,

1 year ago · 8a2d8f2a9a
parent 359effce4c
commit 8a2d8f2a9a
6 changed files with 560 additions and 25 deletions
--- a/浙江行业电量/7月全行业.csv
+++ b/浙江行业电量/7月全行业.csv
@ -0,0 +1,81 @@
+,real,pred
+0,366314897.6,325991300.0
+1,380820733.0,290698080.0
+2,380238801.9,257544720.0
+3,244227473.0,281246980.0
+4,364073058.1,275352350.0
+5,380820733.0,347751680.0
+6,380238801.9,321158200.0
+7,244227473.0,292571360.0
+8,364073058.1,298532380.0
+9,316723431.2,319373630.0
+10,380238801.9,343127650.0
+11,244227473.0,333881150.0
+12,364073058.1,315865570.0
+13,316723431.2,321350340.0
+14,334702116.4,339486850.0
+15,244227473.0,355201730.0
+16,364073058.1,351205660.0
+17,316723431.2,340886050.0
+18,334702116.4,344197900.0
+19,340252127.5,356064770.0
+20,364073058.1,273813060.0
+21,316723431.2,292272900.0
+22,334702116.4,293830340.0
+23,340252127.5,293426180.0
+24,338685923.8,300031800.0
+25,316723431.2,349388480.0
+26,334702116.4,348153700.0
+27,340252127.5,350227100.0
+28,338685923.8,351487040.0
+29,323789488.1,347344860.0
+30,334702116.4,305196900.0
+31,340252127.5,312199360.0
+32,338685923.8,318501820.0
+33,323789488.1,309094460.0
+34,327864337.1,311822370.0
+35,340252127.5,332496220.0
+36,338685923.8,331613600.0
+37,323789488.1,330546100.0
+38,327864337.1,327573700.0
+39,308162792.6,328044900.0
+40,338685923.8,335969060.0
+41,323789488.1,331136770.0
+42,327864337.1,323793180.0
+43,308162792.6,321710080.0
+44,291282417.7,328650000.0
+45,323789488.1,319168770.0
+46,327864337.1,319473630.0
+47,308162792.6,315658180.0
+48,291282417.7,316581950.0
+49,326988587.6,322722780.0
+50,327864337.1,304423680.0
+51,308162792.6,311155600.0
+52,291282417.7,315857400.0
+53,326988587.6,318633470.0
+54,338531879.5,318105200.0
+55,308162792.6,332961340.0
+56,291282417.7,335313200.0
+57,326988587.6,345534460.0
+58,338531879.5,343698900.0
+59,343441750.8,337725980.0
+60,291282417.7,301866880.0
+61,326988587.6,310824100.0
+62,338531879.5,321738430.0
+63,343441750.8,311232700.0
+64,328373555.6,309342560.0
+65,326988587.6,306569470.0
+66,338531879.5,310643680.0
+67,343441750.8,314744000.0
+68,328373555.6,308885660.0
+69,307601302.1,306099400.0
+70,338531879.5,321590140.0
+71,343441750.8,318885440.0
+72,328373555.6,318865300.0
+73,307601302.1,314202980.0
+74,296591378.5,314701920.0
+75,343441750.8,322403870.0
+76,328373555.6,318382370.0
+77,307601302.1,315469380.0
+78,296591378.5,312369950.0
+79,291360430.5,316049300.0
--- a/浙江行业电量/7月第三产业.csv
+++ b/浙江行业电量/7月第三产业.csv
@ -0,0 +1,81 @@
+,real,pred
+0,135488382.9,126543270.0
+1,139728296.4,132820504.0
+2,137821025.8,111796240.0
+3,78600422.1,131913600.0
+4,126378434.7,121779730.0
+5,139728296.4,133275510.0
+6,137821025.8,132085740.0
+7,78600422.1,123508616.0
+8,126378434.7,122062880.0
+9,112584670.1,126883176.0
+10,137821025.8,127550950.0
+11,78600422.1,127659760.0
+12,126378434.7,120534296.0
+13,112584670.1,120153140.0
+14,126043317.8,126331544.0
