基于Java实现PSO算法解决环形拓扑多模问题（源码）.rar

package RPSO; import java.io.IOException; import java.util.Random; public class pso { static int n = 5; double pi = Math.PI; /** * 计算两个个体间的距离 * * @param a * @param b * @return */ public double calculatedis(double a[], double b[]) { double dis = 0.0; for (int i = 0; i < n; i++) { dis += Math.pow((a[i] - b[i]), 2); } return Math.sqrt(dis); } /** * 计算个体之间的距离 * * @param population * [][]种群 * @param dis * [][]距离矩阵 * @param popsize * 种群规模 */ public void distance(double population[][], double dis[][], int popsize) { double temp1[] = new double[n]; double temp2[] = new double[n]; for (int i = 0; i < popsize; i++) { for (int j = 0; j < popsize; j++) { for (int k = 0; k < n; k++) { temp1[k] = population[i][k]; temp2[k] = population[j][k]; } dis[i][j] = calculatedis(temp1, temp2); } } } /** * 更新速度和位置 */ public void updatev(int k, double v[][], double population[][], double pbest[][], double nbest[][], double w, double c1, double c2, int popsize, double low, double high) { // 速度、位置更新公式 double random_pick1 = 0.0; double random_pick2 = 0.0; for (int i = 0; i < popsize; i++) { for (int j = 0; j < n; j++) { Random random = new Random(); random_pick1 = (double) Math.abs(random.nextInt(10000)) / 10000; random_pick2 = (double) Math.abs(random.nextInt(10000)) / 10000; v[i][j] = w * v[i][j] + c1 * random_pick1 * (pbest[i][j] - population[i][j]) + c2 * random_pick2 * (nbest[i][j] - population[i][j]); double Vmax = (high - (low)) * 0.1;// Vmax决定粒子在一个循环中最大的移动距离 double Vmin = -Vmax; if (v[i][j] > Vmax) { v[i][j] = Vmax; // System.out.println("速度超了"); } if (v[i][j] < Vmin) { v[i][j] = Vmin; // System.out.println("速度低了"); } population[i][j] = population[i][j] + v[i][j]; // r if (population[i][j] > high) { population[i][j] = high; // System.out.println("上界超了"); } if (population[i][j] < low) { population[i][j] = low; // System.out.println("下界超了"); } } } } /** * 找最优个体 * * @param fitness * [] 适应值 * @param popsize * 种群数目 * @param population * [][] 所有种群的种群个体 * @param BestIndividual * [] 最优个体 * @return fitness[population_position] 最优适应值 * @author zzx * @create Date 2016-1-4 */ public double FindBestIndividual(double fitness[], int popsize, double population[][], double BestIndividual[]) { // int lay_num = 200; int population_position = 0; double best = 99999; for (int i = 0; i < popsize; i++) { if (best > fitness[i]) { best = fitness[i]; population_position = i; } } for (int i = 0; i < n; i++) BestIndividual[i] = population[population_position][i]; return fitness[population_position]; } /** * 寻找相邻两个粒子的nbest * * @param a * @param b * @param nbest_fitness * @param pbest_fitness * @param nbest * @param pbest * @param n */ public void comparetwo(int a, int b, double nbest_fitness[], double pbest_fitness[], double nbest[][], double pbest[][]) { if (pbest_fitness[a] <= pbest_fitness[b]) { nbest_fitness[a] = pbest_fitness[a]; nbest_fitness[b] = pbest_fitness[a]; for (int i = 0; i < n; i++) { nbest[a][i] = pbest[a][i]; nbest[b][i] = pbest[a][i]; } } else { nbest_fitness[a] = pbest_fitness[b]; nbest_fitness[b] = pbest_fitness[b]; for (int i = 0; i < n; i++) { nbest[a][i] = pbest[b][i]; nbest[b][i] = pbest[b][i]; } } } /** * 找三个粒子的nbest * * @param left * 左粒子 * @param mid * 