并联混合动力汽车的发动机转矩估计

提出采用神经网络对发动机转矩进行估计的方法,以发动机台架性能实验所得数据为基础,在Matlab中建立了多种类型的神经元和层数的发动机输出转矩模型.在网络模型的建立过程中,提出了一种新的训练方式,通过对不同模型的训练及结果分析,筛选出了估计精度和泛化能力都较为理想的模型,并以此模型进行了实际的转矩估计,取得了较满意的效果.

Estimation of Engine Torque for Parallel Hybrid Electric Vehicle

A method of estimating engine torque by using neural network is presented in this paper. Based on the data of engine bench test, several models for engine output torque with different types of neural cells and different number of layers are established in Matlab. During the process of network model establishment, a new training method is adopted. Through the training on different models and analyses on their results, a final model with ideal estimated accuracy and generalization ability is selected, Using this model, the actual torque is estimated with a satisfactory effect obtained