目录

一、算法介绍

1.1 遗传算法

1.2 为什么要使用遗传算法进行改进

二、算法原理

三、算法实现

3.1 算子选择

3.2 代码实现

一、算法介绍

1.1 遗传算法

遗传算法是受启发于自然界中生物对于自然环境 “适者生存”的强大自适应能力，通过对生物演化过程模拟和抽象，构建了以自然界生物演变进化为逻辑基础的遗传算法。遗传算法包括了自然界生物在演变过程中的主要步骤，即选择、（基因）变异和（基因）交叉，对应着遗传算法中的三个运算算子。在具体的优化问题下，遗传算法会产生多个问题的可行解作为种群，然后让种群进行模拟意义上生物进化中的选择、变异、交叉等操作。在种群繁衍（迭代）一定次数之后，通过计算种群的适应度，寻找最终种群中的最优个体，该个体即代表优化问题的近似最优解。上述此即为遗传算法主要思想。其流程图如下：

1.2 为什么要使用遗传算法进行改进

BP算法原理不多赘述，可见我之前博文BP原理介绍，在BP训练过程中，很容易出现陷入局部最小值的情况，所以引入遗传算法进行优化。遗传作为一种模拟生物进化的全局寻优算法，有着优秀的全局寻优能力，能够以一个种群为基础不断的迭代进化，最后获得问题的最优解或近似最优解。BP算法和遗传算法都是人们广泛使用的算法，而且两算法具有明显的优势互补，故而很多研究者都在探索两个算法的融合方法，以期能提高算法性能、提升算法精度。

二、算法原理

基于遗传算法改进的BP神经网络算法（GA-BP算法）的主要思想即为：通过遗传算法的全局寻优能力获得最优的BP网络的初始权值和阈值，将寻优算法获得的最优初始权值和阈值作为BP神经网络的初始权值和阈值，然后进行训练以避免陷入局部最小值。遗传算法改进后的BP神经网络权值不是随机产生的，而是遗传算法寻优模块获得的。BP算法中的初始权值和阈值作为遗传算法个体的基因值，个体长度即为BP神经网络中权值和阈值的个数，每个基因即代表一个权值或阈值，基因上的数值就是BP神经网络中连接权值或阈值的真实值，如此便组成了遗传算法中的一个染色体。一定数量的染色体作为遗传算法训练的初始种群，再经过遗传算法的选择运算、交叉运算、变异运算等迭代过程后获得一个最优个体，然后以最优个体作为BP网络的初始参数进行训练，此即为GA-BP算法的原理。流程图如下：

三、算法实现

3.1 算子选择

对于(e)所述的组织方法，是当影响因子数据和目标数据没有很强的相关性的情况下，用前一时序区间的数据作为该时序数据的影响因子来进行训练。

3.2 代码实现

实例为基于一段时序监测数据的滑坡位移预测，监测影响因子数据有：温度、降雨、风力、灌溉等，监测的目标数据是坡体的裂缝宽度数据。实验表明影响因子数据和目标数据不具有强相关性，所以选择用目标数据本身作为影响因子数据。

将整个算法分成如下模块：

chrom_code # 基因编码模块chrom_mutate # 变异算子模块chrom_cross # 交叉算子模块chrom_select # 选择算子模块chrom_fitness # 染色体适应度计算模块data_prepare # 数据准备模块BP_network # BPNN模块chrom_test # 染色体检测模块new_GA-BP # 改进算法主程序

chrom_test.py 检测生成的染色体基因有没有超限。

# 染色体检查# 检查染色体中有没有超出基因范围的基因def test(code_list,bound):""":param code_list: code_list: 染色体个体:param bound: 各基因的取值范围:return: bool"""for i in range(len(code_list)):if code_list[i] < bound[i][0] or code_list[i] > bound[i][1]:return Falseelse:return True

chrom_code.py 基因编码。

# 基因编码模块import randomimport numpy as npimport chrom_testdef code(chrom_len,bound):""":param chrom_len: 染色体的长度，为一个数，采用实数编码即为基因的个数:param bound: 取值范围，为一个二维数组，每个基因允许的取值范围:return: 对应长度的编码"""code_list = []count = 0while True:pick = random.uniform(0,1)if pick == 0:continueelse:pick = round(pick,3)temp = bound[count][0] + (bound[count][1] - bound[count][0])*picktemp = round(temp,3)code_list.append(temp)count = count + 1if count == chrom_len:if chrom_test.test(code_list,bound):breakelse:count = 0return code_list

