基于高斯模糊和BP神经网络的汽车液压转向系统故障诊断

Fault Diagnosis of Automotive Hydraulic Steering System Based on Gaussian Fuzzy and BP Neural Network

提出了一种结合高斯模糊化和BP神经网络的汽车液压转向系统故障诊断方法。通过采集压力、温度、噪声、流量和泄漏量等信号,并对其进行高斯模糊化处理,将其映射为一组多元模糊集合,用于描述液压转向系统的状态。同时,使用BP神经网络对故障进行分类和回归预测,并通过训练集和测试集验证了模型的准确性和泛化能力。与未高斯模糊化的对照实验相比,高斯模糊化处理的方法表现出了明显的优越性,能够提高预测准确性、降低训练时间和计算量,并减小过拟合的可能性。因此,该方法可以为汽车液压转向系统的故障诊断提供一种有效的解决方案。

A method for diagnosing faults in the hydraulic steering system of automobiles is proposed,which combines Gaussian fuzzification and BP neural network.Signals such as pressure,temperature,noise,flow,and leakage are collected and processed by Gaussian fuzzification to map them into a set of multivariate fuzzy sets for describing the state of the hydraulic steering system.Meanwhile,the BP neural network is used to classify and regress the faults,and the accuracy and generalization ability of the model are verified through training and testing sets.Compared with the control experiment without Gaussian fuzzification,the method of Gaussian fuzzification processing shows obvious superiority,which can improve prediction accuracy,reduce training time and computational complexity,and decrease the possibility of overftting.Therefore,the method can provide an effective solution for fault diagnosis in the hydraulic steering system of automobiles.