#电池老化率测定的神经网络模型 import tensorflow as tf import numpy as np import matplotlib.pyplot as plt import pandas as pd path = 'SOH_Data.xlsx' #训练集读取及归一化 xTrainData = pd.read_excel(path, sheetname = 0) yTrainData = pd.read_excel(path, sheetname = 1) n1 = np.shape(xTrainData)[1] x_data = np.array(xTrainData).astype('float32') for i in range(n1): x_data[:, i] = (x_data[:, i] - np.amin(x_data[:, i]))/(np.amax(x_data[:, i]) - np.amin(x_data[:, i])) y_data = np.array(yTrainData).astype('float32') y_data[:] = (y_data[:] - np.amin(y_data[:]))/(np.amax(y_data[:]) - np.amin(y_data[:])) #测试集读取及归一化 xTestData = pd.read_excel(path, sheetname = 2) yTestData = pd.read_excel(path, sheetname = 3) xTest = np.array(xTestData).astype('float32') n2 = np.shape(xTrainData)[1] xTrain = np.array(xTrainData).astype('float32') for i in range(n2): xTest[:, i] = (xTest[:, i] - np.amin(xTest[:, i]))/(np.amax(xTest[:, i]) - np.amin(xTest[:, i])) yTest = np.array(yTestData).astype('float32') yTest[:] = (yTest[:] - np.amin(yTest[:]))/(np.amax(yTest[:]) - np.amin(yTest[:])) #参数概要 def variable_summaries(var): with tf.name_scope('summaries'): mean = tf.reduce_mean(var)#平均值 tf.summary.scalar('mean', mean) with tf.name_scope('stddev'): stddev = tf.sqrt(tf.reduce_mean(tf.square(var - mean))) tf.summary.scalar('stddev', stddev)#标准差 tf.summary.scalar('max', tf.reduce_max(var))#最大值 tf.summary.scalar('min', tf.reduce_min(var))#最小值 tf.summary.histogram('histogram', var)#直方图 #5层神经网络，每层神经元个数 IHO = [12, 8, 5, 4, 1] #命名空间 with tf.name_scope('input'): #定义两个placeholder x = tf.placeholder(tf.float32, [None, 12], name = 'xInput') y = tf.placeholder(tf.float32, [None, 1], name = 'y') #神经元中间层 with tf.name_scope('layer'): with tf.name_scope('weights_L1'): Weight_L1 = tf.Variable(tf.random_normal([12, 8]), name = 'W1') variable_summaries(Weight_L1) with tf.name_scope('bias_L1'): biases_L1 = tf.Variable(tf.zeros([8]), name = 'b1') variable_summaries(biases_L1) with tf.name_scope('L_1'): Wx_plus_b_L1 = tf.matmul(x, Weight_L1) + biases_L1 L1 = tf.nn.tanh(Wx_plus_b_L1) with tf.name_scope('weights_L2'): Weight_L2 = tf.Variable(tf.random_normal([8, 5]), name = 'W2') variable_summaries(Weight_L2) with tf.name_scope('bias_L2'): biases_L2 = tf.Variable(tf.zeros([5]), name = 'b2') variable_summaries(biases_L2) with tf.name_scope('L_2'): Wx_plus_b_L2 = tf.matmul(L1, Weight_L2) + biases_L2 L2 = tf.nn.tanh(Wx_plus_b_L2) with tf.name_scope('weights_L3'): Weight_L3 = tf.Variable(tf.random_normal([5, 4]), name = 'W3') variable_summaries(Weight_L3) with tf.name_scope('bias_L3'): biases_L3 = tf.Variable(tf.zeros([4]), name = 'b3') variable_summaries(biases_L3) with tf.name_scope('L_3'): Wx_plus_b_L3 = tf.matmul(L2, Weight_L3) + biases_L3 L3 = tf.nn.tanh(Wx_plus_b_L3) #神经元输出层 with tf.name_scope('weights_L4'): Weight_L4 = tf.Variable(tf.random_normal([4, 1]), name = 'W4') variable_summaries(Weight_L4) with tf.name_scope('bias_L4'): biases_L4 = tf.Variable(tf.zeros([1]), name = 'b4') variable_summaries(biases_L4) with tf.name_scope('prediction'): Wx_plus_b_L4 = tf.matmul(L3, Weight_L4) + biases_L4 prediction = tf.nn.tanh(Wx_plus_b_L4) #二次代价函数 with tf.name_scope('loss'): loss = tf.reduce_mean(tf.square(y - prediction), name = 'loss') tf.summary.scalar('loss', loss) #使用梯度下降法训练 with tf.name_scope('train'): train_step = tf.train.GradientDescentOptimizer(0.01).minimize(loss) #合并所有summary merged = tf.summary.merge_all() with tf.Session() as sess: #变量初始化 sess.run(tf.global_variables_initializer()) writer = tf.summary.FileWriter('logs/', sess.graph) for