改进粒子群算法及其在输电网规划中的应用

根据粒子群算法收敛性受初始粒子分布影响较大的特点，结合非线性单纯体法(nonlinear simplex method NSM)，提出了一种粒子群初始化方法。该方法克服了随机初始化粒子时不能保证粒子合理分布的缺点，对提高初始粒子质量、加速PSO算法的收敛速度起到了有效的作用；同时分析了罚函数形式的适应度函数使迭代过程产生振荡的现象，根据对偶拉格朗日法，提出了一种交叉迭代法，克服了罚函数法的不足。最后通过算例证明了这2种方法应用于电网规划的有效性和正确性，为PSO算法的进一步改进拓展了思路。

IMPROVED PARTICLE SWARM OPTIMIZATION METHOD AND ITS APPLICATION IN POWER TRANSMISSION NETWORK PLANNING

Particle initialization distribution has great effect on its convergence performance, so a particle initialization method is put forth according to nonlinear simplex method. This method can improve the quality of initialization particles, and accelerate the convergence rapidity of PSO. Furthermore the oscillation phenomenon bring about by the use of penalty function is also analyzed, and then put forward an intercross iteration method for PSO calculation based on Dual Lagrange method. This method overcomes the shortcoming of penalty function. Numerical simulation results of power transmission network planning demonstrate they are feasible and efficient, and also shed new light on the further improving of PSO.