基于改进AFSA优化BP神经网络的汽车故障诊断方法

Automobile fault diagnosis based on BP optimized by improved artificial fish swarm algorithm

为提高汽车故障诊断的准确率,提出一种基于改进人工鱼群算法优化BP神经网络的汽车故障诊断方法。在分析BP神经网络基本原理和局限的基础上,利用改进人工鱼群算法对BP神经网络参数寻优,从而进一步提高汽车故障诊断的准确率。仿真结果表明,基于改进人工鱼群算法优化BP神经网络的迭代次数为13次,少于未经参数优化的BP神经网络迭代次数,并可有效识别11种不同类型的汽车故障。研究表明,基于改进人工鱼群算法优化BP神经网络在提高汽车故障诊断的准确率上可行。

In order to improve the accuracy of automobile fault diagnosis,an automobile fault diagnosis method based on improved artificial fish swarm algorithm and optimized BP neural network is proposed.On the basis of analyzing the basic principles and limitations of BP neural network,the improved artificial fish swarm algorithm is used to optimize the parameters of BP neural network,so as to further improve the accuracy of automobile fault diagnosis.The simulation results show that the iteration times of BP neural network optimized by improved artificial fish swarm is 13 times,which is less than that of the BP neural network without parameter optimization.The BP neural network model is applied to automobile fault diagnosis,thus 11 kinds of automobile faults can be effectively identified.Through the above results,it is proved that the proposed automobile fault diagnosis model based on BP neural network optimized by improved artificial fish swarm algorithm is feasible.