基于距离度量和健康指数的电子设备健康评估方法

摘要

Health state of a system can be acquired through assessing the monitored parameters,which is a foundation to perform preventive maintenance for ensuring the operating safety before failure occurs.For electronic system,false alarm often happens due to random disturbance which causes unnecessary replacement.An integrated health evaluation method is proposed to construct a health index based on the statistics distribution of multiple performance parameters.Distance measurement is used to reduce the dimension of the time series of multiple parameters by transform them into a Mahalanobis distance.A system health index is built by the weighted sum of each bin's contribution of the histogram with an optimal kernel bin-width.A case study is performed on computer simulation and experiments.Results are given for the situation of single and multiple parameters abnormal,which demonstrates the effectiveness of the method and the ability to decrease false alarm.%依据监测参数对系统进行健康评估,可以了解系统在工作过程中的健康状态,并通过预防维修避免系统故障带来的安全隐患;电子设备的健康状态容易受到随机干扰的影响产生虚警从而导致不必要的维修,为此提出一种通过多参数距离度量来构建健康指数从而进行系统综合健康评估的方法;利用马氏距离对监测的多个系统性能参数时间序列进行降维,基于核函数最优分段间隔估计对数据的直方图分布进行统计,利用加权分数贡献率来构建健康指数;通过电脑的实验与仿真数据进行了验证,给出了单参数异常和多参数同时异常情况下的评估结果;研究表明,该方法不仅能对系统的健康状态进行有效的评估,而且能够降低随机突发干扰产生的虚警.