+15,78600422.1,124620616.0
+16,126378434.7,125789710.0
+17,112584670.1,120741016.0
+18,126043317.8,122845240.0
+19,127571038.2,126469840.0
+20,126378434.7,89965520.0
+21,112584670.1,97125790.0
+22,126043317.8,97640040.0
+23,127571038.2,97961870.0
+24,126803223.9,100093896.0
+25,112584670.1,118327300.0
+26,126043317.8,119570870.0
+27,127571038.2,120226640.0
+28,126803223.9,122849880.0
+29,121261267.3,119103896.0
+30,126043317.8,112915770.0
+31,127571038.2,113519840.0
+32,126803223.9,115918376.0
+33,121261267.3,112486350.0
+34,124085800.1,111817100.0
+35,127571038.2,122143170.0
+36,126803223.9,119428750.0
+37,121261267.3,117216230.0
+38,124085800.1,115505320.0
+39,111506694.9,116549080.0
+40,126803223.9,123125660.0
+41,121261267.3,118972860.0
+42,124085800.1,113853950.0
+43,111506694.9,114263896.0
+44,106343825.4,117712260.0
+45,121261267.3,118652850.0
+46,124085800.1,116123000.0
+47,111506694.9,112103380.0
+48,106343825.4,114997496.0
+49,123690748.8,118147590.0
+50,124085800.1,113974550.0
+51,111506694.9,114976400.0
+52,106343825.4,114994730.0
+53,123690748.8,119030696.0
+54,126477907.1,118913830.0
+55,111506694.9,128059680.0
+56,106343825.4,128777160.0
+57,123690748.8,132264690.0
+58,126477907.1,134432380.0
+59,127784089.0,130678440.0
+60,106343825.4,117309140.0
+61,123690748.8,120235380.0
+62,126477907.1,124647930.0
+63,127784089.0,121395256.0
+64,121978961.0,120084936.0
+65,123690748.8,115211520.0
+66,126477907.1,116868504.0
+67,127784089.0,118030080.0
+68,121978961.0,115720190.0
+69,115593904.4,115036584.0
+70,126477907.1,122232480.0
+71,127784089.0,120755330.0
+72,121978961.0,119494640.0
+73,115593904.4,118518104.0
+74,108270758.4,118762740.0
+75,127784089.0,121758630.0
+76,121978961.0,119449144.0
+77,115593904.4,117008670.0
+78,108270758.4,116879120.0
+79,106730768.7,118586610.0
--- a/浙江行业电量/7月预测1.csv
+++ b/浙江行业电量/7月预测1.csv
@ -0,0 +1,81 @@
+,real,pred
+0,50.73,-3852886.8
+1,47.88,19293490.0
+2,48.31,-787532.5
+3,57.75,17889912.0
+4,49.95,14566443.0
+5,47.88,-2105174.2
+6,48.31,9359497.0
+7,57.75,-433206.5
+8,49.95,7017683.0
+9,39.61,7703169.0
+10,48.31,-808642.3
+11,57.75,4597720.0
+12,49.95,-360913.0
+13,39.61,3072979.0
+14,56.33,4244487.0
+15,57.75,-51642.7
+16,49.95,2367718.2
+17,39.61,-302210.2
+18,56.33,1611450.0
+19,52.92,2502897.2
+20,49.95,276224.3
+21,39.61,1291416.2
+22,56.33,-257380.97
+23,52.92,980514.25
+24,21.66,1559440.0
+25,39.61,357924.16
+26,56.33,741489.3
+27,52.92,-223402.89
+28,21.66,639626.94
+29,17.11,1005113.4
+30,56.33,332503.4
+31,52.92,443570.0
+32,21.66,-191401.42
+33,17.11,425019.1
+34,21.02,662274.2
+35,52.92,272184.7
+36,21.66,272402.28
+37,17.11,-157965.39
+38,21.02,281494.4
+39,21.65,444933.7
+40,21.66,216378.39
+41,17.11,174190.5
+42,21.02,-121474.26
+43,21.65,189702.75
+44,17.95,310332.28
+45,17.11,170874.25
+46,21.02,115950.27
+47,21.65,-86960.06
+48,17.95,130457.57
+49,32.64,225947.27
