基于灰色理论与神经网络的柴油机相继增压系统故障预测与诊断

摘要

增压柴油机是一个具有不确定性的复杂系统,灰色理论是一种处理不确定性的理论.为了保证柴油机的可靠运行,使用灰色预测理论与神经网络技术对柴油机相继增压系统进行了故障预测与诊断.采用MAT-LAB语言编制了灰色预测程序并训练了神经网络,训练好的神经网络具有良好的泛化性.采用GM(1,1)模型对相继增压柴油机运行参数进行预测,对于波动数据,使用改进GM(1,1)模型进行预测,预测参数与试验数据的相对误差均在5%以内.预测参数送入已培训好的神经网络中,对柴油机相继增压系统可能发生的故障进行预测与诊断.%Turbocharged diesel engine is a complex system which involving the uncertainty. Grey theory is one of the methods to deal with systems with uncertain information. To ensure a reliable diesel engine operation, the grey forecast theory and neural network were used to forecast and diagnose the faults of sequential turbocharging system. MATLAB language was used to train the neural network and write program of grey forecast. The neural network has good generalization and is suitable to diagnose faults of sequential turbocharging system. Parameters of turbocharged diesel engine are forecasted by GM (1,1) model, then the improved GM(1,1) models are used to forecast fluctuant parameters with relative deviation under 5% between the forecasted data and experimental data. Furthermore, the forecasted parameters are inputted into the artificial neural networks to forecast and diagnose the faults of sequential turbocharging system.