基于改进MOPSO的MMC-HVDC控制器PI参数分层优化

Layered Parameter Optimization of MMC-HVDC PI Controller Based on Improved MOPSO

针对模块化多电平换流器型高压直流输电工程系统控制目标多、多个PI控制器之间相互影响、PI参数优化困难的问题,提出一种基于改进多目标粒子群算法的MMCHVDC控制器PI参数分层优化方法。为了提高种群的多样性同时免于粒子陷入局部最优解,对外部存储器中的粒子进行变异操作;为了改善算法收敛性并获得良好的Pareto前沿,提出一种基于隶属度函数的领导粒子选取方法;为了适应MMC-HVDC控制系统结构,对多个控制器PI参数进行分层优化。本文实现了MATLAB与PSCAD/EMTDC联合调用和数据交互。算法性能分析和电磁暂态程序仿真证明了本文所提出方法可以快速收敛并且获得均匀分布的Pareto前沿,有效改善系统的动态性能。

For MMC-HVDC（modular multilevel converterhigh voltage direct current,MMC-HVDC）controller optimization,such problems as multiple control objectives,the relative influence among multiple PI controllers and the difficulty of PI parameter optimization are obvious.In this paper,layered parameter optimization of MMC-HVDC PI controller is proposed based on improved MOPSO method.In order to increase the diversity of the particles to avoid the population sinking into the local optimal solution,a mutation operation is introduced to the particles in the external memory.In order to improve the convergence of the algorithm and obtain a good Pareto frontier,a leading particle selection method is proposed based on membership function.Meanwhile,a layered parameter optimization method is proposed to fit with the structure of MMC-HVDC controller.The combined invocation and data interaction between MATLAB and PSCAD/EMTDC are implemented in this paper.Algorithm performance analysis and electromagnetic transient simulation results show that the proposed method converges fast and can obtain the uniform Pareto frontier.It can also improve the dynamic performance of the system effectively.