摘要
针对某炼油厂油品车间柴油调和过程这个多输入多输出复杂对象进行了神经网络内模控制的仿真研究,其中在线优化算法采用线性规划的方法。神经网络预测控制正是克服了传统控制思想的束缚,通过对象的输入输出特性建立对象的数学模型,而不必通过复杂的系统辨识来建立过程的模型。对仿真结果进行了比较,结果表明神经网络预测控制算法对复杂对象具有较好的控制作用。
A neural network predictive control simulation is made, aimed at a diesel oil blending process which is a strong non-linearity, multi-perturbation, MIMO system, a linear programming is applied on optimal calculation. Conquering the traditional control idea model predictive control (MPC) establish object model through the input-output characteristic. Simulation results show that the predictive control strategy can takes on the good performances of both set-point tracking and anti-perturbations.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第24期5931-5934,共4页
Computer Engineering and Design
关键词
神经网络
预测控制
算法
内模控制
线性规划
neural network
model predictive control
arithmetic
predictive control
linear programming