i in range(10000): summary, _ = sess.run([merged, train_step], feed_dict = {x: x_data, y: y_data}) writer.add_summary(summary, i) curr_loss = sess.run(loss, feed_dict = {x: x_data, y: y_data}) if (i + 1)%100 == 0: print('第%d次迭代loss:'%(i + 1), curr_loss) #训练集预测集 prediction_value = sess.run(prediction, feed_dict = {x: x_data}) #测试集预测集 prediction_value_test = sess.run(prediction, feed_dict = {x: xTest}) test_loss = sess.run(loss, feed_dict = {x: xTest, y: yTest}) print('测试误差：', test_loss) print(prediction_value_test)
#电池老化率测定的神经网络模型 import tensorflow as tf import numpy as np import matplotlib.pyplot as plt import pandas as pd from numpy import * import os np.set_printoptions(suppress=True) np.set_printoptions(threshold=np.NaN) BATCH_SIZE = 256 with open("D:\\bs\\finall_data\\marry2.11\\train\\nc_all.txt","rb") as fa,open("D:\\bs\\finall_data\\marry2.11\\train\\calipso_all.txt","rb") as fb,open("D:\\bs\\finall_data\\marry2.11\\test\\nc_text.txt","rb") as fc,open("D:\\bs\\finall_data\\marry2.11\\test\\calipso_text.txt","rb") as fd: #训练集读取及归一化 # xTrainData = pd.read_excel(path, sheetname = 0) # yTrainData = pd.read_excel(path, sheetname = 1) # # print(xTrainData) # # print(yTrainData) # n1 = np.shape(xTrainData)[1] # # print(n1) # x_data = np.array(xTrainData).astype('float32') # print(x_data) # # def Polyfit(x, y, degree): # results = {} # coeffs = np.polyfit(x, y, degree) # # results['polynomial'] = coeffs.tolist() # # # r-squared # p = np.poly1d(coeffs) # # print(p) # # # fit values, and mean # # yhat = p(x) # or [p(z) for z in x] # # ybar = np.sum(y) / len(y) # or sum(y)/len(y) # # ssreg = np.sum((yhat - ybar) ** 2) # or sum([ (yihat - ybar)**2 for yihat in yhat]) # # sstot = np.sum((y - ybar) ** 2) # or sum([ (yi - ybar)**2 for yi in y]) # # results['determination'] = ssreg / sstot # 准确率 # return results list_x = [] for i in fa.readlines(): # print(str(i)) x_data_1 = str(i).split(" ")[2:18] for x_data_12 in x_data_1: x_data_12=float(x_data_12) list_x.append(x_data_12) mat_x = mat(list_x) x_data = mat_x.reshape(-1, 16) for i in range(16): x_data[i, :] = (x_data[i, :] - np.amin(x_data[i, :])) / (np.amax(x_data[i, :]) - np.amin(x_data[i, :])) list_y=[] for v in fb.readlines(): y_data_1 = str(v).split(" ")[2].split(" ")[0] y_data_1=1/(1+float(y_data_1)) # print(y_data) list_y.append(float(y_data_1)) # print(list_y) mat_y=mat(list_y) y_data = mat_y.reshape(-1, 1) # print(y_data) # y_data = np.array(yTrainData).astype('float32') # y_data[:] = (y_data[:] - np.amin(y_data[:]))/(np.amax(y_data[:]) - np.amin(y_data[:])) # # # # print(y_data[:]) # z1 = Polyfit(x_data, y_data, 2) # plt.plot(x_data, y_data, 'o') # # plt.plot(x_data, np.polyval(z1, x_data)) # plt.show() # # #测试集读取及归一化 list_t_x = [] for m in fc.readlines(): x_data_t_1 = str(m).split(" ")[2:18] for x_data_t_12 in x_data_t_1: # print(x_data_t_12) x_data_t_12 = float(x_data_t_12) list_t_x.append(x_data_t_12) mat_t_x = mat(list_t_x) xTest = mat_t_x.reshape(-1, 16) # print(xTest.shape) #(1598,16) # xTestData = pd.read_excel(path, sheetname = 2) # yTestData = pd.read_excel(path, sheetname = 3) # xTest = np.array(xTestData).astype('float32') # n2 = np.shape(xTrainData)[1] # xTrain = np.array(xTrainData).astype('float32') for i in range(16): xTest[i, :] = (xTest[i, :] - np.amin(xTest[i, :]))/(np.amax(xTest[i, :]) - np.amin(xTest[i, :])) list_t_y = [] for n in fd.readlines(): y_data_t_1 = str(n).split(" ")[2].split(" ")[0] # print(y_data) y_data_t_1=1/(1+float(y_data_t_1)) list_t_y.append(float(y_data_t_1)) # print(list_y) mat_t_y = mat(list_t_y) yTest = mat_t_y.reshape(-1, 1) # print(yTest) # yTest = np.array(yTestData).astype('float32') # yTest[:] = (yTest[:] - np.amin(yTest[:]))/(np.amax(yTest[:]) - np.amin(yTest[:])) # print(np.amax(yTest[:])) # print(np.amax(y_Test[:])) #参数概要 def variable_summaries(var): with tf.name_scope('summaries'): mean = tf.reduce_mean(var)#平均值 tf.summary.scalar('mean', mean) with tf.name_scope('stddev'): stddev = tf.sqrt(tf.reduce_mean(tf.square(var - mean))) tf.summary.scalar('stddev', stddev)#标准差 tf.summary.scalar('max', tf.reduce_max(var))#最大值 tf.summary.scalar('min', tf.reduce_min(var))#最小值 tf.summary.histogram('histogram', var)#直方图 #5层神经网络，每层神经元个数 IHO = [16, 8, 5, 4, 1] #命名空间 with tf.name_scope('input'): #定义两个placeholder x = tf.placeholder(tf.float32, [None, 16], name = 'xInput') y = tf.placeholder(tf.float32, [None, 1], name = 'y') #神经元中间层 with tf.name_scope('layer'): with tf.name_scope('weights_L1'): Weight_L1 = tf.Variable(tf.random_normal([16, 8]), name = 'W1') variable_summaries(Weight_L1) with tf.name_scope('bias_L1'): biases_L1 = tf.Variable(tf.zeros([8]), name = 'b1') variable_summaries(biases_L1) with tf.name_scope('L_1'): Wx_plus_b_L1 = tf.matmul(x, Weight_L1) + biases_L1 L1 = tf.nn.sigmoid(Wx_plus_b_L1) with tf.name_scope('weights_L2'): Weight_L2 = tf.Variable(tf.random_normal([8, 5]), name = 'W2') variable_summaries(Weight_L2) with tf.name_scope('bias_L2'): biases_L2 = tf.Variable(tf.zeros([5]), name = 'b2') variable_summaries(biases_L2) with tf.name_scope('L_2'): Wx_plus_b_L2 = tf.matmul(L1, Weight_L2) + biases_L2 L2 = tf.nn.sigmoid(Wx_plus_b_L2) with tf.name_scope('weights_L3'): Weight_L3 = tf.Variable(tf.random_normal([5, 4]), name = 'W3') variable_summaries(Weight_L3) with tf.name_scope('bias_L3'): biases_L3 = tf.Variable(tf.zeros([4]), name = 'b3') variable_summaries(biases_L3) with tf.name_scope('L_3'): Wx_plus_b_L3 = tf.matmul(L2, Weight_L3) + biases_L3 L3 = tf.nn.sigmoid(Wx_plus_b_L3) #神经元输出层 with tf.name_scope('weights_L4'): Weight_L4 = tf.Variable(tf.random_normal([4, 1]), name = 'W4') variable_summaries(Weight_L4) with tf.name_scope('bias_L4'): biases_L4 = tf.Variable(tf.zeros([1]), name = 'b4') variable_summaries(biases_L4) with tf.name_scope('prediction'): Wx_plus_b_L4 = tf.matmul(L3, Weight_L4) + biases_L4 prediction = tf.nn.sigmoid(Wx_plus_b_L4) #二次代价函数 with tf.name_scope('loss'): loss = tf.reduce_mean(tf.square(y - prediction), name = 'loss') tf.summary.scalar('loss', loss) #使用梯度下降法训练 with tf.name_scope('train'): train_step = tf.train.GradientDescentOptimizer(0.01).minimize(loss) #合并所有summary merged = tf.summary.merge_all() with tf.Session() as sess: #变量初始化 sess.run(tf.global_variables_initializer()) writer = tf.summary.FileWriter('logs/', sess.graph) for i in range(1000): summary, _ = sess.run([merged, train_step], feed_dict = {x: x_data, y: y_data}) writer.add_summary(summary, i) curr_loss = sess.run(loss, feed_dict = {x: x_data, y: y_data}) if (i + 1)%100 == 0: print('第%d次迭代loss:'%(i + 1), curr_loss) #训练集预测集 prediction_value = sess.run(prediction, feed_dict = {x: x_data}) #测试集预测集 prediction_value_test = sess.run(prediction, feed_dict = {x: xTest}) test_loss = sess.run(loss, feed_dict = {x: xTest, y: yTest}) print('测试误差：', test_loss) # print(len(prediction_value_test)) #text数据的个数 # print(yTest) print(prediction_value_test)
