这是用python在本地做的数据库连接操作，源代码如下：
# -*- coding: utf-8 -*-
"""
Created on Fri Jul 19 09:23:19 2019@author: sunwf1114
"""from tensorflow import keras
import numpy as np # 用于转换数据
import pandas as pd # 用于分析数据集
import time
import pymysqlglobal conn, cur
conn = pymysql.Connect(host="127.0.0.1", port=3306, user="***", passwd="****", db="energy",charset='utf8',cursorclass=pymysql.cursors.DictCursor) # ,cursorclass = MySQLdb.cursors.DictCursor
cur = conn.cursor()powerResult = {}
index = ['lengji1', 'lengji2', 'lengji3', 'lengji4']for i in index:powerResult[i] = 0class model_pre():def load_model1(self): # 加载模型with open(r'./lengjiPower/1/model/model.json', 'r') as file:model_json1 = file.read()model = keras.models.model_from_json(model_json1) # 返回模型结构model.load_weights("./lengjiPower/1/model/model.hdf5", by_name=False) # 加载模型的权重return model# 加载基本数据进行归一化def read_min_max(self):filename = "./lengjiPower/1/model/data.txt"with open(filename, 'r+', encoding='utf-8') as f:for lines in f.readlines():ss = lines.strip('[]').split(', ')ds = np.asarray(ss)li_max = []li_min = []for i in range(0, len(ds), 2): # 分成最大最小分别存储li_max.append(ds[i])li_min.append(ds[i + 1])nu_max = np.asarray(li_max) # 转换为数组nu_min = np.asarray(li_min)return nu_max, nu_mindef check1(self):while True:self.data_pre()# print("depty")time.sleep(2)def load_model2(self): # 加载模型with open(r'./lengjiPower/2/model/model.json', 'r') as file:model_json1 = file.read()model = keras.models.model_from_json(model_json1) # 返回模型结构model.load_weights("./lengjiPower/2/model/model.hdf5", by_name=False) # 加载模型的权重return modeldef check2(self):while True:self.data_pre()# print("depty")time.sleep(2)def load_model3(self): # 加载模型with open(r'./lengjiPower/3/model/model.json', 'r') as file:model_json1 = file.read()model = keras.models.model_from_json(model_json1) # 返回模型结构model.load_weights("./lengjiPower/3/model/model.hdf5", by_name=False) # 加载模型的权重return modeldef check3(self):while True:self.data_pre()# print("depty")time.sleep(2)def load_model4(self): # 加载模型with open(r'./lengjiPower/4/model/model.json', 'r') as file:model_json1 = file.read()model = keras.models.model_from_json(model_json1) # 返回模型结构model.load_weights("./lengjiPower/4/model/model.hdf5", by_name=False) # 加载模型的权重return modeldef check4(self):while True:self.data_pre()# print("depty")time.sleep(2)# 参数是一个列表，每一个元素代表一组数据：[冷冻水回水温度,冷冻水出水温度,冷却水回水温度,冷水机组蒸发侧压力（kg）,冷水机组冷凝侧压力（kg）,冷冻水流量,输入功率]def data_pre_lengji1gonglv(self):global cur, conn# y_max = 439.0 #根据原始数据最大值做为基准# y_min = 345.0# read = pd.read_excel('data_input.xlsx')# ts = read.iloc[:,0:read.columns.size].values #读取前4列，分别是流量、进水温度、出水温度、COP###############lengji1Gonglv######################try:conn.ping()except pymysql.OperationalError:conn = pymysql.Connect(host="127.0.0.1", port=3306, user="**", passwd="***", db="energy",charset='utf8',cursorclass=MySQLdb.cursors.DictCursor) # ,cursorclass = MySQLdb.cursors.DictCursorcur = conn.cursor()sql_select = 'Select * from lengji1_power order by id desc limit 1'cur.execute(sql_select)conn.commit()results = cur.fetchall()listSequence = []for i in results[0]:if i != 'power' and i != 'id':listSequence.append(results[0][i])ts = []ts.append(listSequence)ds = np.asarray(ts) # ts转换为数组矩阵# wnd_sz = read.columns.size #定义矩阵宽度wnd_sz = len(results[0]) - 2print(ts, wnd_sz)x_1 = ds[:, 0:wnd_sz] # 输出前n - 1维的特征max, min = self.read_min_max() # 加载数据中的最大最小值用于还原归一化x_nomal = np.full((len(x_1), wnd_sz - 1), -1000., dtype=float) # 创建一个空数组,用来存储float类型的归一值for i in range(0, len(x_1)):for j in range(0, len(x_1[0])):x = x_1[i]new = x_nomal[i]up = float(x[j]) - float(min[j])down = float(max[j]) - float(min[j])new[j] = up / down # 归一化#x_nomal = x_1 # MinMaxScaler().fit_transform(x_1) #归一化model = self.load_model1() # 加载模型y = model.predict(x_nomal) # 预测加载值# y = y #y*(y_max - y_min) + y_min# y = np.asarray(y)# if os.path.exists('data_out.txt') != True:# path_txt = "data_out.txt"# file = open(path_txt, 'w')# file.write(str(y))powerResult['lengji1'] = int(y)###############lengji2Gonglv######################try:conn.ping()except pymysql.OperationalError:conn = pymysql.Connect(host="127.0.0.1", port=3306, user="***", passwd="****", db="energy",charset='utf8',cursorclass=MySQLdb.cursors.DictCursor) # ,cursorclass = MySQLdb.cursors.DictCursorcur = conn.cursor()sql_select = 'Select * from lengji2_power order by id desc limit 1'cur.execute(sql_select)conn.commit()results = cur.fetchall()listSequence = []for i in results[0]:if i != 'power' and i != 'id':listSequence.append(results[0][i])ts = []ts.append(listSequence)ds = np.asarray(ts) # ts转换为数组矩阵# wnd_sz = read.columns.size #定义矩阵宽度wnd_sz = len(results[0]) - 2print(ts, wnd_sz)x_1 = ds[:, 0:wnd_sz] # 输出前n - 1维的特征x_nomal = x_1 # MinMaxScaler().fit_transform(x_1) #归一化model = self.load_model2() # 加载模型y = model.predict(x_nomal) # 预测加载值# y = y #y*(y_max - y_min) + y_min# y = np.asarray(y)# if os.path.exists('data_out.txt') != True:# path_txt = "data_out.txt"# file = open(path_txt, 'w')# file.write(str(y))powerResult['lengji2'] = int(y)def dataWrite_lengji1gonglv(self):global conn, curtablename = 'powerdata'try:conn.ping()except pymysql.OperationalError:conn = pymysql.Connect(host="127.0.0.1", port=3306, user="****", passwd="****", db="energy",charset='utf8') # ,cursorclass = MySQLdb.cursors.DictCursorcur = conn.cursor()sql_delete = 'DELETE FROM powerdata WHERE id = 0'cur.execute(sql_delete)conn.commit()sql = "INSERT INTO %s" % tablename + "(id,lengji1)\VALUES (%s,%s)"param = (0, powerResult['lengji1'])cur.execute(sql, param)conn.commit()if __name__ == '__main__':# print("hello world")back = model_pre()print('System Started!')while True:time.sleep(2)back.data_pre_lengji1gonglv()back.dataWrite_lengji1gonglv()