论文部分内容阅读
根据CD级柴油机的模拟实验与台架试验的基础数据,用人工神经网络的反向传播(BP)方法,对两者的关系进行了研究,给出了人工神经网络的学习速率为0.2,动量因子为0.9,对人工神经网络的拓扑结构也进行了研究,得到了合适的5-7-2拓扑结构及各节点间的权重系数。探讨了用模拟实验数据预测台架试验结果的可能性,通过检验,证明了用人工神经网络方法建立的模型能准确预报柴油机油的台架试验结果。
According to the basic data of CD-class diesel engine simulation test and bench test, the relationship between the two is studied by artificial neural network’s back propagation (BP) method. The learning rate of artificial neural network is 0.2 , The momentum factor is 0.9, the topology of artificial neural network is also studied, and the suitable 5-7-2 topology and the weight coefficient between nodes are obtained. The feasibility of predicting bench test results by using simulated experimental data is discussed. The test results show that the model established by artificial neural network can accurately predict the bench test results of diesel engine oil.