基于多目标遗传算法的汽轮机转速PI控制器参数优化

Multi-objective PI Controller Tuning for a Steam Turbine Speed Governing System Using NSGA-Ⅱ

针对常规PID参数整定方法不能兼顾系统快速性、稳定性和准确性要求的缺陷,提出了一种基于非支配排序多目标遗传算法-Ⅱ(Non-dominated Sorted Genetic Algorithm-Ⅱ,NSGA-Ⅱ)和多属性决策技术(Multiple Attribute Decision Making,MADM)的PID参数的优化设计方法。以系统的超调量、调节时间和时间乘以误差绝对值积分(ITAE)指标最小作为优化目标,以频域稳定性为约束条件,以PID参数为设计变量,建立多目标优化模型。利用NSGA-Ⅱ算法求出Pareto最优解集,采用逼近理想解的排序方法(Technique for Order Preference by Similarity to an Ideal Solution,TOPSIS)进行Pareto最优解排序,进而得到最终的最优解。以某型工业汽轮机转速PI控制器参数的多目标优化设计为仿真算例,验证了该文设计方法的有效性和优越性。

It is difficult for the conventional PID controller tuning methods to meet requirements for quickness, stability and accuracy. In order to solve this problem, a tuning method based on Non-Dominated Sorted Genetic Algorithm-If （NSGA-Ⅱ ） and Multi-Attribute Decision Making （MADM） technique was proposed in the paper. A multi-objective opti- mization model was developed, in which minimum overshoot, setting time and ITAE criterion were taken as objectives, frequency stability serve as constraints, and PID parameters were the variables. The Pareto optimal solutions were ob- tained using NSOA-lI and ranked by Technique for Order Preference by Similarity to an Ideal Solution （TOPSIS）, and then the final optimal solution was obtained. The proposed Optimization method was exemplified in the case of the PI con- troller parameters tuning for an industrial steam turbine speed governing system. Simulation results demonstrated the ef- fectiveness and superiority of the proposed method.