BP_network.py完成网络结构的构建。

# BP模块 借助PyTorch实现import torch# 引入了遗传算法参数的BP模型class BP_net(torch.nn.Module):def __init__(self, n_feature, n_hidden, n_output, GA_parameter):super(BP_net, self).__init__()# 构造隐含层和输出层self.hidden = torch.nn.Linear(n_feature, n_hidden)self.output = torch.nn.Linear(n_hidden, n_output)# 给定网络训练的初始权值和偏执等self.hidden.weight = torch.nn.Parameter(GA_parameter[0])self.hidden.bias = torch.nn.Parameter(GA_parameter[1])self.output.weight = torch.nn.Parameter(GA_parameter[2])self.output.bias = torch.nn.Parameter(GA_parameter[3])def forward(self, x):# 前向计算hid = torch.tanh(self.hidden(x))out = torch.tanh(self.output(hid))return out# 传统的BP模型class ini_BP_net(torch.nn.Module):def __init__(self, n_feature, n_hidden, n_output):super(ini_BP_net, self).__init__()# 构造隐含层和输出层self.hidden = torch.nn.Linear(n_feature, n_hidden)self.output = torch.nn.Linear(n_hidden, n_output)def forward(self, x):# 前向计算hid = torch.tanh(self.hidden(x))out = torch.tanh(self.output(hid))return outdef train(model, epochs, learning_rate, x_train, y_train):""":param model: 模型:param epochs: 最大迭代次数:param learning_rate:学习率:param x_train:训练数据（输入）:param y_train:训练数据（输出）:return: 最终的loss值（MSE）"""# path = "log.txt"# f = open(path, 'w',encoding='UTF-8')# f.write("train log\n------Train Action------\n"# "Time：{}\n".format(time.ctime()))loss_fc = torch.nn.MSELoss(reduction="sum")optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate)loss_list = []for i in range(epochs):model.train()# 前向计算data = model(x_train)# 计算误差loss = loss_fc(data, y_train)loss_list.append(loss)# 更新梯度optimizer.zero_grad()# 方向传播loss.backward()# 更新参数optimizer.step()# print("This is {} th iteration,MSE is {}。".format(i+1,loss))loss_ls = [loss_list[i].detach().numpy() for i in range(len(loss_list))]return loss_ls

chrom_fitness.py 适应度计算

# 适应度计算模块# 功能；传入一个编码，返回一个适应度值from torchvision.transforms import transformsimport torchimport BP_networkimport numpy as np# 最小二乘思想获得两组数据的误差def zxec_PC(X, Y):X = np.array(X, dtype=np.float).flatten()Y = np.array(Y, dtype=np.float).flatten()if len(X) != len(Y):print("Wrong!")n = len(X)Wc = 0for i in range(n):Wc = Wc + (X[i] - Y[i]) * (X[i] - Y[i])return Wcdef calculate_fitness(code,n_feature,n_hidden,n_output,epochs,learning_rate,x_train,y_train):""":param code: 染色体编码:param n_feature: 输入层个数:param n_hidden: 隐含层个数:param n_output: 输出层个数:param epochs: 最多迭代次数:param learning_rate: 学习率:param x_train: 训练（输入）数据:param y_train: 训练（输出）数据:return: fitness 适应度值"""Parameter = code[:]# 参数提取hidden_weight = Parameter[0:n_feature * n_hidden]hidden_bias = Parameter[n_feature * n_hidden:n_feature * n_hidden + n_hidden]output_weight = Parameter[n_feature * n_hidden + n_hidden:n_feature * n_hidden + n_hidden + n_hidden * n_output]output_bias = Parameter[n_feature * n_hidden + n_hidden + n_hidden * n_output:n_feature * n_hidden + n_hidden + n_hidden * n_output + n_output]# 类型转换tensor_tran = transforms.ToTensor()hidden_weight = tensor_tran(np.array(hidden_weight).reshape((n_hidden, n_feature))).to(torch.float32)hidden_bias = tensor_tran(np.array(hidden_bias).reshape((1, n_hidden))).to(torch.float32)output_weight = tensor_tran(np.array(output_weight).reshape((n_output,n_hidden))).to(torch.float32)output_bias = tensor_tran(np.array(output_bias).reshape((1, n_output))).to(torch.float32)# 形装转换hidden_weight = hidden_weight.reshape((n_hidden,n_feature))hidden_bias = hidden_bias.reshape(n_hidden)output_weight = output_weight.reshape((n_output,n_hidden))output_bias = output_bias.reshape(n_output)# 带入模型计算GA = [hidden_weight, hidden_bias, output_weight, output_bias]BP_model = BP_network.BP_net(n_feature,n_hidden,n_output,GA)loss = BP_network.train(BP_model,epochs,learning_rate,x_train,y_train)# 计算适应度prediction = BP_model(x_train)fitness = 10 - zxec_PC(prediction.detach().numpy(),y_train.detach().numpy())return round(fitness,4)