from __future__ import print_function import numpy as np np.random.seed(1337) from keras.datasets import mnist from keras.models import Sequential from keras.layers.core import Dense, Dropout, Activation from keras.optimizers import RMSprop from keras.utils import np_utils from keras.optimizers import SGD batch_size = 128 nb_classes = 10 nb_epoch = 20 (X_train, y_train), (X_test, y_test) = mnist.load_data() X_train = X_train.reshape(60000, 784) X_test = X_test.reshape(10000, 784) X_train = X_train.astype('float32') X_test = X_test.astype('float32') X_train /= 255 X_test /= 255 print(X_train.shape[0], 'train samples') print(X_test.shape[0], 'test samples') Y_train = np_utils.to_categorical(y_train, nb_classes) Y_test = np_utils.to_categorical(y_test, nb_classes) model1 = Sequential() model1.add(Dense(256, activation='relu', input_dim=784)) model1.add(Dropout(0.2)) model1.add(Dense(256, activation='relu')) model1.add(Dropout(0.2)) model1.add(Dense(10, activation='softmax')) sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True) model1.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy']) history1 = model1.fit(X_train, Y_train, batch_size = batch_size, epochs = nb_epoch, verbose = 2, validation_data = (X_test, Y_test)) model2 = Sequential() model2.add(Dense(256, activation='relu', input_dim=784)) model2.add(Dense(256, activation='relu')) model2.add(Dense(10, activation='softmax')) sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True) model2.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy']) history2 = model2.fit(X_train, Y_train, batch_size = batch_size, epochs = nb_epoch, verbose = 2, validation_data = (X_test, Y_test)) model3 = Sequential() model3.add(Dense(256, activation='relu', input_dim=784)) model3.add(Dense(10, activation='softmax')) sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True) model3.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy']) history3 = model3.fit(X_train, Y_train, batch_size = batch_size, epochs = nb_epoch, verbose = 2, validation_data = (X_test, Y_test)) import matplotlib.pyplot as plt # list all data in history print(history.history.keys()) # summarize history for accuracy plt.plot(history1.history['acc']) plt.plot(history1.history['val_acc']) plt.title('model1 accuracy') plt.ylabel('accuracy') plt.xlabel('epoch') plt.legend(['train', 'test'], loc='upper left') plt.show() plt.plot(history2.history['acc']) plt.plot(history2.history['val_acc']) plt.title('model2 accuracy') plt.ylabel('accuracy') plt.xlabel('epoch') plt.legend(['train', 'test'], loc='upper left') plt.show() plt.plot(history3.history['acc']) plt.plot(history3.history['val_acc']) plt.title('model3 accuracy') plt.ylabel('accuracy') plt.xlabel('epoch') plt.legend(['train', 'test'], loc='upper left') plt.show() plt.plot(history1.history['val_acc']) plt.plot(history2.history['val_acc']) plt.plot(history3.history['val_acc']) plt.title('model accuracy') plt.ylabel('accuracy') plt.xlabel('epoch') plt.legend(['model1', 'model2', 'model3'], loc='upper left') plt.show() # summarize history for loss plt.plot(history1.history['loss']) plt.plot(history1.history['val_loss']) plt.title('model1 loss') plt.ylabel('loss') plt.xlabel('epoch') plt.legend(['train', 'test'], loc='upper left') plt.show() plt.plot(history2.history['loss']) plt.plot(history2.history['val_loss']) plt.title('model2 loss') plt.ylabel('loss') plt.xlabel('epoch') plt.legend(['train', 'test'], loc='upper left') plt.show() plt.plot(history3.history['loss']) plt.plot(history3.history['val_loss']) plt.title('model3 loss') plt.ylabel('loss') plt.xlabel('epoch') plt.legend(['train', 'test'], loc='upper left') plt.show()
