基于人工神经网络预测柴油机油台架试验结果

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。

Pridiction of Engine Test of Disel Engine Oil with ANN Method

According to the data of bench test and engine test of CD grade engine oil, the relationship between engine test and bench test was studied by back-propagation(BP) method of artificial neural network(ANN). The learning rate of ANN was 0.2, and the mementum factor of ANN was 0.9. The topology of ANN was discussed.Appropriate 5-7-2 topology and weights among nodes were obtained.The feasibility to predict the results of engine test from the data of bench test was also studied.It is shown that the model of ANN method can predict results of engine test of CD grade engine oil with much accuracy.