Python 实现粒子群算法

粒子群算法原理很简单，用matlab和python都很快实现编程。

程序：

参数部分，需要修改的可以修改。这个程序实现的是基本粒子群算法，对于提升粒子群算法的表现，

可以在上面进行更多的功能添加。

  
  

   
   import numpy as np
import random 
import matplotlib.pyplot as plt
#----------------------PSO参数设置---------------------------------
class PSO():
    def __init__(self,pN,dim,max_iter):
        self.w = 0.8  
        self.c1 = 2   
        self.c2 = 2   
        self.r1= 0.6
        self.r2=0.3
        self.pN = pN                #粒子数量
        self.dim = dim              #搜索维度
        self.max_iter = max_iter    #迭代次数
        self.X = np.zeros((self.pN,self.dim))       #所有粒子的位置和速度
        self.V = np.zeros((self.pN,self.dim))
        self.pbest = np.zeros((self.pN,self.dim))   #个体经历的最佳位置和全局最佳位置
        self.gbest = np.zeros((1,self.dim))
        self.p_fit = np.zeros(self.pN)              #每个个体的历史最佳适应值
        self.fit = 1e10             #全局最佳适应值
        
#---------------------目标函数Sphere函数-----------------------------
    def function(self,x):
        sum = 0
        length = len(x)
        x = x**2
        for i in range(length):
            sum += x[i]
        return sum
  
  
  
  


  
  
  
  

   
   #---------------------初始化种群----------------------------------
    def init_Population(self):
        for i in range(self.pN):
            for j in range(self.dim):
                self.X[i][j] = random.uniform(0,1)
                self.V[i][j] = random.uniform(0,1)
            self.pbest[i] = self.X[i]
            tmp = self.function(self.X[i])
            self.p_fit[i] = tmp
            if(tmp < self.fit):
                self.fit = tmp
                self.gbest = self.X[i]
    
#----------------------更新粒子位置----------------------------------
    def iterator(self):
        fitness = []
        for t in range(self.max_iter):
            for i in range(self.pN):         #更新gbest\pbest
               temp = self.function(self.X[i])
               if(temp<self.p_fit[i]):      #更新个体最优
                   self.p_fit[i] = temp
                   self.pbest[i] = self.X[i]
                   if(self.p_fit[i] < self.fit):  #更新全局最优
                       self.gbest = self.X[i]
                       self.fit = self.p_fit[i]
            for i in range(self.pN):
                self.V[i] = self.w*self.V[i] + self.c1*self.r1*(self.pbest[i] - self.X[i])\
                       + self.c2*self.r2*(self.gbest - self.X[i])
                self.X[i] = self.X[i] + self.V[i]
            fitness.append(self.fit)
            print(self.fit)                   #输出最优值
        return fitness
  
  
  
  


  
  
  
  

   
   
#----------------------程序执行-----------------------
my_pso = PSO(pN=30,dim=5,max_iter=100)
my_pso.init_Population()
fitness = my_pso.iterator()
  
  
  
  


  
  
  
  

   
   #-------------------画图--------------------
plt.figure(1)
plt.title("Figure1")
plt.xlabel("iterators", size=14)
plt.ylabel("fitness", size=14)
t = np.array([t for t in range(0,100)])
fitness = np.array(fitness)
plt.plot(t,fitness, color='b',linewidth=3)
plt.show()
  
  

计算结果：fitness = 8.64346866606e-07