为解决现有方法对1 000 MW机组给水系统建模复杂、算法收敛速度慢、精度低等问题，提出一种改进遗传算法融合混沌粒子群算法(Genetic Algorithm-Chaotic Particle Swarm Optimization，GA-CPSO)。首先，粒子群算法(Particle Swarm Optimization，PSO)中引入了自适应权重和收缩因子，提升粒子寻优能力；在一维Logistic的基础上提出二维Logistic混沌映射，避免寻优过程中陷入局部最优解；采用轮盘赌选择方法，选取粒子进行下一步的遗传算法优化，提升了全局寻优能力。其次，通过实验仿真数据和现场实际数据验证了改进GA-CPSO算法的精度。将该算法用于1 000 MW机组给水系统，建模精度提高了88.65%，仅需要迭代7次左右即完成收敛。然后，利用数据中加干扰实验进一步挖掘改进GA-CPSO算法的抗干扰能力。实验表明：加入外部大扰动建模误差仅有0.385，算法抗干扰能力强。最后，用皮尔逊相关系数方法验证了机组直流阶段模型间的相关性，相关系数达到了0.9以上，可用一个模型代表。

To solve the problems of complicated modeling, slow convergence speed and low accuracy of the existing methods for 1 000 MW unit feedwater system, an improved genetic algorithm-chaotic particle swarm optimization (GA-CPSO) is proposed. First, adaptive weights and shrinkage factors are introduced in the particle swarm optimization (PSO) algorithm to improve the particle search capability; a twodimensional Logistic chaotic mapping is proposed on the basis of onedimensional Logistic to avoid falling into local optimal solutions in the search process; a roulette wheel selection method is used to select particles for the next step of genetic algorithm optimization and to improve the global optimizationseeking ability. Second, the accuracy of the improved GA-CPSO algorithm is verified through experimental simulation data and actual field data.This algorithm is used for the feedwater system of 1 000 MW unit to improve the modeling accuracy by 88.65% and to complete the convergence after only 7 iterations. Then, the improved GA-CPSO algorithm is further explored for its antidisturbance capability using the experiment of adding disturbances in the data.The experiment shows that the modeling error of adding large external disturbances is only 0,385 and the algorithm is highly resistant to disturbances. Finally, the correlation between the DC stage models of the units is verified by Pearson correlation coefficient method, and the correlation coefficient reachs more than 0.9, which can be represented by one model.

TK39

上海市2021年度“科技创新行动计划”科技支撑碳达峰碳中和专项（21DZ1207302）;国家自然科学基金青年科学基金项目（51607111）；上海市科学技术委员会科研计划项目（19DZ1205700）