基于成长遗传算法的非线性模型预测控制及仿真

Nonlinear Model Predictive Control Based on Growth Genetic Algorithm and Simulation

非线性预测控制在每个控制周期需要求取控制量,其实质为非线性优化问题。标准遗传算法时间消耗较大,难以用于控制周期较小的系统。首先证明了基于种子策略和精英保存策略的遗传算法能够保证闭环控制系统的渐进稳定性;继而模拟自然界成长过程,利用成长算子改进算法框架,并用爬山法进行实现。在具有强烈非线性的连续搅拌釜式反应器模型上进行仿真试验。试验结果表明,在不损失控制效果的情况下,成长遗传算法有效的降低了时间消耗。

To calculate control sequence, nonlinear model predictive control involves optimization to a nonlinear programming problem. Suffered from the unsatisfied time consumption, simple genetic algorithm is not suit for the system with small control cycle. The stabilization of close-loop system with seed and elitism strategies was proved. By simulating the growth process in living nature, the structure of simple genetic algorithm was improved by growth operator, which was realized by the hill climbing method. Simulation results to the continuous stirred tank reactor model, which has strong nonlinear property, show that growth genetic algorithm decreases time consumption and obtains the same performance.