+50,21.02,136433.27
+51,21.65,80172.625
+52,17.95,-57612.453
+53,32.64,93036.65
+54,36.2,172486.55
+55,21.65,111274.96
+56,17.95,57095.465
+57,32.64,-35822.484
+58,36.2,68347.86
+59,36.29,137399.97
+60,17.95,91150.695
+61,32.64,40596.707
+62,36.2,-20513.984
+63,36.29,51582.69
+64,34.19,112940.664
+65,32.64,78077.64
+66,36.2,31135.309
+67,36.29,-9841.227
+68,34.19,40612.53
+69,14.42,97921.65
+70,36.2,70152.695
+71,36.29,25340.3
+72,34.19,-927.8079
+73,14.42,34880.83
+74,17.77,88737.21
+75,36.29,62387.992
+76,34.19,19532.104
+77,14.42,3516.0425
+78,17.77,29056.807
+79,126.08,80504.98
--- a/浙江行业电量/7月预测2.csv
+++ b/浙江行业电量/7月预测2.csv
@ -0,0 +1,81 @@
+,real,pred
+0,2704531.598,-2316342.0
+1,2847401.395,20676706.0
+2,2853452.254,621314.75
+3,1445328.766,19298740.0
+4,2859827.803,15913489.0
+5,2847401.395,160415.81
+6,2853452.254,11473635.0
+7,1445328.766,1780781.9
+8,2859827.803,9122751.0
+9,2596527.222,9748031.0
+10,2853452.254,1740487.1
+11,1445328.766,7060606.5
+12,2859827.803,2215774.0
+13,2596527.222,5502353.0
+14,2493576.394,6607667.0
+15,1445328.766,2590090.2
+16,2859827.803,4988868.5
+17,2596527.222,2414713.2
+18,2493576.394,4189292.0
+19,2590630.491,5007022.5
+20,2859827.803,2663833.5
+21,2596527.222,3738563.2
+22,2493576.394,2267029.0
+23,2590630.491,3384996.8
+24,2542066.764,3895295.0
+25,2596527.222,2791455.5
+26,2493576.394,3239150.8
+27,2590630.491,2324025.0
+28,2542066.764,3109880.2
+29,2348001.351,3393537.5
+30,2493576.394,2799275.0
+31,2590630.491,2971549.0
+32,2542066.764,2372881.5
+33,2348001.351,2928644.5
+34,2358799.425,3085286.2
+35,2590630.491,2734414.0
+36,2542066.764,2794325.2
+37,2348001.351,2387036.8
+38,2358799.425,2778122.0
+39,2280758.623,2866578.2
+40,2542066.764,2720565.0
+41,2348001.351,2730321.8
+42,2358799.425,2451633.2
+43,2280758.623,2726225.2
+44,2178461.623,2774300.0
+45,2348001.351,2671150.2
+46,2358799.425,2668631.5
+47,2280758.623,2480967.5
+48,2178461.623,2668544.2
+49,2333089.921,2699806.2
+50,2358799.425,2573983.0
+51,2280758.623,2581621.8
+52,2178461.623,2458319.0
+53,2333089.921,2591283.8
+54,2464330.824,2604670.0
+55,2280758.623,2550985.0
+56,2178461.623,2560103.5
+57,2333089.921,2483460.2
+58,2464330.824,2577773.5
+59,2501636.637,2571584.0
+60,2178461.623,2486985.5
+61,2333089.921,2497864.5
+62,2464330.824,2443193.2
+63,2501636.637,2499882.5
+64,2391662.7,2499099.0
+65,2333089.921,2403869.2
+66,2464330.824,2416259.8
+67,2501636.637,2374731.2
+68,2391662.7,2414558.5
+69,2132738.014,2411228.2
+70,2464330.824,2412634.2
+71,2501636.637,2417201.2
+72,2391662.7,2387954.0
+73,2132738.014,2420058.0
+74,2089421.05,2411550.0
+75,2501636.637,2422625.8
+76,2391662.7,2422028.2
+77,2132738.014,2402778.5
+78,2089421.05,2425684.2
+79,2169828.292,2417022.0
--- a/浙江行业电量/行业电量_输出为5.py
+++ b/浙江行业电量/行业电量_输出为5.py
@ -86,7 +86,7 @@ for excel in os.listdir(file_dir)[1:]:
        dataset_y = np.concatenate((dataset_y,y))


