基于MATLAB神经网络的柴油机涡流比的预测

Investigation on the Predicting of Diesel Engine Intake Swirl Ratio Based on Neural Network Toolbox of MATLAB

研究利用MATLAB神经网络工具箱图形用户界面(GUI)来开展柴油机进气涡流比的预测。首先,利用MATLAB神经网络工具箱图形用户界面(GUI)建立了生产工艺参数与进气涡流比之间关系的数学模型,经过反复训练之后利用该网络来预测进气涡流比。随后,将该模型用于玉柴G2000气缸盖进气涡流比的预测时,当要求相对误差在4%时,命中率可达60%;相对误差在10%时,命中率可达90%,从而为控制进气涡流比提供了一套行之有效的方法。

This paper investigates the application of GUI of Neural Network Toolbox of MATLAB in predicting intake swirl ratio for diesel engines. First, a mathematical model between the process parameters and the intake swirl ratio is set up with GUI of Neural Networks Toolbox of MATLAB. When the model is verified, it can be used to predict the intake swirl ratio. According to this model, experiments are conducted in Yuchai company, and a large number of experiments data are obtained. When the required relative error is limited to 4% ,the accuracy can reach 60%; and when the required relative error is limited to 10%,the accuracy can reach 90%. By using this method, the diesel engine intake swirl ratio can be controlled in a higher accuracy.