Security Performance of Integrated Chassis Structures of New Energy Vehicles Based on Convolutional Neural Networks

Abstract

To address the issue of new energy vehicles' inadequate safety performance, the study creates a hybrid algorithm by combining stacked auto encoders with convolutional neural networks. It then suggests a model for diagnosing chassis mechanism faults based on the hybrid algorithm. The fusion method outperforms the other algorithms employed in the comparison test, having an accuracy and average error of 98.01% and 5.00%, respectively, according to a comparative examination of the performance of the hybrid algorithm based on convolutional neural networks. Additionally, it is discovered that the algorithm's running time is 0.063s, which is faster than the times of the other comparison algorithms. The accuracy and loss values of the integrated chassis structural defect diagnosis model, which was included in the study for new energy vehicles, were later discovered to be 98.53% and 0.041 respectively, which were better than the comparison models. According to the aforementioned results, the proposed integrated chassis structure fault diagnosis model for new energy vehicles has higher diagnostic accuracy than the conventional model, which has significant implications for the integrated chassis structure of new energy vehicles' ability to perform safely.