-
+print(dataset_x.shape,dataset_y.shape)
 # 训练
 device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

@ -112,22 +112,23 @@ train_y = torch.from_numpy(train_y).to(device).type(torch.float32)
 model = LSTM_Regression(DAYS_FOR_TRAIN, 32, output_size=5, num_layers=2).to(device)  # 导入模型并设置模型的参数输入输出层、隐藏层等


-train_loss = []
-loss_function = nn.MSELoss()
-optimizer = torch.optim.Adam(model.parameters(), lr=0.005, betas=(0.9, 0.999), eps=1e-08, weight_decay=0)
-for i in range(1500):
-    out = model(train_x)
-    loss = loss_function(out, train_y)
-    loss.backward()
-    optimizer.step()
-    optimizer.zero_grad()
-    train_loss.append(loss.item())
-    if i % 100 == 0:
-        print(f'epoch {i+1}: loss:{loss}')
+# train_loss = []
+# loss_function = nn.MSELoss()
+# optimizer = torch.optim.Adam(model.parameters(), lr=0.005, betas=(0.9, 0.999), eps=1e-08, weight_decay=0)
+# for i in range(1500):
+#     out = model(train_x)
+#     loss = loss_function(out, train_y)
+#     loss.backward()
+#     optimizer.step()
+#     optimizer.zero_grad()
+#     train_loss.append(loss.item())
+#     if i % 100 == 0:
+#         print(f'epoch {i+1}: loss:{loss}')

-# 保存模型
-torch.save(model.state_dict(),'hy5.pth')
+# 保存/读取模型
+# torch.save(model.state_dict(),'hy5.pth')

+model.load_state_dict(torch.load('hy5.pth'))
 # for test
 model = model.eval()  # 转换成测试模式
 # model.load_state_dict(torch.load(os.path.join(model_save_dir,model_file)))  # 读取参数
@ -139,25 +140,40 @@ pred_test = model(dataset_x)  # 全量训练集
 pred_test = pred_test.view(-1)
 pred_test = np.concatenate((np.zeros(DAYS_FOR_TRAIN), pred_test.cpu().detach().numpy()))

-plt.plot(pred_test.reshape(-1), 'r', label='prediction')
-plt.plot(dataset_y.reshape(-1), 'b', label='real')
-plt.plot((train_size*5, train_size*5), (0, 1), 'g--')  # 分割线 左边是训练数据 右边是测试数据的输出
-plt.legend(loc='best')
-plt.show()
+# plt.plot(pred_test.reshape(-1), 'r', label='prediction')
+# plt.plot(dataset_y.reshape(-1), 'b', label='real')
+# plt.plot((train_size*5, train_size*5), (0, 1), 'g--')  # 分割线 左边是训练数据 右边是测试数据的输出
+# plt.legend(loc='best')
+# plt.show()


 # 创建测试集
-# result_list = []
-# 以x为基础实际数据，滚动预测未来3天
-# x = torch.from_numpy(df[-14:-4]).to(device)
-# pred = model(x.reshape(-1,1,DAYS_FOR_TRAIN)).view(-1).detach().numpy()

+df_eval = pd.read_excel(r'C:\Users\user\Desktop\浙江各地市行业电量数据\ 杭州 .xlsx',index_col='stat_date')
+df_eval.columns = df_eval.columns.map(lambda x:x.strip())
+df_eval.index = pd.to_datetime(df_eval.index)
+
+x,y = create_dataset(df_eval.loc['2023-7']['第三产业'],10)
+
+x = (x - min_value) / (max_value - min_value)
+x = x.reshape(-1,1,10)
+
+x = torch.from_numpy(x).type(torch.float32).to(device)
+pred = model(x)
+
+# 反归一化
+pred = pred * (max_value - min_value) + min_value
+# df = df * (max_value - min_value) + min_value
+
+df = pd.DataFrame({'real':y.reshape(-1),'pred':pred.view(-1).cpu().detach().numpy()})
+print(df)
+df.to_csv('7月第三产业.csv',encoding='gbk')