chrom_mutate.py 选择算子

# 变异算子import randomdef mutate(chrom_sum, size, p_mutate, chrom_len, bound, maxgen, nowgen):""":param chrom_sum: 染色体群，即种群，里面为一定数量的染色体 类型为一个二维列表:param size: 种群规模，即染色体群里面有多少个染色体 为一个数:param p_mutate: 交叉概率 为一个浮点数:param chrom_len: 种群长度，即一条染色体的长度，即基因的个数 为一个数:param bound: 各基因的取值范围:param maxgen: 最大迭代次数:param nowgen: 当前迭代次数:return: 变异算子后的种群"""count = 0# print("\n---这是第{}次遗传迭代...".format(nowgen))while True:# 随机选择变异染色体# print("{}-{}".format(nowgen,count+1))seek = random.uniform(0,1)while seek == 1:seek = random.uniform(0,1)index = int(seek * size)# print("可能变异的染色体号数为：",index)# 判断是否变异flag = random.uniform(0,1)if p_mutate >= flag:# 选择变异位置# print("发生变异中...")seek1 = random.uniform(0,1)while seek1 == 1:seek1 = random.uniform(0,1)pos = int(seek1 * chrom_len)# print("变异的基因号数为：",pos)# 开始变异seek3 = random.uniform(0,1)fg = pow(seek3*(1-nowgen/maxgen),2) # 约到迭代后期，其至越接近0，变异波动就越小# print("变异前基因为：",chrom_sum[index][pos])if seek3 > 0.5:chrom_sum[index][pos] = round(chrom_sum[index][pos] +(bound[pos][1] - chrom_sum[index][pos])*fg,3)else:chrom_sum[index][pos] = round(chrom_sum[index][pos] -(chrom_sum[index][pos] - bound[pos][0])*fg,3)# print("变异后基因为：", chrom_sum[index][pos])count = count + 1else:# print("未发生变异。")count = count + 1if count == size:breakreturn chrom_sum

chrom_cross.py 交叉算子

# 交叉算子import randomimport chrom_testdef cross(chrom_sum, size, p_cross, chrom_len, bound):""":param chrom_sum:种群集合，为二维列表:param size:种群总数，即染色体的个数:param p_cross:交叉概率:param chrom_len:染色提长度，每个染色体含基因数:param bound:每个基因的范围:return: 交叉后的种群集合"""count = 0while True:# 第一步 先选择要交叉的染色体seek1 = random.uniform(0,1)seek2 = random.uniform(0,1)while seek1 == 0 or seek2 == 0 or seek1 == 1 or seek2 == 1:seek1 = random.uniform(0, 1)seek2 = random.uniform(0, 1)# index_1(2)为选中交叉的个体在种群中的索引index_1 = int(seek1 * size)index_2 = int(seek2 * size)if index_1 == index_2:if index_2 == size - 1:index_2 = index_2 - 1else:index_2 = index_2 + 1# print("可能交叉的两个染色体为：",index_1,index_2)# 第二步 判断是否进行交叉flag = random.uniform(0,1)while flag == 0:flag = random.uniform(0,1)if p_cross >= flag:# 第三步 开始交叉# print("开始交叉...")p_pos = random.uniform(0, 1)while p_pos == 0 or p_pos == 1:p_pos = random.uniform(0, 1)pos = int(p_pos * chrom_len)# print("交叉的极影位置为：",pos)var1 = chrom_sum[index_1][pos]var2 = chrom_sum[index_2][pos]pick = random.uniform(0,1)# print("交叉前染色体为：")# print(chrom_sum[index_1])# print(chrom_sum[index_2])chrom_sum[index_1][pos] = round((1-pick) * var1 + pick * var2,3)chrom_sum[index_2][pos] = round(pick * var1 + (1-pick) * var2,3)# print("交叉后染色体为：")# print(chrom_sum[index_1])# print(chrom_sum[index_2])if chrom_test.test(chrom_sum[index_1],bound) and chrom_test.test(chrom_sum[index_2],bound):count = count + 1else:continueelse:# print("没有发生交叉。")count = count + 1# print("本次循环结束\n")if count == size:breakreturn chrom_sum

