基于多特征信息融合的模拟电路故障预测方法研究

Research on analog circuit fault prediction method based on multi feature information fusion

针对模拟电路元件构成复杂、故障多变及多特征表征的特质,采用以往的单一信息故障预测方法已经无法精准识别潜在的故障隐患,为此,本文通过提取多个测试点的故障参数来最大限度获取故障信息,并运用AR预测模型和灰色预测模型的综合优势完成故障的有效预测,最终引入加权马氏距离对多特征数据进行加权融合,从而完成故障的事先预测,有效降低模拟电路故障发生率。

for the complex components, variable fault and changeable features of the analog circuit, using single information as the fault prediction methods has been unable to identify potential problems accurately, therefore, in this paper, through extraction of fault information from multiple test points to obtain maximum fault information, and use the comprehensive advantages of the AR predictior~ model and grey forecasting model to complete fault prediction effectively,and finally take the weighted Mahalanobis distance to weighted fusion of multi feature data, so as to make fault prediction and reduce the incidence of fault effectively.