一种适合于电力系统机组组合问题的混合粒子群优化算法

该文针对机组组合问题，提出了一种新的混合粒子群优化算法。该算法包含3个重要的方面：通过在算法迭代过程中对松弛后的0、1变量与机组有功出力变量并行地进行优化，避免了由于决策变量过多造成的维数灾难题；引入变动阈值，解决了在寻优过程中粒子的值出现振荡时可能会丢失机组有效启停状态的问题：在粒子群优化算法中引入启发式变异技术，有效地处理了机组启、停时间的约束并提高了粒子群优化算法的全局收敛能力。文中通过对2个算例的计算及与其他算法进行的比较结果，验证了所提出的混合粒子群优化算法具有更好的全局收敛性。

A HYBRID PARTICLE SWARM OPTIMIZATION METHOD FOR UNIT COMMITMENT PROBLEM

This paper proposes a new hybrid particle swarm optimization method for unit commitment problem.The new method includes three important aspects: to employ parallel optimization of relaxed zero-one variable and the active power continuous variable to overcome the curse of dimensionality problem;to introduce the variable threshold to ensure all unit operating states taken into account when some zero-one variables are oscillated in iterative optimization;to construct heuristic mutation operation to cope with the start-up and shut-down ramp constraints and to improve the global convergence of the method efficiently.Compared with other methods,numerical simulation results demonstrate that the hybrid particle swarm optimization method possesses strong global convergence capability.