基于神经网络的汽车行驶工况识别方法研究

Research on Driving Cycle Recognition Based on Neural Network

混动车辆能量管理实现自适应控制策略的前提是对汽车行驶工况的正确识别。针对此问题提出了一种基于神经网络的车辆行驶工况识别方法。首先通过采集车辆实际行驶路谱,整理出3种典型工况。在此基础上,运用方差分析工具,筛选提取10项特征,用于路谱的辨别,并且整理成用于训练的数据。采用Matlab建立目标神经网络,经过训练和测试,模型识别率达到98.5%。经过综合工况的验证,模型识别率良好,达到预期效果。

In order to realize the adaptive control strategy of hybrid vehicle,the driving cycle recognition is required.A neural network algorithm is proposed to recognize the driving cycle.Firstly,three typical driving cycles are obtained by collecting the actual driving cycles.The 10 characteristic parameters of driving cycle are extracted by the variance analysis,and the training data is compiled.The target neural network was established by the Matlab,the model recognition rate reached 98.5%after training and test.Through the validation of the comprehensive driving cycle,the model has the high identification effect and reaches the expectation.