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。 Aiming at the characteristics that the analog circuit components are complex, fault-changing and multi-feature characterization, it is impossible to accurately identify the potential fault hidden by the single information fault prediction method. Therefore, by extracting the fault parameters of multiple test points, Get the fault information, and use the comprehensive advantage of AR prediction model and gray prediction model to complete the effective prediction of the fault. Finally, the weighted Mahalanobis distance is used to weighted fusion of multi-feature data to complete the prediction of the fault and effectively reduce the occurrence rate of the analog circuit fault .