chrom_select.py 选择算子

# 选择算子import numpy as npimport randomdef select(chrom_sum,fitness_ls):""":param chrom_sum:种群:param fitness_ls: 各染色体的适应度值:return: 更新后的种群"""# print("种群适应度分别为：",fitness_ls)fitness_ls = np.array(fitness_ls,dtype=np.float64)sum_fitness_ls = np.sum(fitness_ls,dtype=np.float64)P_inh = []M = len(fitness_ls)for i in range(M):P_inh.append(fitness_ls[i]/sum_fitness_ls)# 将概率累加for i in range(len(P_inh)-1):P_temp = P_inh[i] + P_inh[i+1]P_inh[i+1] = round(P_temp, 2)P_inh[-1] = 1# 轮盘赌算法选择染色体account = []for i in range(M):rand = random.random()for j in range(len(P_inh)):if rand <= P_inh[j]:account.append(j)breakelse:continue# 根据索引号跟新种群# print("轮盘赌的结果为：",account)new_chrom_sum = []for i in account:new_chrom_sum.append(chrom_sum[i])return new_chrom_sum

data_prepare.py 数据准备

# 数据准备import numpy as npimport pandas as pddef Data_loader():# 文件路径ENU_measure_path = "18-10-25至19-3-25三方向位移数据.xlsx"t_path = "天气数据.xls"M_path = "data.csv"# 三方向数据df_1 = pd.read_excel(ENU_measure_path)ENU_df = pd.DataFrame(df_1)ENU_E = ENU_df["E/m"]ENU_E = np.array(ENU_E)ENU_N = ENU_df["N/m"]ENU_N = np.array(ENU_N)ENU_U = ENU_df["U/m"]ENU_U = np.array(ENU_U)ENU_R = ENU_df['R/m']ENU_R = np.array(ENU_R)df_2 = pd.read_excel(t_path)t_df = pd.DataFrame(df_2)# 最大温度数据max_tem = t_df["bWendu"]max_tem_ls = []for i in range(len(max_tem)):temp = str(max_tem[i])temp = temp.replace("℃","")max_tem_ls.append(eval(temp))max_tem = np.array(max_tem_ls)# 最低温度数据min_tem = t_df["yWendu"]min_tem_ls = []for i in range(len(min_tem)):temp = str(min_tem[i])temp = temp.replace("℃","")min_tem_ls.append(eval(temp))min_tem =np.array(min_tem_ls)# 天气数据tianqi = t_df["Tian_Qi"]tianqi = np.array(tianqi)# 风力数据Feng = t_df["Feng"]Feng = np.array(Feng)# 降雨数据rain = t_df["rainfall"]rain = np.array(rain)# 灌溉数据guangai = t_df["guangai"]guangai = np.array(guangai)# 获取时间数据namels = t_df["ymd"]name_ls = []for i in range(len(namels)):temp = str(namels[i])temp = temp.replace(" 00:00:00","")name_ls.append(str(temp))# 读取另一文件数据，该数据为位移计和GNSS监测数据df_3 = pd.read_csv(M_path)M_df = pd.DataFrame(df_3)M_data = M_df["Measurerel"]R_data = M_df["R"]M_data = np.array(M_data)R_data = np.array(R_data)return [ENU_R, M_data, R_data, ENU_U, ENU_E, ENU_N,max_tem,min_tem,name_ls]