 # 反归一化
 # pred = pred * (max_value - min_value) + min_value
 # df = df * (max_value - min_value) + min_value

-# print(pred)
+
 # # 打印指标
 # print(abs(pred - df[-3:]).mean() / df[-3:].mean())
 # result_eight = pd.DataFrame({'pred': np.round(pred,1),'real': df[-3:]})
--- a/电压等级_输出为5.py
+++ b/电压等级_输出为5.py
@ -0,0 +1,195 @@
+import numpy as np
+import pandas as pd
+import torch
+from torch import nn
+from multiprocessing import Pool
+import matplotlib.pyplot as plt
+import os
+os.environ["KMP_DUPLICATE_LIB_OK"]="TRUE"
+DAYS_FOR_TRAIN = 10
+torch.manual_seed(42)
+class LSTM_Regression(nn.Module):
+
+    def __init__(self, input_size, hidden_size, output_size=1, num_layers=2):
+        super().__init__()
+
+        self.lstm = nn.LSTM(input_size, hidden_size, num_layers)
+        self.fc = nn.Linear(hidden_size, output_size)
+
+    def forward(self, _x):
+        x, _ = self.lstm(_x)  # _x is input, size (seq_len, batch, input_size)
+        s, b, h = x.shape  # x is output, size (seq_len, batch, hidden_size)
+        x = x.view(s * b, h)
+        x = self.fc(x)
+        x = x.view(s, b, -1)  # 把形状改回来
+        return x
+
+
+def create_dataset(data, days_for_train=5) -> (np.array, np.array):
+    dataset_x, dataset_y = [], []
+    for i in range(len(data) - days_for_train-5):
+        dataset_x.append(data[i:(i + days_for_train)])
+        dataset_y.append(data[i + days_for_train:i + days_for_train+5])
+        # print(dataset_x,dataset_y)
+    return (np.array(dataset_x), np.array(dataset_y))
+
+def normal(nd):
+    high = nd.describe()['75%'] + 1.5*(nd.describe()['75%']-nd.describe()['25%'])
+    low = nd.describe()['25%'] - 1.5*(nd.describe()['75%']-nd.describe()['25%'])
+    return nd[(nd<high)&(nd>low)]
+
+def data_preprocessing(data):
+    data.columns = data.columns.map(lambda x: x.strip())
+    data.index = pd.to_datetime(data.index)
+    data.sort_index(inplace=True)
+    data = data.loc['2021-01':'2023-08']
+    data.drop(columns=[i for i in data.columns if (data[i] == 0).sum() / len(data) >= 0.5], inplace=True)  # 去除0值列
+    data = data[data.values != 0]
+    data = data.astype(float)
+    for col in data.columns:
+        data[col] = normal(data[col])
+
+    return data
+
+
+if __name__ == '__main__':
+    # 拼接数据集
+    file_dir = r'C:\Users\user\Desktop\浙江各地市分电压日电量数据'
+    excel = os.listdir(file_dir)[0]
+
+    data = pd.read_excel(os.path.join(file_dir, excel), sheet_name=0, index_col=' stat_date  ')
+
+    data = data_preprocessing(data)
+
+    df = data[data.columns[0]]
+    df.dropna(inplace = True)
+    dataset_x, dataset_y = create_dataset(df, DAYS_FOR_TRAIN)
+
+    for level in data.columns[1:]:
+        df = data[level]
+        df.dropna(inplace=True)
+        x, y = create_dataset(df, DAYS_FOR_TRAIN)
+        dataset_x = np.concatenate((dataset_x, x))
+        dataset_y = np.concatenate((dataset_y, y))
+
+
+    for excel in os.listdir(file_dir)[1:]:
+        data = pd.read_excel(os.path.join(file_dir,excel), sheet_name=0,index_col=' stat_date  ')
+        data = data_preprocessing(data)
+
+        for level in data.columns:
+            df = data[level]
+            df.dropna(inplace=True)
+            x,y = create_dataset(df,DAYS_FOR_TRAIN)
+            dataset_x = np.concatenate((dataset_x,x))
+            dataset_y = np.concatenate((dataset_y,y))
+
+    print(dataset_x,dataset_y,dataset_x.shape,dataset_y.shape)
+
+    # 训练
