基于PSO-控制变量参数化混合策略的间歇化工过程优化控制

摘要

As a gradient search algorithm for dynamic optimization of chemical process, the efficiency of control vector parameterization depends on the initial given trajectory deeply. At present, the initial trajectory is usually set at the boundary value or the intermediate value, which does not have enough scientific reason and it affects the convergence speed of the algorithm. To solve this problem, a hybrid strategy combined with particle swarm optimization and control vector parameterization method is proposed in this paper, it uses particle swarm optimization to achieve the value of control variables before employing the method of control vector parameterization to reoptimize the process. The two-layer optimization hybrid strategy improves the convergence speed of the control vector parameterization method and the precision of the particle swarm optimization. The hybrid strategy is applied to two examples of batch chemical process optimization control, and the simulation results show that the algorithm is feasible and effective for solving dynamic optimization problems of chemical process.%控制变量参数化方法作为一种化工过程动态优化的梯度搜索算法,其求解效率过于依赖初始给定轨迹.目前初始轨迹一般都是设定在边界值或中间值,缺乏科学依据,从而大大影响了算法的收敛速度.针对这一问题,提出了一种粒子群优化(PSO)与控制变量参数化方法混合的策略,首先利用粒子群优化对间歇化工过程最优控制量进行求解,结果作为控制变量参数化方法初始给定轨迹,进行二次优化.双层优化的混合策略提高了控制变量参数化方法的收敛速度和粒子群优化算法的求解精度.将混合策略应用于两个间歇化工过程优化控制实例,仿真结果表明了该算法对求解化工过程动态优化问题具有可行性和有效性.