主程序！！！！

# 改进算法主程序import sysimport chrom_code # 基因编码模块import chrom_mutate # 变异算子模块import chrom_cross # 交叉算子模块import chrom_select # 选择算子模块import chrom_fitness # 染色体适应度计算模块import data_prepare # 数据准备模块import BP_network # BPNN模块import torchimport torch.nn.functional as Ffrom torchvision.transforms import transformsimport numpy as npimport matplotlib.pyplot as pltimport timeplt.rcParams['font.sans-serif'] = ['SimHei']plt.rcParams['axes.unicode_minus'] = False# -----参数设置-----epochs = 300 # 神经网络最大迭代次数learning_rate = 0.01 # 学习率n_feature = 6 # 输入层个数n_hidden = 9 # 隐含层个数n_output = 1 # 输出层个数chrom_len = n_feature * n_hidden + n_hidden + n_hidden * n_output + n_output # 染色体长度size = 15 # 种群规模bound = np.ones((chrom_len, 2))sz = np.array([[-1, 0], [0, 1]])bound = np.dot(bound, sz) # 各基因取值范围p_cross = 0.4 # 交叉概率p_mutate = 0.01 # 变异概率maxgen = 30 # 遗传最大迭代次数# 数据准备# ========================================= #data_set = data_prepare.Data_loader()displace = data_set[1]name_ls = data_set[-1]in_train_data = []in_test_data = []# 数目分配train_num = 120test_num = len(displace) - train_num - n_featurefor i in range(len(displace)):temp = []if i <= train_num-1: # 用于控制训练数据和预测数据的分配temp = [round(displace[i + j], 5) for j in range(n_feature)]in_train_data.append(temp)else:temp = [round(displace[i + j], 5) for j in range(n_feature)]in_test_data.append(temp)if i == len(displace)-n_feature-1:break# 格式转化in_train_data = np.array(in_train_data)in_test_data = np.array(in_test_data)# 数据分割，用于建模和预测out_train_data = displace[n_feature:train_num+n_feature]out_test_data = displace[train_num+n_feature:len(displace)]# 测试输出# print(in_train_data)# print(out_train_data)# print(in_test_data)# print(out_test_data)# print(train_num)# print(test_num)# 数据格式转换及数据归一化tensor_tran = transforms.ToTensor()# 训练过程中的输入层数据in_train_data = tensor_tran(in_train_data).to(torch.float)in_train_data = F.normalize(in_train_data)in_train_data = in_train_data.reshape(train_num, n_feature)# 预测过程中的输入层数据in_test_data = tensor_tran(in_test_data).to(torch.float)in_test_data = F.normalize(in_test_data)in_test_data = in_test_data.reshape(test_num, n_feature)# 训练过程中的输出层数据out_train_data = out_train_data.reshape(len(out_train_data), 1)out_train_data = tensor_tran(out_train_data).to(torch.float)un_norm1 = out_train_data[0][0]out_train_data = F.normalize(out_train_data)norm1 = out_train_data[0][0]out_train_data = out_train_data.reshape(train_num, n_output)fanshu_train = round(float(un_norm1 / norm1), 4) # 建模时，训练数据中输出数据的范数# 预测中用于检验的输出层数据out_test_data = out_test_data.reshape(len(out_test_data), 1)out_test_data = tensor_tran(out_test_data).to(torch.float)un_norm = out_test_data[0][0] # 归一化前out_test_data = F.normalize(out_test_data)norm = out_test_data[0][0] # 归一化后out_test_data = out_test_data.reshape(test_num, n_output)fanshu = round(float(un_norm / norm), 4) # 预测时，测试数据中输出数据的范数# 建模训练数据x_train = in_train_datay_train = out_train_datax_test = in_test_datay_label = out_test_data# ========================================== #chrom_sum = [] # 种群，染色体集合for i in range(size):chrom_sum.append(chrom_code.code(chrom_len, bound))account = 0 # 遗传迭代次数计数器best_fitness_ls = [] # 每代最优适应度ave_fitness_ls = [] # 每代平均适应度best_code = [] # 迭代完成适应度最高的编码值# 适应度计算fitness_ls = []for i in range(size):fitness = chrom_fitness.calculate_fitness(chrom_sum[i], n_feature, n_hidden, n_output,epochs, learning_rate, x_train, y_train)fitness_ls.append(fitness)# 收集每次迭代的最优适应值和平均适应值fitness_array = np.array(fitness_ls).flatten()fitness_array_sort = fitness_array.copy()fitness_array_sort.sort()best_fitness = fitness_array_sort[-1]best_fitness_ls.append(best_fitness)ave_fitness_ls.append(fitness_array.sum() / size)while True:# 选择算子# print("\n这是第{}次遗传迭代。".format(account+1))# print("平均适应度为：",fitness_array.sum()/size)chrom_sum = chrom_select.select(chrom_sum, fitness_ls)# 交叉算子chrom_sum = chrom_cross.cross(chrom_sum, size, p_cross, chrom_len, bound)# 变异算子chrom_sum = chrom_mutate.mutate(chrom_sum, size, p_mutate, chrom_len, bound, maxgen, account + 1)# 适应度计算fitness_ls = []for i in range(size):fitness = chrom_fitness.calculate_fitness(chrom_sum[i], n_feature, n_hidden, n_output,epochs, learning_rate, x_train, y_train)fitness_ls.append(fitness)# 收集每次迭代的最优适应值和平均适应值fitness_array = np.array(fitness_ls).flatten()fitness_array_sort = fitness_array.copy()fitness_array_sort.sort()best_fitness = fitness_array_sort[-1] # 获取最优适应度值best_fitness_ls.append(best_fitness)ave_fitness_ls.append(fitness_array.sum() / size)# 计数器加一account = account + 1if account == maxgen:index = fitness_ls.index(max(fitness_ls)) # 返回最大值的索引best_code = chrom_sum[index] # 通过索引获得对于染色体break# 参数提取hidden_weight = best_code[0:n_feature * n_hidden]hidden_bias = best_code[n_feature * n_hidden:n_feature * n_hidden + n_hidden]output_weight = best_code[n_feature * n_hidden + n_hidden:n_feature * n_hidden + n_hidden + n_hidden * n_output]output_bias = best_code[n_feature * n_hidden + n_hidden + n_hidden * n_output:n_feature * n_hidden + n_hidden + n_hidden * n_output + n_output]# 类型转换tensor_tran = transforms.ToTensor()hidden_weight = tensor_tran(np.array(hidden_weight).reshape((n_hidden, n_feature))).to(torch.float32)hidden_bias = tensor_tran(np.array(hidden_bias).reshape((1, n_hidden))).to(torch.float32)output_weight = tensor_tran(np.array(output_weight).reshape((n_output, n_hidden))).to(torch.float32)output_bias = tensor_tran(np.array(output_bias).reshape((1, n_output))).to(torch.float32)# 形装转换hidden_weight = hidden_weight.reshape((n_hidden, n_feature))hidden_bias = hidden_bias.reshape(n_hidden)output_weight = output_weight.reshape((n_output, n_hidden))output_bias = output_bias.reshape(n_output)GA = [hidden_weight, hidden_bias, output_weight, output_bias]# 带入模型计算BP_model = BP_network.BP_net(n_feature, n_hidden, n_output, GA)ini_BP_model = BP_network.ini_BP_net(n_feature, n_hidden, n_output)# 网络训练loss = BP_network.train(BP_model, epochs, learning_rate, x_train, y_train)ini_loss = BP_network.train(ini_BP_model, epochs, learning_rate, x_train, y_train)# 建模效果model_x = BP_model(x_train)ini_model_x = ini_BP_model(x_train)# 网络预测prediction = BP_model(x_test)ini_prediction = ini_BP_model(x_test)# 建模数据反归一化(都换算到厘米级)y_train = y_train.detach().numpy() * fanshu_trainmodel_x = model_x.detach().numpy() * fanshu_trainini_model_x = ini_model_x.detach().numpy() * fanshu_train# 建模绘图train_name_ls = name_ls[6:126]xlabel = [i for i in range(0, 120, 14)]plt.plot(y_train, markersize=4, marker='.', label="真值", c='r')plt.plot(model_x, markersize=4, marker='.', label="GA-BP预测值", c='b')plt.title("GA-BP算法建模情况")plt.ylabel("累计裂缝宽度（mm）")plt.xticks(xlabel, [train_name_ls[i] for i in xlabel], rotation=25)plt.grid(linestyle='-.') # 设置虚线plt.legend()f2 = plt.figure()plt.plot(y_train, markersize=4, marker='.', label="真值", c='r')plt.plot(ini_model_x, markersize=4, marker='.', label="BP预测值", c='g')plt.title("BP算法建模情况")plt.ylabel("累计裂缝宽度（mm）")plt.xticks(xlabel, [train_name_ls[i] for i in xlabel], rotation=25)plt.grid(linestyle='-.')plt.legend()# 预测数据格式转换（厘米级）GABP_prediction = prediction.detach().numpy()BP_prediction = ini_prediction.detach().numpy()y_label = y_label.detach().numpy()# 预测数据反归一化（厘米级）GABP_prediction = GABP_prediction * fanshuBP_prediction = BP_prediction * fanshuy_label = y_label * fanshu# 计算预测结果的SSE误差def get_MSE(argu1, argu2):if len(argu1) != len(argu2):return 0error = 0for i in range(len(argu1)):error = error + pow((argu1[i] - argu2[i]), 2)error = float(error[0])return round(error, 5)error_BP = get_MSE(y_label, BP_prediction)error_GA_BP = get_MSE(y_label, GABP_prediction)print("BP算法预测MSE误差为：", error_BP)print("GA-BP算法预测MSE误差为：", error_GA_BP)# 将巡行情况和运行结果写入日志f = open("log.txt",'a',encoding='UTF-8')# 追加写打开文件f.write("运行时间：" + str(time.ctime()) + '\n')f.write("训练数据长度为：" + str(train_num) + '\n'+ "测试数据长度为：" + str(test_num) + '\n')f.write("网络结构层数为：{}、{}、{}\n".format(n_feature,n_hidden,n_output))f.write("遗传迭代所获得的最优权值为：" + str(best_code) + "\n")f.write("======预测结果如下======\n真值数据为：" + str(y_label.flatten()) + '\n')f.write("BP预测结果为：" + str(BP_prediction.flatten()) + "\n"+ "GA-BP预测结果为：" + str(GABP_prediction.flatten()) + '\n')f.write("-->>BP预测MSE误差为：" + str(error_BP) + '平方厘米\n'+ "-->>GA-BP预测MSE误差为：" + str(error_GA_BP) + '平方厘米\n\n')f.close()# 预测绘图test_name_ls = name_ls[126:152]xlabel2 = [i for i in range(0, 26, 4)]f3 = plt.figure()plt.plot(y_label, markersize=4, marker='.', label="真值", c='r')plt.plot(GABP_prediction, markersize=4, marker='*', label="GA-BP预测值", c='b')plt.plot(BP_prediction, markersize=4, marker='^', label="BP预测值", c='g')plt.title("算法预测情况对比")plt.ylabel("累计裂缝宽度（mm）")plt.xticks(xlabel2, [test_name_ls[i] for i in xlabel2], rotation=20)plt.legend()plt.grid(linestyle='-.')f4 = plt.figure()plt.plot(y_label, markersize=4, marker='.', label="真值", c='r')plt.plot(BP_prediction, markersize=4, marker='^', label="BP预测值", c='g')plt.title("BP算法预测情况")plt.ylabel("累计裂缝宽度（mm）")plt.xticks(xlabel2, [test_name_ls[i] for i in xlabel2], rotation=20)plt.legend()plt.grid(linestyle='-.')f5 = plt.figure()plt.plot(y_label, markersize=4, marker='.', label="真值", c='r')plt.plot(GABP_prediction, markersize=4, marker='*', label="GA-BP预测值", c='b')plt.title("GA-BP算法预测情况")plt.ylabel("累计裂缝宽度（mm）")plt.xticks(xlabel2, [test_name_ls[i] for i in xlabel2], rotation=20)plt.legend()plt.grid(linestyle='-.')plt.show()

对比结果确实有提升：

资源获取：

关注公众号 小堂同学（公众号微信号：gh_d99af1e66e91） 回复 GABP

才疏学浅，水平有限。敬请批评指正！

共勉！