基于数据分析方法的动力电池系统滥用故障诊断

为了提高对于电池滥用故障的检测能力和诊断效率,针对电动汽车动力电池系统以电压、电流和电功率作为主要控制参数的特点,提出了一种基于数据分析方法的动力电池系统滥用故障检测方法。该方法采用局部异常因子(LOF)算法和感知器网络,研究由于电池滥用对电压、电流等数据的一致性和离散分布的影响。结果表明,由于故障造成的异常数据点,其LOF值远大于正常数据,在感知器分类结果中通常输出为“0”。利用局部异常因子算法可以有效找出数据集中的异常数据点,利用LOF算法处理后的数据对感知器进行训练,可以对数据集进行快速分类,进而判断电池是否发生了滥用故障。该方法可为动力电池系统故障检测提供参考。

Abuse Fault Diagnosis Method of Power Battery System Based on Data Analysis Method

An abuse fault detection of power battery system based on data analysis method, aiming at the characteristics that electric vehicle battery system put voltage, current and electric power as its main control parameters, was proposed to improve fault detection capability and diagnostic efficiency for battery abuse failures. By using local anomaly factor(LOF) algorithm and perceptron network, the effects of battery abuse fault on data consistency and discrete distribution of voltage and current were studied. The results showed that the LOF value of the abnormal data points caused by the fault was much larger than the normal data, and the output was usually “0” in the perceptron classification result. Obviously, local anomaly factor algorithm can effectively find the abnormal data points in the data set. Using the data processed by the LOF algorithm to train the sensor, the data set can be quickly classified to determine whether the battery has been abused. Therefore, the method can provide a reference for fault detection of the power battery system.