Optimization and prediction of a novel preignition in hydrogen direct injection engines through experimentation and the Random forest algorithms

Abstract

The potential use of hydrogen in automotive internal combustion engines could aid in the energy transition and contribute to achieving carbon neutrality. This paper has investigated the impact of operating parameters boundary for a novel preignition event on a 2.0L hydrogen direct injection (HDI) engine to address the challenge of abnormal combustion limiting the application of HDI engines. The identified characteristics of the event and its potential harm to the engine has been summarized. Experiments have shown that the novel preignition event predominantly occurred mainly during 170−80°CA BTDC, hence the name PEMC (preignition in the early/mid compression stroke) event, and caused a surge in cylinder pressure. In 77.2 % of the selected cases, the PEMC events resulted in increased NOx emissions. In 93.1 % and 91.5 % of the cases, there was a decrease in brake power and brake thermal efficiency, respectively. Unlike the low speed, high load preignition of gasoline engines, PEMC events were more likely to occur in the torque range of 70–100 Nm and could be monitored in a lean mixture environment with λ = 1.5–3.3. PEMC events were found in the interval of 180−320° CA BTDC at injection timing and were most intense at 240−250° CA BTDC. To prevent PEMC events completely, it is recommended to delay the injection timing beyond 180° CA BTDC. The paper has employed a model based on experimental data and Random Forest algorithms to predict PEMC events. The model demonstrated an accuracy of 98.76 % and 97.69 % for the training and test data, respectively. It has been effective in predicting the occurrence of PEMC events under various operating conditions. The paper has offered an alternative solution for the novel preignition in HDI engines.