某扫雷车扫雷犁电液伺服系统辨识与控制  

Modeling and Control for Mine Sweeping Plough Electro-hydraulic Servo System of Certain Mine-clearing Vehicle

作　　者：高强[1] 金勇 侯远龙[1] 季丽君[1] 

机构地区：[1]南京理工大学机械工程学院,江苏南京210094 [2]总装备部工程兵军代局武汉军代室,湖北武汉430073

出　　处：《南京理工大学学报》2012年第2期238-244,共7页Journal of Nanjing University of Science and Technology

摘　　要：针对某扫雷车扫雷犁电液伺服系统存在非线性及时变性的问题,研究了其辨识和控制方案。提出了离线训练与在线微调相结合的神经网络辨识方案,较好地解决了反向传播BP神经网络易陷入局部最小的问题,加速了网络的收敛速度,且离线训练后的权值参数为合理值,避免了在线微调时的振荡现象。研究了神经网络间接模型参考自适应控制方案,该方案利用神经网络在线辨识器为神经网络控制器实时提供梯度信息,使神经网络控制器的学习修正能够正常进行。仿真研究和样机试验结果证明了所提出辨识和控制方案的有效性和正确性。To control the nonlinearity and time-variation of the mine sweeping plough electro-hydraulic servo system in a certain mine-clearing vehicle,the modeling and control techniques are studied here.First,the neural network identification scheme of the off-line training and on-line minor adjustment is presented.The scheme can offer a better solution to the problem that the back-propagation（BP）neural network is easy to fall into the local minimum and accelerate the convergence of BP algorithm.By keeping the weight value reasonable after the off-line training,the scheme avoids the oscillation when using the on-line minor adjustment.Secondly,the neural network model reference adaptive control is introduced.The on-line neural network identifier offers the real-time gradient information to the neural network controller to guarantee its proper learning and modification.The results of the simulation and the prototype test prove that the proposed modeling scheme and control scheme are effective and suitable.

关 键 词：扫雷车 扫雷犁 电液伺服系统 神经网络 在线辨识 离线辨识 自适应控制 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]