The invention relates to the technical field of new energy automobile anomaly detection, in particular to a new energy automobile electric drive system driving state monitoring method based on machine learning, and the method comprises the steps: firstly screening a reference clustering center point according to the distribution density and relative position of voltage temperature data points in a neighborhood; further obtaining a clustering influence weight according to a data point density change on a connecting line between the voltage temperature data point and the reference clustering center point and a connecting line slope; the clustering distance during clustering is influenced by the clustering influence weight, so that abnormal voltage and temperature data points can be distributed to abnormal voltage and temperature data clusters, namely, a more accurate clustering analysis result is obtained; therefore, a more accurate clustering anomaly degree is obtained according to the data point density and the relative outlier condition of the clustering cluster; and the effect of monitoring the driving state of the electric driving system of the new energy automobile according to the clustering abnormal degree is better.
本发明涉及新能源汽车异常检测技术领域,具体涉及基于机器学习的新能源汽车电驱系统驱动状态监测方法,该方法首先根据电压温度数据点在邻域分布密集程度以及相对位置,筛选参考聚类中心点;进一步地根据电压温度数据点与参考聚类中心点之间连线上的数据点密度变化,以及连线斜率得到聚类影响权重;通过聚类影响权重影响聚类时的聚类距离,使得异常的电压温度数据点更能分配到异常的电压温度数据聚类簇中,也即得到更加准确的聚类分析结果;从而使得根据数据点密集程度和聚类簇相对离群情况,得到更加准确的聚类异常程度;使得根据聚类异常程度进行新能源汽车电驱系统驱动状态监测的效果更好。
New energy automobile electric drive system driving state monitoring method based on machine learning
2025-06-17
Patent
Electronic Resource
Chinese
DRIVING BATTERY SYSTEM FOR AUTOMOBILE AND AUTOMOBILE HAVING ELECTRIC DRIVE EQUIPMENT
European Patent Office | 2022
|Driving motor rotation state monitoring method and system for electric automobile
European Patent Office | 2020
|Driving state monitoring method and system, and automobile
European Patent Office | 2021
|ELECTRIC AUTOMOBILE DRIVING SYSTEM AND ELECTRIC AUTOMOBILE
European Patent Office | 2024
|