电机与控制学报
電機與控製學報
전궤여공제학보
ECTRIC MACHINES AND CONTROL
2010年
3期
104-108
,共5页
混合动力汽车%神经网络%转矩%估计%最优停止法
混閤動力汽車%神經網絡%轉矩%估計%最優停止法
혼합동력기차%신경망락%전구%고계%최우정지법
hybrid electric vehicle%neural network%torque%estimation%optimal stopping rule
针对混合动力汽车控制系统的开发过程,提出一种应用改进BP神经网络对发动机输出转矩进行估计的方法.根据在发动机实验台中所测得的部分样本数据,将传统的BP网络误差函数进行改进,建立了发动机输出转矩估计模型,并利用最优停止法对网络进行训练,避免了过拟合现象.实验结果表明,利用改进的BP网络对发动机输出转矩进行估计,减轻了网络训练负担,降低了网络训练的误差,提高了发动机输出转矩估计的精确度.
針對混閤動力汽車控製繫統的開髮過程,提齣一種應用改進BP神經網絡對髮動機輸齣轉矩進行估計的方法.根據在髮動機實驗檯中所測得的部分樣本數據,將傳統的BP網絡誤差函數進行改進,建立瞭髮動機輸齣轉矩估計模型,併利用最優停止法對網絡進行訓練,避免瞭過擬閤現象.實驗結果錶明,利用改進的BP網絡對髮動機輸齣轉矩進行估計,減輕瞭網絡訓練負擔,降低瞭網絡訓練的誤差,提高瞭髮動機輸齣轉矩估計的精確度.
침대혼합동력기차공제계통적개발과정,제출일충응용개진BP신경망락대발동궤수출전구진행고계적방법.근거재발동궤실험태중소측득적부분양본수거,장전통적BP망락오차함수진행개진,건립료발동궤수출전구고계모형,병이용최우정지법대망락진행훈련,피면료과의합현상.실험결과표명,이용개진적BP망락대발동궤수출전구진행고계,감경료망락훈련부담,강저료망락훈련적오차,제고료발동궤수출전구고계적정학도.
According to the development process of the control system for Hybrid Electic Vehicle,the estimation method to the torque of the engine is presented.Based on partial sample experiment result which came from engine experiment frame,the error function of the traditional BP neural network was improved.Then the model of engine torque was built.Meanwhile,optimal stopping rule was used to avoid over-fitting.The experiment results indicated that improved BP neural network not only alleviates burden of the network training,reduces error of the network training but also improves precision of engine output torque estimation.
