记忆径向基神经网络在冷轧液压自动位置系统的优化控制

Optimal Control in Cold Rolling Hydraulic Automatic Position System Based on RBF Neural Network with Memory Factor

针对冷轧液压自动位置控制系统多变量、强耦合、高阶次和时变性等特点，提出一种引入记忆因子的径向基函数神经网络在线自适应调节PID参数的系统。为提高网络精度，利用改进的混洗蛙跳算法离线全优化记忆径向基神经网络，在获得网络结构的同时得到初始参数，避免网络模型训练的繁琐，并利用测试函数证明优化后的网络具有良好的逼近能力。然后利用优化后记忆径向基神经网络的自校正功能在线细调PID参数，仿真结果表明，该控制系统跟踪快、超调小、适应性强，控制品质优于传统PID和普通径向基神经网络PID控制方法。

Aiming at the characteristics for cold rolling hydraulic aotomatic positom control system with multi-variable, strong coupling, higher order and time-varying, a radial basis function neural network introduced in memory factor is proposed which can adaptive tune PID parameters online. To improve network accuracy, improved shuffled frog leaping algorithm is used to offline fully optimize radial basis fanction neural network with memory factor, which can obtain the network structure and initial parameters simultaneously, and avoid the tedious network model training. And the test functions are applied to demonstrate the optimized network has good approximation ability. Then the optimized radial basis fanction neural network with memory factor that has self-correction function is used to finely tune PID parameters online, and simulation results show that the control system with fast track, small overshoot, strong adaptability is better than the traditional PID control and general radial basis function neural network PID control methods, which has practical value.