具有混合约束的多变量系统的优化控制

给出了多变量预测控制的算法描述 ,解决了工业生产中存在的混合约束问题。并运用协调优化的原则对带有硬约束的操作变量进行了协调。当找不到满足所有约束条件的可行解时 ,对被控变量进行约束软化处理。采取线性和二次型相结合的惩罚函数 ,对预测时域上每个时刻的激活值进行惩罚 ,不仅可以保证可行解的存在 ,而且改进了控制系统的性能。理论分析证明 ,该算法能够保证闭环系统的稳定性 ,并具有较强的鲁棒性。仿真结果也证实了算法的有效性。

Optimized control of multivariable process with mixed constraints

An optimized control strategy of multivariable process in the presence of hard and soft constraints is presented. On the basis of an algorithm for a multivariable constrained model predictive control, a coordinate scheme is proposed to construct a coordinated predictive control system. When the manipulated variables are subject to hard constraints, the priorities of the manipulated and controlled variables can be used to adjust the tuning means and controlled objectives. In order to ensure the feasibility of the constrained control problem, a more effective constraint softening approach is adopted. The objective penalizes the sum of both linear and quadratic measures of constraint violations of the controlled variables. Some theoretical results prove that this algorithm can garantee the closed loop stability and robustness of the control system. Simulated results on the shell control problem demonstrate that the optimized control algorithm possesses a good performance.