+    device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+
+    # 标准化到0~1
+    max_value = np.max(dataset_x)
+    min_value = np.min(dataset_x)
+    dataset_x = (dataset_x - min_value) / (max_value - min_value)
+    dataset_y = (dataset_y - min_value) / (max_value - min_value)
+
+    # 划分训练集和测试集
+    train_size = int(len(dataset_x)*0.7)
+
+    train_x = dataset_x[:train_size]
+    train_y = dataset_y[:train_size]
+
+    # 将数据改变形状，RNN 读入的数据维度是 (seq_size, batch_size, feature_size)
+    train_x = train_x.reshape(-1, 1, DAYS_FOR_TRAIN)
+    train_y = train_y.reshape(-1, 1, 5)
+
+    # 转为pytorch的tensor对象
+    train_x = torch.from_numpy(train_x).to(device).type(torch.float32)
+    train_y = torch.from_numpy(train_y).to(device).type(torch.float32)
+
+    model = LSTM_Regression(DAYS_FOR_TRAIN, 32, output_size=5, num_layers=2).to(device)  # 导入模型并设置模型的参数输入输出层、隐藏层等
+
+
+    train_loss = []
+    loss_function = nn.MSELoss()
+    optimizer = torch.optim.Adam(model.parameters(), lr=0.005, betas=(0.9, 0.999), eps=1e-08, weight_decay=0)
+
+    # for i in range(1500):
+    #     out = model(train_x)
+    #     loss = loss_function(out, train_y)
+    #     loss.backward()
+    #     optimizer.step()
+    #     optimizer.zero_grad()
+    #     train_loss.append(loss.item())
+    #     # print(loss)
+    # # 保存模型
+    # torch.save(model.state_dict(),'dy5.pth')
+
+
+    model.load_state_dict(torch.load('dy5.pth'))
+
+    # for test
+    model = model.eval()  # 转换成测试模式
+    # model.load_state_dict(torch.load(os.path.join(model_save_dir,model_file)))  # 读取参数
+    dataset_x = dataset_x.reshape(-1, 1, DAYS_FOR_TRAIN)  # (seq_size, batch_size, feature_size)
+    dataset_x = torch.from_numpy(dataset_x).to(device).type(torch.float32)
+
+    pred_test = model(dataset_x)  # 全量训练集
+
+    # 模型输出 (seq_size, batch_size, output_size)
+    pred_test = pred_test.view(-1)
+    pred_test = np.concatenate((np.zeros(DAYS_FOR_TRAIN), pred_test.cpu().detach().numpy()))
+
+    # plt.plot(pred_test.reshape(-1), 'r', label='prediction')
+    # plt.plot(dataset_y.reshape(-1), 'b', label='real')
+    # plt.plot((train_size*5, train_size*5), (0, 1), 'g--')  # 分割线 左边是训练数据 右边是测试数据的输出
+    # plt.legend(loc='best')
+    # plt.show()
+
+
+
+
+
+
+
+# 创建测试集
+# result_list = []
+# 以x为基础实际数据，滚动预测未来3天
+df_eval = pd.read_excel(r'C:\Users\user\Desktop\浙江各地市分电压日电量数据\杭州.xlsx',index_col=' stat_date  ')
+df_eval.columns = df_eval.columns.map(lambda x:x.strip())
+df_eval.index = pd.to_datetime(df_eval.index)
+
+x,y = create_dataset(df_eval.loc['2023-7']['10kv以下'],10)
+
+x = (x - min_value) / (max_value - min_value)
+x = x.reshape(-1,1,10)
+
+x = torch.from_numpy(x).type(torch.float32).to(device)
+pred = model(x)
+
+
+# 反归一化
+pred = pred * (max_value - min_value) + min_value
+# df = df * (max_value - min_value) + min_value
+
+print(pred,y)
+df = pd.DataFrame({'real':y.reshape(-1),'pred':pred.view(-1).cpu().detach().numpy()})
+df.to_csv('7月全行业.csv',encoding='gbk')
+
+# 打印指标
+# print(abs(pred - df[-3:]).mean() / df[-3:].mean())
+# result_eight = pd.DataFrame({'pred': np.round(pred,1),'real': df[-3:]})
+# target = (result_eight['pred'].sum() - result_eight['real'].sum()) / df[-31:].sum()
+# result_eight['loss_rate'] = round(target, 5)
+# result_eight['level'] = level
+# list_app.append(result_eight)
+# print(target)
+# print(result_eight)
+# final_df = pd.concat(list_app,ignore_index=True)
+# final_df.to_csv('市行业电量.csv',encoding='gbk')
+# print(final_df)
+
+
+