基于神经网络的行驶工况智能识别方法研究

Research on Intelligent Identification Method of Driving ConditionBased on Neural Network

针对目前工况识别率低降低了能量控制策略在实际行驶中的控制效果的问题,提出了神经网络算法优化行驶工况识别的方法。首先,利用降维后的特征值对典型工况进行分类,构建综合行驶工况。其次,建立模糊C均值聚类、概率神经网络、学习向量化神经网络、BP神经网络的工况识别模型并训练。最后对工况智能识别模型进行仿真验证。结果表明,BP神经网络算法识别效果最好,可优选BP神经网络工况识别模型为设计能量管理策略提供基础。

Aiming at the problem that the low identification rate of current driving condition reduces the control effect of energy control strategy in actual driving condition,a neural network algorithm is proposed to optimize the driving condition identification.Firstly,typical working conditions are classified by characteristic value of the dimensionality reduction and the comprehensive driving condition is constructed.Secondly,condition identification models are respectively established and trained which are fuzzy C-means clustering,probabilistic neural network,learning vector quantization neural network and BP neural network.Finally,intelligent identification models are verified by simulation.The results show that the BP neural network algorithm has the best identification effect,and BP neural network condition identification model can be preferentially selected to provide a basis for designing the energy management strategy.