甲醇-柴油双燃料发动机喷醇 MAP 优化

摘要

在优化双燃料发动机喷醇 MAP中应用了一种新方法，根据4B26甲醇-柴油双燃料发动机台架试验数据，用 Matlab 编写 BP 人工神经网络（BP-ANN）程序模拟发动机输入输出对应关系，利用遗传算法（GA）程序对神经网络模型中的喷醇 MAP 寻优。仿真结果显示，人工神经网络能简单快速地模拟发动机的实际运行情况。人工神经网络、遗传算法联合运行，能找到不同转速不同负荷下的最佳喷醇量。稳态工况试验表明，在保证双燃料发动机循环变动不致过大的情况下，应用该方法得到喷醇 MAP 的发动机醇耗比原机提高24．3％，炭烟排放降低7．5％，同时消除了循环变动。%According to the dual-fuel engine bench experimental data,engine input and output relations were simulated with the compiled back propagation artificial neural network program with Matlab software and the methanol injection MAP was op-timized with genetic algorithm program.The results show that artificial neural network can simulate engine actual running easi-ly and quickly.With the two network programs,the best methanol injection quantities at different loads and speeds can be de-termined.Further steady state experiment shows that the engine with a new MAP gets no cyclical fluctuation,and consumes 24.3% more methanol and gets 7.5% lower soot emission under steady operating conditions.