中间粒子 * @param right * 右粒子 * @param nbest_fitness * []nbest的适应度值 * @param pbest_fitness * []pbest的适应度值 * @param nbest * [][]nbest种群 * @param pbest * [][]pbest种群 * @param n染色体长度 */ public void Findnbest(int left, int mid, int right, double nbest_fitness[], double pbest_fitness[], double nbest[][], double pbest[][]) { comparetwo(left, mid, nbest_fitness, pbest_fitness, nbest, pbest); comparetwo(mid, right, nbest_fitness, pbest_fitness, nbest, pbest); } /** * 计算适应度值 * * @param I_popoulation * @return */ public double caculte(double I_population[]) { double x1 = 0; double t[] = new double[n]; double y[] = new double[n + 1]; double x[] = new double[n + 1]; for (int i = 0; i < n; i++) { x[i + 1] = I_population[i]; } for (int i = 1; i < n + 1; i++) { x1 += Math.pow(x[i] - 1, 2) + Math.pow(x[1] - x[i] * x[i], 2); } return x1; } public static void main(String[] args) throws IOException { // TODO Auto-generated method stub pso pso = new pso(); double w = 0.8; // 0.8 保持原来速度的系数，所以叫做惯性权重 double c1 = 2; // 2 粒子跟踪自己历史最优值的权重系数，它表示粒子自身的认识，所以叫“认知”。通常设置为2 double c2 = 2; // 2粒子跟踪群体最优值的权重系数，它表示粒子对整个群体知识的认识，所以叫做“社会知识”，经常叫做“社会”。通常设置为2。 // public static double r=1;//对位置更新的时候，在速度前面加的一个系数，这个系数我们叫做约束因子。通常设置为1 int popsize = 40;// 粒子 数目 int maxGen = 5000;// 迭代次数 double low = 0;// 下界 double high = 10;// 上界 /** * * 初始化部分 */ // 根据井的数据用于生成Vp，Vs，ρ的上下域 double population[][] = new double[popsize][n];// 粒子的位置，求解问题多少维，则此数组为多少维 double pbest[][] = new double[popsize][n];// 粒子的历史最好位置 double pbest_fitness[] = new double[popsize];// 历史最优解 double v[][] = new double[popsize][n];// 粒子的速度，维数同位置 double nbest[][] = new double[popsize][n];// nbest double nbest_fitness[] = new double[popsize]; double dis[][] = new double[popsize][popsize]; // pso.GeneratePopulation(population, popsize, v, pbest);// 根据上下域生成初始种群 // 种群初始化 for (int i = 0; i < popsize; i++) { for (int j = 0; j < n; j++) { Random random = new Random(); double val = ((double) Math.abs(random.nextInt(10000000)) / 10000000) * (high - low) + low; population[i][j] = val; } } /** * 存储初始种群的数据、适应值的数据 */ double GA_per_pop_sample[] = new double[n];// 存储单个种群个体所拥有的603个数据 double fitness[] = new double[popsize];// 种群大小的适应值数组 for (int i = 0; i < popsize; i++) { for (int j = 0; j < n; j++) { GA_per_pop_sample[j] = population[i][j]; } fitness[i] = pso.caculte(GA_per_pop_sample); pbest_fitness[i] = fitness[i]; } double gbest[] = new double[n];// 所有粒子找到的最好位置 double gbest_fitness = 99999; gbest_fitness = pso.FindBestIndividual(fitness, popsize, population, gbest); System.out.println("初始 gbest_fitness: " + gbest_fitness); // 初始pbest为自身 for (int i = 0; i < popsize; i++) { for (int j = 0; j < n; j++) { pbest[i][j] = population[i][j]; } } for (int i = 0; i < 40 / 3; i++) { // 三个体不重叠 pso.Findnbest(i * 3, i * 3 + 1, i * 3 + 2, nbest_fitness, pbest_fitness, nbest, pbest); } nbest_fitness[popsize - 1] = fitness[popsize - 1]; nbest[popsize - 1] = pbest[popsize - 1]; // 开始迭代 for (int k = 0; k < maxGen; k++) { pso.updatev(k, v, population, pbest, nbest, w, c1, c2, popsize, low, high);// 更新速度和位置 for (int i = 0; i < popsize; i++) { for (int j = 0; j < n; j++) { GA_per_pop_sample[j] = population[i][j]; } fitness[i] = pso.caculte(GA_per_pop_sample);// 计算适应度值 if (fitness[i] < pbest_fitnes