基于AP-HMM混合模型的充电桩故障诊断

The Fault Diagnosis of Charging Piles Based on Hybrid AP-HMM Model

确定性的相似性传播(AP)聚类方法和统计性的隐马尔可夫模型(HMM)是2种常用的设备故障诊断方法,但电动汽车充电桩结构设计复杂且目前积累的故障样本不多,使用上述2种方法均不够理想。针对充电桩故障诊断本身具有的特点,结合AP聚类快速、准确提取故障的特征和HMM强大的故障分类能力,本文提出一种基于AP-HMM混合模型的充电桩故障诊断方法。为了研究充电桩长期工作的状态性质,采用马尔可夫平衡方程组求得充电桩发生故障的稳态概率值。实验结果表明,与传统模型相比,AP-HMM混合模型的充电桩故障诊断学习精度提高了3%以上。本文提出的混合模型具有一定的可行性与普适性,可在一定程度上用于速度要求低但精度要求高的其他电子设备故障诊断。

Affinity propagation （AP） of uncertainty hidden Markov model （HMM） and statistical clustering method are the two commonly used methods for fault diagnosis of equipment. However, since the structure of a electric car charging pile is complex and there are few fault samples on it, the above two methods are not ideal for fault diagnosis. According to the characteristics of charging pile with fault diagnosis, taking into account the AP clustering fast and accurate fault feature extraction and HMM powerful capability of fault classification, a fault diagnosis method of charging pile is presented based on AP-HMM hybrid model in this paper. At the same time, in order to discuss the long-term nature of the charging pile, the Markov equilibrium equations are used to obtain the stable probability of fault. The experimental results verify the correctness of the above theoretical analysis, and the results show that the AP-HMM hybrid model has the advantage of high diagnostic accuracy compared with the traditional model. The hybrid model proposed in this paper has certain feasibility and universality, and can be applied to the fault diagnosis of other electronic equipments with low speed and high precision.