神经网络内模控制在汽车曲轴圆角滚压系统中的应用  

Use of Neural Network Internal Model Control in Automotive Crankshaft Fillet Rolling System

作　　者：张锐[1] 胡荣强[1] 

机构地区：[1]武汉理工大学自动化学院,湖北武汉430070

出　　处：《材料开发与应用》2009年第6期89-92,共4页Development and Application of Materials

摘　　要：介绍了汽车发动机曲轴滚压强化和校直的先进加工工艺,提出了圆角滚压强化校直工艺中的3个关键问题。针对滚压强化过程压力控制中压力波动的问题,分析了扰动产生的原因,采用了内模控制方法克服压力扰动。构建了一个基于神经网络建模的内模控制系统框架,选用了去饱和特性后的PID神经网络与按拍延迟线组成辨识器来建立比例液压系统模型,推导了去饱和特性后的PID神经网络等效的脉冲传递函数形式并给出了逆模型,指出了该神经网络的应用范围,也指出了在液压系统建模过程中为获得有效数据而应注意的问题。将神经网络内模控制应用于曲轴圆角滚压机床液压系统中的比例压力阀的输出压力控制上,解决了由于滚压动钳摆动时的机械惯性力干扰引起的压力波动问题。This paper reviewed the advanced processing technique of automobile engine crankshaft wiling aggrandizement and alignment. Three key problems about straightening and alignment, of fillet rolling process was put forward. For the control of pressure fluctuation pressure in rolling process, the causes of pressure disturbance was analyzed and the internal model control meth- od was proposed to overcome stress disturbance. A neural network modeling of internal model control system framework was constructed, the PID neural network removed the saturation was chcosen and delay line identifier to proportion hydraulic system model was established. Equivalent pulse transfer function of PID neural network removed saturated properties was deduced and its inverse model was established. The application scope of the neural network was pointed out, and in order to obtain the effective data, the points to which some attention should be paid were also covered. Neural network internal model control was applied to control output pressure of proportional pressure valves of crankshaft rounded rolling machine in the hydraulic system, so the pressure fluctuations problem was solved, due to the interference caused by mechanical force of the swing of rolling dynamic clamp.

关 键 词：圆角滚压 内模控制 神经网络 M序列 逆模型 

分 类 号：TG31[金属学及工艺—金属压力加工]