论文部分内容阅读
为了有效地控制粒子群优化算法的全局搜索和局部搜索,基于递减惯性权值的基本思想,在现有的线性递减权值策略的基础上,提出了开口向下抛物线、开口向上抛物线和指数曲线3种非线性的权值递减策略,并采用Sphere、Rosenbrock、Griewank和Rastrigrin这4个标准测试函数测试这些策略对算法的影响.试验结果表明,对于多数连续优化问题,在初始权值和最终权值相同的情况下,凹函数递减策略优于线性策略,而线性策略优于凸函数策略,凹函数递减策略能够在不影响收敛精度的情况下较大幅度地提高粒子群算法的收敛速度.
In order to control the global search and local search of Particle Swarm Optimization algorithm effectively, based on the basic idea of decreasing inertia weight, based on the existing linear weight decreasing strategy, a parabolic parabola, an upward parabola and an exponential curve Three kinds of non-linear weight reduction strategies and tested the impact of these strategies on the algorithm using four standard test functions: Sphere, Rosenbrock, Griewank and Rastrigrin.The experimental results show that for most continuous optimization problems, the initial weights and final weights In the case of the same value, the concave function reduction strategy is better than the linear strategy, but the linear strategy is better than the convex function strategy. The concave function reduction strategy can greatly improve the convergence speed of PSO without affecting the convergence accuracy.