基于GA-BP神经网络的柴油机NO_x瞬态排放预测

开发了一种柴油机NOx瞬态排放预测研究的方法,该方法采用BP神经网络进行建模,并加入遗算法优化网络权值弥补BP网络缺陷。模型选取影响NOx排放的控制参数作为输入量。网络训练中采用早停止策略防止训练网络过拟合,保证网络泛化能力。模型总排放预测相对误差为3.63%,尤其对加减速等瞬态变化工况的预测效果更优,相对误差小于3%,可以用于NOx瞬态排放预测,并且模型输入参数多为控制参数,有利于对模型进行深入研究。

Predictive Model of Diesel Transient NOx Emission Based on GA-BP Neural Network

A method of prediction was introduced.The method used the BP neural network and added genetic algorithms to optimize the network weights and made up for shortcomings of BP network.The current parameters and historical parameters were selected as input in the model.To ensure network generalization,early stop strategy was exploited to prevent the training network over fitting.The relative error of total emissions prediction was 3.63%,especially the relative error of acceleration and deceleration transient conditions emissions prediction was less than 3%.The results showed that the model could be used to predict transient NOx emission.As the input was mostly control parameters,the model for in-depth study could be carried out.