基于燃烧模型的增压柴油机动态指示转矩预测

Dynamic Indicated Torque Prediction for Turbocharged Diesel Engine Based on Combustion Model

为了提高发动机动态转矩估计精度,提出了基于零维燃烧模型的增压柴油机转矩在线预测方法.首先从增压柴油机动态燃烧特性出发,分析了影响动态指示转矩的因素;通过稳态和动态试验获取燃烧放热率参数,并利用神经网络训练出工况参数与燃烧模型参数的关系式,得到能够预测燃烧放热规律的燃烧模型;然后在此基础上对缸内热力过程进行数值求解和算法简化,实现了在线指示转矩估计,具有良好的实时性.由于该方法解耦了工况参数与指示转矩之间的强非线性关系,获得了较高的预测精度,稳态平均相对误差为2%,,动态平均相对误差为5%,.

For accurately estimating the indicated torque of turbocharged diesel engine under transient operation, a new on-line prediction algorithm based on the 0-D combustion model was proposed. Firstly, the factors that affect transient indicated torque were analyzed through transient combustion characteristics. The experiment under steady and transient operation was conducted to obtain the heat release and state parameters. These data were subsequently used to train the neural networks to get the function of state parameters and heat release. The combustion model can predict heat release in both steady and transient states. The in-cylinder thermodynamic process is numerically calculated using the combustion model. To implement its online, the numerical algorithm was simplified and shows good real-time capability. Due to the decoupling of the strong nonlinear correlation between state parameters and indicated torque, the algorithm possesses high predictability. Bench test result shows the mean average relative error （MARE） of torque estimation is 2% in the steady state, and 5% in transient state.