低气压环境下柴油机涡轮增压器转速实车估算方法  

Measuring the Turbosupercharger Speed of Actual Vehicle in Low Pressure Conditions

作　　者：王宪成[1] 杨莹[1] 何星[1] 刘海涛[2] 

机构地区：[1]装甲兵工程学院机械工程系,北京100072 [2]武警沈阳指挥学院,辽宁沈阳110113

出　　处：《装甲兵工程学院学报》2011年第3期35-38,共4页Journal of Academy of Armored Force Engineering

基　　金：军队科研计划项目

摘　　要：针对涡轮增压器转速实车测量中存在安装困难、精度低的问题,提出了利用RBF神经网络和BP神经网络估算涡轮增压器转速的方法。分别建立了基于RBF神经网络和BP神经网络增压器转速估算模型,通过与台架试验测试样本比较,模型误差分别2.25%和2.27%。同时,RBF神经网络较BP神经网络,具有训练次数少,收敛速度快、结果稳定的优点,更适合实车涡轮增压器转速估算。There are some problems when measuring turbosupercharger speed in actual vehicle.Two methods for mesauring the turbosupercharger speed by means of the RBF neural network and BP neural network are introduced.Accordingly,the RBF neural network and BP neural network estimation modles are set up.Compared with the bench test sample,the model errors are 2.25% and 2.27%.Meanwhile,in comparation with the BP neural network,the RBF neural network has less training times and fast convergence speed and stable result,which is more suitable for the actual vehicle turbosupercharger speed estimation.

关 键 词：人工神经网络 增压器转速 估算 

分 类 号：TK421.8[动力工程及工程热物理—动力机械及工程]