【摘要】 群优化是一种随机的群体搜索策略。针对粒子群算法易陷入局部最优和收敛速度慢等不足,提出了根据粒子的能量自适应调整参数的改进算法。该算法基于动力学和热力学的理论,计算每个粒子的能量值,并将优化过程中的群体视为热力学的某一状态,通过退火温度和粒子的能量动态调整算法中的惯性参数,达到对"惰性"粒子的原速度方向给予较大的牵引力的目的。优化过程中,随着系统温度的降低,惯性参数逐渐减小,有利于问题的收敛。算法中采用了带极值扰动策略,加速粒子跳出局部最优的能力。数值实验结果表明,该算法具有收敛精度高和收敛速度快的特点,可快速有效的求解约束和非约束优化问题。更多还原

【Abstract】 Swarm optimization algorithm is a stochastic,population-based searching strategy.The original particle swarm optimizer(PSO) has some deficiencies,such as falling in the local optimal region,slow convergence velocity,and so on.An improvement on the PSO algorithm was proposed,called the Energy Adaptive Particle Swam Optimizer(EAPSO),by employing the particle’s energy to adaptively adjust the parameters.The energy of particle was calculated based on the dynamics theory and thermodynamics theory.In the proceeding of optimization,the swarm was viewed as a state in the thermodynamics.And the inertia weight varies according to the annealing temperature and the particle’s energy in order to give the stronger traction on the original speed direction of inertia particle.With the reducing of the temperature,the decreasing inertia weight conduces to the convergence issue.The disturbed extremum scheme accelerates the particles to overstep the local extremum.Application of the EAPSO on several optimization problems shows the EAPSO algorithm holds the fast convergence velocity and good precise.EAPSO algorithm can effectively and quickly solve the constrained and unconstrained optimization problems.更多还原