【摘要】 RBF神经网络进行船舶主机燃油喷射系统故障和诊断的过程中存在着精度不高、误诊率高的缺点。针对这种情况,引入以Elman神经网络为基本识别模型,使用改进遗传算法(GA)对网络的权值和阈值进行优化;使用船舶主机燃油喷射系统故障样本对优化后的算法进行训练并对待识别故障样本进行仿真。对比普通Elman神经网络模型、GA-RBF神经网络模型、GA-Elman神经网络模型的诊断结果。仿真结果显示,改进的GA-Elman神经网络不易陷入局部最小值,误差小,在故障诊断方面优于Elman神经网络和GA-RBF神经网络。更多还原

【Abstract】 RBF neural network has the disadvantages of low accuracy and high misdiagnosis rate in fault diagnosis of fuel injection system of marine main engine. In view of this situation, the Elman neural network was introduced as the basic recognition model, and the improved genetic algorithm(GA) was used to optimize the weight and threshold of the network. The fault samples of the fuel injection system of marine main engine were used to train the optimized algorithm and simulate the fault samples. The diagnosis results of Elman neural network model, GA-RBF neural network model and GA-Elman neural network model were compared. The simulation results show that the improved GA-Elman neural network is not easy to fall into the local minimum value with small error, and it is better than Elman neural network and GA-RBF neural network in fault diagnosis.更多还原