeling under normal working conditions and fault conditions, Strong ability to diagnose.In order to solve the inequality period problem encountered by using this kind of method, according to the data characteristic of the hydrostatic test process, the sub-period header data isometric method is used to adjust all the batches to the average length. The MFDA model was established for each sub-period, combined with the recursive thinking to combine the diagnostic results of the three periods to improve the diagnostic accuracy.This method was tested by the four lock-valve fault data of the actual production process of hydraulic pressure testing machine, The results verify the effectiveness of this method.