基于遗传BP神经网络的电控发动机故障诊断研究

Fault Diagnosis of Electronic Control Engine Based on Genetic BP Neural Network

为提高发动机故障诊断的正确率与精确度,提出遗传算法和BP神经网络相结合的故障诊断模型。将发动机部分尾气信息和传感器数据作为BP神经网络诊断模型的输入变量,利用遗传算法的全局搜索能力优化BP神经网络的初始权值和阈值,采用优化后的BP神经网络建立发动机故障的诊断模型。实验结果表明该诊断模型可提高发动机故障诊断的正确率。

To improve the correctness and accuracy of the engine fault diagnosis, a fault diagnosis modelbased on BP neural network combining genetic algorithm was proposed. With the engine emissions in-formation and sensors data as the input variables of BP neural network diagnosis model, BP neural net-work,s initial weights and thresholds were optimized using the global search ability of genetic algo-rithm. A model of engine fault diagnosis was built by the optimized BP neural network. The simulationresults show that the proposed method can improve the accuracy of engine fault diagnosis.