CAJ | 학술논문

针对标准粒子群算法易陷入局部最优、收敛过早的缺陷，提出了一种模糊自修正粒子群算法。通过利用模糊推理机制建立了粒子适应度值隶属度函数，在每次寻优过程中，使得各粒子根据自身当前适应度隶属度函数值来修正惯性权重的取值，而不是把惯性权重作为全局变量，对同一代粒子使用相同的惯性权重；这充分考虑了各粒子自身的性能，可以进一步改善早熟的缺陷，增强全局搜索能力，从而可以获取更好的目标值。将该算法用于求解电力系统经济负荷分配问题，兼顾考虑了燃料成本和环境成本；在求解此问题时，为了更精确地处理功率平衡约束，根据寻优过程中等式约束偏差量的大小不断调整罚系数取值，并以此建立相应的罚函数。算例结果表明，模糊自修正粒子群算法对比标准粒子群算法有较强的全局搜索能力，有更可靠的优化计算结果，进而体现了该方法的有效性和优越性。
침대표준입자군산법역함입국부최우、수렴과조적결함，제출료일충모호자수정입자군산법。통과이용모호추리궤제건립료입자괄응도치대속도함수，재매차심우과정중，사득각입자근거자신당전괄응도대속도함수치래수정관성권중적취치，이불시파관성권중작위전국변량，대동일대입자사용상동적관성권중；저충분고필료각입자자신적성능，가이진일보개선조숙적결함，증강전국수색능력，종이가이획취경호적목표치。장해산법용우구해전력계통경제부하분배문제，겸고고필료연료성본화배경성본；재구해차문제시，위료경정학지처리공솔평형약속，근거심우과정중등식약속편차량적대소불단조정벌계수취치，병이차건립상응적벌함수。산례결과표명，모호자수정입자군산법대비표준입자군산법유교강적전국수색능력，유경가고적우화계산결과，진이체현료해방법적유효성화우월성。
According to the shortage that particle swarm optimization (PSO) algorithm easily falls into local optimum and premature convergence, a fuzzy self-correction particle swarm optimization algorithm is proposed. By using the fuzzy reasoning mechanism, a particle fitness membership function is established, which makes the particles base on their current fitness membership function values to modify the value of inertia weight in the process of optimization, instead of seeing the inertia weight as a global variable, then a generation of particles use the same inertia weight. This optimization fully considers the features of the particle itself, which can further improve the defect of prematurity, enhance the global search ability and get a better target value. The algorithm is used to solve the economic load distribution problems in power system, both considering fuel cost and environmental cost. In solving this problem, to exactly deal with power balance constraints, it uses the size of the deviation value of equality constraint in the optimization process to constantly adjust the value of penalty coefficients, and then establishes corresponding penalty function. Numerical example results show that the proposed algorithm has strong global search ability and more reliable optimization calculation results compared to the standard particle swarm algorithm, which shows the effectiveness and superiority of this method.