基于进化状态判定的模糊自适应二进制粒子群优化算法

随着迭代过程的推进,二进制粒子群算法容易陷入局部最优解,后期收敛性较差.针对此缺点,文中提出基于进化状态判定的模糊自适应二进制粒子群优化算法.采用隶属函数进行模糊分类的方法,判定种群进化状态.在迭代过程前期采用S形映射函数和较大的惯性权重值,提高收敛速度,保证算法的稳定性.后期采用V形映射函数和动态增减的惯性权重值,增强算法后期全局探索能力,避免其陷入局部最优.仿真实验表明,文中算法的收敛速度较快,精度较高,搜索能力较好,可以避免早熟现象.

Fuzzy Adaptive Binary Particle Swarm Optimization Algorithm Based on Evolutionary State Determination

Since the binary particle swarm algorithm is easy to fall into local optimal solution and its convergence performance during later period is poor, a fuzzy adaptive binary particle swarm optimization algorithm based on evolutionary state determination(EFBPSO) is proposed. Population evolution state is determined by fuzzy classification method based on membership function. S-shaped mapping function and large inertia weight value are adopted to improve convergence speed and ensure stability of the algorithm in the earlier stage of the iterative process. V-shaped mapping function and the smaller inertia weight are employed to enhance global exploration ability of the algorithm and avoid the algorithm falling into local optimization in the later stage of iterative process. Simulation experimental results show that EFBPSO possesses higher convergence speed and accuracy and obtains better searching ability to avoid prematurity.