基于改进HMM的潜在电子故障状态识别模型

Potential fault recognition based on improved hidden Markov model

针对复杂电子装备隐性故障难以诊断的难题,在深入分析隐马尔可夫模型的核心问题及基本算法的基础上,探讨了其在故障诊断应用中存在的主要问题,建立了多状态电子装备可靠性评估模型,利用系统可靠性评估结果作为隐马尔可夫模型的初始模型特征量,改进了传统的隐马尔可夫模型,并对Baum-Welch训练算法进行了优化,形成了一套适于复杂电子装备潜在故障状态跟踪识别的数学模型。实验结果显示,理论方法及模型能够更好地识别潜在故障状态,加快了模型训练速度,提高了故障状态识别率。

Hidden failures are difficult to diagnose in complex electronic equipment.Based on in-depth analysis of the core issues and basic algorithm of hidden Markov model,the main problems of the algorithm in fault diagnosis application is studied,a reliable multi-state electronic equipment assessment model is established,the system reliability assessment result is used as the initial model features of hidden Markov model;and traditional hidden Markov model is improved,and the Baum-Welch training algorithm is optimized.A mathematical model of potential failure status track and identification suitable for complex electronic equipment is formed.Experiment results show that the theoretical method and model can better identify potential fault condition,speed up model training speed and improve fault condition recognition rate.