forked from shizizhixin/Photovoltaic-power-prediction
-
Notifications
You must be signed in to change notification settings - Fork 0
/
power_pred.py
138 lines (92 loc) · 4.19 KB
/
power_pred.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
import tensorflow as tf
import pandas as pd
import numpy as np
#读取训练数据
df=pd.read_csv(r"C:\Users\Administrator\Desktop\shujujingsai\salor\public.train.csv",header=None)
datatrain=np.array(df)
#提取特征值，形成输入数据
#dataxs=datatrain[1:,[0,1,2,3,4,8,9,10,11,12,13,17,18,19]]#9000x8
dataxs=datatrain[1:,:20]
dataxshlen=len(dataxs)#训练输入数据的行数
dataxsllen=len(dataxs[0])#训练输入数据的列数
for i in range(dataxshlen):
for j in range(dataxsllen):
dataxs[i][j]=float(dataxs[i][j])
#形成输出数据
datays=datatrain[1:,[20]]#9000x1
datayshlen=dataxshlen#训练输出数据的行数
dataysllen=len(datays[0])#训练输出数据的列数
for i in range(datayshlen):
for j in range(dataysllen):
datays[i][j]=float(datays[i][j])
print(datays[0][0])
#归一化,datays取值范围为（0.05，0.95）
#datays_nor=0.05+(datays-np.min(datays))/(np.max(datays)-np.min(datays))*(0.95-0.05)
print(dataxs,datays,dataxshlen,dataxsllen,datayshlen,dataysllen)
#读取测试数据
df1=pd.read_csv(r"C:\Users\Administrator\Desktop\shujujingsai\salor\public.test.csv",header=None)
datatest=np.array(df1)
#提取特征值，形成输入数据
#dataxs_test=datatest[1:,[0,1,2,3,4,8,9,10,11,12,13,17,18,19]]#9000x8
dataxs_test=datatest[1:,:20]
dataxs_testhlen=len(dataxs_test)#训练输入数据的行数
dataxs_testllen=len(dataxs_test[0])#训练输入数据的列数
for i in range(dataxs_testhlen):
for j in range(dataxs_testllen):
dataxs_test[i][j]=float(dataxs_test[i][j])
#测试比对数据
df2=pd.read_csv(r"C:\Users\Administrator\Desktop\shujujingsai\salor\submit_example.csv",header=None)
datatest1=np.array(df2)
datays_test=datatest1[0:,1:2]#9000x1
datays_testhlen=dataxs_testhlen#训练输出数据的行数
datays_testllen=len(datays_test[0])#训练输出数据的列数
for i in range(datays_testhlen):
for j in range(datays_testllen):
datays_test[i][j]=float(datays_test[i][j])
print(datays_test[0][0])
#归一化,datays取值范围为（0.05，0.95）
#datays_nor_test=0.05+(datays_test-np.min(datays_test))/(np.max(datays_test)-np.min(datays_test))*(0.95-0.05)
print(dataxs_test,datays_test,dataxs_testhlen,dataxs_testllen,datays_testhlen,datays_testllen)
def add_layer(inputs,in_size,out_size,activation_function=None):
Weights=tf.Variable(tf.truncated_normal([in_size,out_size],stddev=0.1))
Biases=tf.Variable(tf.constant(0.1,shape=[out_size]))
xW_plus_b=tf.matmul(inputs,Weights)+Biases
if activation_function is None:
outputs=xW_plus_b
else:
outputs=activation_function(xW_plus_b)
return outputs
#define placeholder for inputs to network
xs=tf.placeholder(tf.float32,[None,20])#
ys=tf.placeholder(tf.float32,[None,1])
keep_prop=tf.placeholder(tf.float32)
#layer1
datays_nor_layer1=add_layer(xs,20,10,activation_function=tf.sigmoid)
#layer2
datays_nor_layer2=add_layer(datays_nor_layer1,10,20,activation_function=tf.nn.relu)
#perdiction
prediction=add_layer(datays_nor_layer2,20,1,activation_function=None)
#the error between prediction and real data
loss=tf.sqrt(tf.reduce_mean(tf.square(prediction-ys)))
train_step=tf.train.GradientDescentOptimizer(0.9).minimize(loss)
#train_step=tf.train.AdamOptimizer(1e-4).minimize(loss)
#初始化
sess=tf.Session()
sess.run(tf.global_variables_initializer())
#模型精确度
def compute_accuracy(v_xs,v_ys):
global prediction
y_pre=sess.run(prediction,feed_dict={xs:v_xs,keep_prop:0.5})
#y_pre_tru=(y_pre-0.05)*(np.max(datays)-np.min(datays))/(0.95-0.05)
accuracy_prediction=1/(1+tf.sqrt(tf.reduce_mean(tf.square(y_pre-ys))))
#accuracy_prediction_tru=1/(1+tf.sqrt(tf.reduce_mean(tf.square(y_pre_tru-datays_test))))
result=sess.run(accuracy_prediction,feed_dict={xs:v_xs,ys:v_ys,keep_prop:0.5})
print(y_pre)
return result
for k in range(2000):
#print(batch_xs,batch_ys,len(batch_xs))
sess.run(train_step,feed_dict={xs:dataxs,ys:datays,keep_prop:0.5})
#sess.run(train_step,feed_dict={xs:dataxs_test,ys:datays_nor_test})
if k%500==0:
#print(k,sess.run(loss,feed_dict={xs:dataxs,ys:datays_nor}))
print(compute_accuracy(dataxs_test,datays_test))