🍁🥬🕒摘要🕒🥬🍁
针对监测区域内含有障碍物的无线传感器网络(Wireless Sensor Networks,WSNs)异构节点部署优化问题,在花朵授粉算法(Flower Pollination Algorithm,FPA)的基础之上,提出了一种改进的 花朵授粉算法(Improved Flower Pollination Algorithm,IFPA)用于改善原有算法收敛速度慢、精度不够高的不足。设计非线性收敛因子以约束原有的缩放因子,采用 Tent 映射以维持迭代后期种群的多样性,而贪心交叉策略则是以较优的个体辅助较差个体搜索。基准函数实验验证了 IFPA 具有较好的收敛性能,而 WSN 部署的仿真实验表明 IFPA 可得到较高的覆盖率,可节约网络部署成本。
✨🔎⚡部分运行结果⚡🔎✨
运行中如果提示安装工具箱,安装即可,例如Symbolic Math Toolbox,如果用2014b,则不需要安装。
无障碍物:
有障碍物:
💂♨️👨🎓Matlab代码👨🎓♨️💂
%三角形和菱形障碍物 geshu_x = [20,25,30,35,40,45,50]; init_y = [54.21,61.09,67.44,73.07,77.24,79.15,82.61]/100; ga_y = [80.71,87.40,94.69,96.74,98.23,99.17,100]/100; pso_y = [83.37,92.37,95.13,96.79,98.56,99.31,100]/100; dea_y = [85.08,93.03,95.79,97.62,99.28,99.50,100]/100; fa_y = [85.69,93.57,96.96,98.61,99.28,100,100]/100; ifa_y = [86.02,94.25,97.12,98.73,99.34,100,100]/100; figure(1); % plot(geshu_x,init_y,'color','k'); % hold on; plot(geshu_x,ga_y,'color','r'); hold on; plot(geshu_x,pso_y,'color','g'); hold on; plot(geshu_x,dea_y,'color','b'); hold on; plot(geshu_x,fa_y,'color','c'); hold on; plot(geshu_x,ifa_y,'color','m'); hold on; legend('ga','pso','dea','fa','ifa'); hold on;
%%主程序 clc; clear ; close all; %删除相应的文件 global N; global M; global L; global W; global Grid_cen_x; global Grid_cen_y; global Grid_cen_x_and_y; global ger; p=0.8;%判断是否是全局优化还是局部优化 L = 50;%长 W = 50;%宽 %假设1平方米一个网格 M = 2500;%网格总数 r_max = 7;%感知半径为5 r_mid = 6; r_min = 5; energy_max = 100;%最大的能量 energy_mid = 90; energy_min = 80; per_sersons_radius_type = [r_max,r_mid,r_min]; %假设大、中为5,剩下为小 N = 25;%30个传感器节点 sizepop = 50;%种群规模 dimension = 2;% 空间维数 前行放x、y,第三行放半径 ger = 10;% 最大迭代次数 pos_limit = [0, 50]; % 设置位置参数限制 %个数限制 r_max_num = 1;%序号为1-5 r_mid_num = 2;%序号为6-10 r_min_num = N - r_max_num - r_mid_num; %序号为11-N struct_pop_per = struct('per',[],'radius',[],'energy_init',[],'energy_end',[],'sersons_num',[]);%结构体类型 struct_pops_temp = repmat(struct_pop_per,[1 sizepop]);%临时的一个种群 energy_init_arr = zeros(1,N); energy_end_arr = zeros(1,N); radius_arr = zeros(1,N); %求出梯形的四个点 syms x y;%先定义一个变量 %左上角 k1 = 1; b1 = 35; x1_up = solve(k1*x+b1==50,x);%左上角的斜线的上个交点 y1_down = solve(k1*0+b1==y,y); %左下角 k2 = -1; b2 = 15; y2_up = solve(k2*0+b2==y,y); x2_down = solve(k2*x+b2==0,x); %右上角 k3 = -1; b3 = 85; x3_up = solve(k3*x+b3==50,x); y3_down = solve(k3*50+b3==y,y); %右下角 k4 = 1; b4 = -35; y4_up = solve(k4*50+b4==y,y); x4_down = solve(k4*x+b4==0,x); %以下数据验证完毕,完全正确 point = zeros(8,2);%存储这些点 从左 从上往下 point(1,:) = [x1_up,50]; point(2,:) = [0,y1_down]; point(3,:) = [0,y2_up]; point(4,:) = [x2_down,0]; point(5,:) = [x3_up,50]; point(6,:) = [50,y3_down]; point(7,:) = [50,y4_up]; point(8,:) = [x4_down,0]; %菱形的计算 point_diamond = zeros(2,4);%菱形的四个点,方位是顺时针 第一列为上 二列为右 %求出新菱形形的四个点 syms x y;%先定义一个变量 %左上角 k5 = 1; b5 = 10; %别搞什么计算了 直接可以看出来 point_diamond(1,1) = 25; point_diamond(2,1) = 35; %右上角 k6 = -1; b6 = 60; point_diamond(1,2) = 35; point_diamond(2,2) = 25; %右下角 k7 = 1; b7 = -10; point_diamond(1,3) = 25; point_diamond(2,3) = 15; %左下角 k8 = 1; b8 = 40; point_diamond(1,4) = 15; point_diamond(2,4) = 25; load struct_pop_public.mat;%加载该种群 struct_pops = struct_pop_public;%得到种群数据 load struct_first_init_public.mat%加载最开始的一个个体数据 struct_first_init = struct_first_init_public;%得到初始化个体数据 %%初始的部署后画图 拿第一个粒子拿去初始画图
📜📢🌈参考文献🌈📢📜
[1]王振东,谢华茂,胡中栋,李大海,王俊岭.改进花朵授粉算法的无线传感器网络部署优化[J].系统仿真学报,2021,33(03):645-656.DOI:10.16182/j.issn1004731x.joss.19-0580.