Optimization of Hydrogen-fueled Engine Ignition Timing Based on L-M Neural Network Algorithm

Lijun Wang (North China University of Water Resources and Electric Power)
Yuan Liu (North China University of Water Resources and Electric Power)
Yahui Liu (North China University of Water Resources and Electric Power)
Wei Wang (North China University of Water Resources and Electric Power)
Yanan Zhao (North China University of Water Resources and Electric Power)
Zhenzhong Yang (North China University of Water Resources and Electric Power)

TELKOMNIKA (Telecommunication Computing Electronics and Control)
Vol 14, No 3: September 2016

In view of the improvement measures of the optimization control algorithm for the ignition system of the hydrogen-fueled engine, the L-M neural network algorithm, Powell neural network algorithm and the traditional BP neural network algorithm are used to optimize the ignition system. The results showed that L-M algorithm not only can accurately predict the hydrogen-fueled engine ignition timing, but also has high precision, high convergence speed, a simple model and other outstanding advantages in the training process, which can greatly reduce the workload of human engine bench tests. Only a small amount of engine bench test is carried out, and the obtained sample data can be used to predict the ignition timing under the whole working conditions. The mean square error of the optimization results based on L-M algorithm arrives at 0.0028 after 100 times of calculation, the maximum value of absolute error arrives at 0.2454, and the minimum value of absolute error arrives at 0.00426.