Abnormal fault detection of automotive miniaturized start-stop starter based on AdaBoost algorithm

With the wide application of automotive miniaturized start-stop starter in new energy vehicles, abnormal fault detection becomes the key to ensure its safety and reliability. In this paper, an effective abnormal fault detection method is proposed based on AdaBoost algorithm. First, the operating signals of the miniaturized start-stop starter are feature extracted by combining time domain and frequency domain analysis, and the accuracy of the features is enhanced by using the fuzzy entropy algorithm. Subsequently, the AdaBoost algorithm is used to construct a strong classifier to improve the accuracy and robustness of fault detection. The experimental results show that the proposed method performs well in abnormal fault detection, with F-score values above 0.95, which are significantly higher than other comparison methods (such as the affiliation function algorithm, the sigmoid function algorithm, and the threshold detection algorithm), and the fault detection time is significantly shortened to 6 seconds, which suggests that the method has a high degree of efficiency and reliability in practical applications.