内燃机油清净剂复配实验的模拟与预测  

The Simulation and Forecast of the Experimentation on the Lustration Complex Additives of Gasoline Engine Oil

作　　者：仝秋红[1] 王毓民[1] 姚顽强[2] 

机构地区：[1]长安大学汽车学院,陕西西安710064 [2]西安科技大学测绘工程系,陕西西安710054

出　　处：《润滑与密封》2006年第8期109-111,共3页Lubrication Engineering

摘　　要：探讨了汽车内燃机油清净剂复配实验的模拟与预测。首先实验研究了T102、T106、T113的复配性能,获得原始数据,然后采用附加动量法、变学习率算法以及动量法与自适应学习率组合算法分别对优化后的BP神经网络进行训练,对实验进行模拟与预测。结果表明,采用附加动量法无法达到预测目标,而采用动量法与自适应学习率组合算法对清净剂复配实验有较好的模拟与预测。The simulation and forecast of the experimentation on the lustration complex additives of gasoline engine oil was discussed. The experiments of T102 , T106 , T113 complex additives were made to obtain the complex properties. By using the Momentum arithmetic, Variable Learning Rate arithmetic and the combining of Momentum and Variable Learning Rate arithmetic, the optimized BP nerve network was trained to simulate and forecast the experiment. The conclu- sion shows that using Momentum arithmetic can not reach the aim, and the combining of Momentum and Variable Learning Rate arithmetic can give better simulation and forecast of the lustration complex additives experimentation.

关 键 词：内燃机油 清净剂复配 附加动量算法 自适应学习率算法 BP网络 

分 类 号：TK418.9[动力工程及工程热物理—动力机械及工程] TP183[自动化与计算机技术—控制理论与控制工程]