markilue
/
self_example
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

leecode更新

This commit is contained in:
markilue 2022-10-13 13:55:33 +08:00
parent f888fdcc78
commit 8d8994786c
6 changed files with 946 additions and 113 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

117
Leecode/src/main/java/com/markilue/leecode/backtrace/IetterCombinations.java Normal file
View File

@ -0,0 +1,117 @@
package com.markilue.leecode.backtrace;
import com.markilue.leecode.stackAndDeque.EvalRPN;
import org.junit.Test;
import java.util.*;
/**
* @BelongsProject: Leecode
* @BelongsPackage: com.markilue.leecode.backtrace
* @Author: markilue
* @CreateTime: 2022-10-13 11:08
* @Description: TODO 力扣17题 电话号码的字母组合：
* 给定一个仅包含数字 2-9 的字符串，返回所有它能表示的字母组合。答案可以按 任意顺序 返回。
* 给出数字到字母的映射如下（与电话按键相同）。注意 1 不对应任何字母。
* @Version: 1.0
*/
public class IetterCombinations {
@Test
public void test() {
System.out.println(letterCombinations(""));
}
List<List<String>> dict = new ArrayList<List<String>>();
//构造字典
public void construct() {
dict.add(new ArrayList<String>(Arrays.asList("a", "b", "c")));//2
dict.add(new ArrayList<String>(Arrays.asList("d", "e", "f")));//3
dict.add(new ArrayList<String>(Arrays.asList("g", "h", "i")));//4
dict.add(new ArrayList<String>(Arrays.asList("j", "k", "l")));//5
dict.add(new ArrayList<String>(Arrays.asList("m", "n", "o")));//6
dict.add(new ArrayList<String>(Arrays.asList("p", "q", "r", "s")));//7
dict.add(new ArrayList<String>(Arrays.asList("t", "u", "v")));//8
dict.add(new ArrayList<String>(Arrays.asList("w", "x", "y", "z")));//9
}
List<String> result = new ArrayList<>();
StringBuilder cur = new StringBuilder();
/**
* 回溯算法：速度击败48.76%,内存击败15.77%
* @param digits
* @return
*/
public List<String> letterCombinations(String digits) {
if(digits.length()==0){
return result;
}
construct();
backtracking(digits.length(), digits, 0);
return result;
}
public void backtracking(int n, String digits, int val) {
if (cur.length() == n) {
result.add(new String(cur.toString()));
return;
}
// System.out.println(digits.charAt(val)); //2
// System.out.println(Integer.valueOf(digits.charAt(val))); 50 =>ASCLL码
// System.out.println(Integer.valueOf(digits.charAt(val)).getClass());
for (String s : dict.get(Integer.valueOf(String.valueOf(digits.charAt(val)))-2)) {
cur.append(s);
backtracking(n, digits, val + 1);
cur.deleteCharAt(cur.length()-1);
}
}
/**
* 官方回溯算法：使用map记录，相比之下避免了一系列的类型转换
* @param digits
* @return
*/
public List<String> letterCombinations1(String digits) {
List<String> combinations = new ArrayList<String>();
if (digits.length() == 0) {
return combinations;
}
Map<Character, String> phoneMap = new HashMap<Character, String>() {{
put('2', "abc");
put('3', "def");
put('4', "ghi");
put('5', "jkl");
put('6', "mno");
put('7', "pqrs");
put('8', "tuv");
put('9', "wxyz");
}};
backtrack(combinations, phoneMap, digits, 0, new StringBuffer());
return combinations;
}
public void backtrack(List<String> combinations, Map<Character, String> phoneMap, String digits, int index, StringBuffer combination) {
if (index == digits.length()) {
combinations.add(combination.toString());
} else {
char digit = digits.charAt(index);
String letters = phoneMap.get(digit);
int lettersCount = letters.length();
for (int i = 0; i < lettersCount; i++) {
combination.append(letters.charAt(i));
backtrack(combinations, phoneMap, digits, index + 1, combination);
combination.deleteCharAt(index);
}
}
}
}

137
Leecode/src/main/java/com/markilue/leecode/backtrace/combinationSum.java Normal file
View File

@ -0,0 +1,137 @@
package com.markilue.leecode.backtrace;
import org.junit.Test;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;
/**
* @BelongsProject: Leecode
* @BelongsPackage: com.markilue.leecode.backtrace
* @Author: markilue
* @CreateTime: 2022-10-13 11:59
* @Description: TODO 力扣39题 组合总和：
* 给你一个 无重复元素 的整数数组 candidates 和一个目标整数 target ，找出 candidates 中可以使数字和为目标数 target 的 所有 不同组合 ，并以列表形式返回。你可以按 任意顺序 返回这些组合。
* candidates 中的 同一个 数字可以 无限制重复被选取 。如果至少一个数字的被选数量不同，则两种组合是不同的。 
* 对于给定的输入，保证和为 target 的不同组合数少于 150 个。
* @Version: 1.0
*/
public class combinationSum {
@Test
public void test() {
int[] candidates = {2, 3, 6, 7};
int target = 7;
System.out.println(combinationSum(candidates, target));
}
@Test
public void test1() {
int[] candidates = {2, 3, 5};
int target = 8;
System.out.println(combinationSum(candidates, target));
}
/**
* 回溯算法
* 速度击败93.59%，内存击败78.22%
*
* @param candidates
* @param target
* @return
*/
public List<List<Integer>> combinationSum(int[] candidates, int target) {
backtracking(candidates, target, 0);
return result;
}
List<List<Integer>> result = new ArrayList<>();
List<Integer> cur = new ArrayList<>();
int sum = 0;
/**
* @param candidates
* @param target
* @param val 当前candidates数组需要查看的数字的位置
*/
public void backtracking(int[] candidates, int target, int val) {
if (sum > target) {
return;
}
if (sum == target) {
ArrayList<Integer> list = new ArrayList<>(cur);
result.add(list);
return;
}
if (val >= candidates.length) {
return;
}
//定义边界，以用来剪枝
//加几个candidates[val]
for (int i = 0; i <= (target - sum) / candidates[val]; i++) {
for (int j = 0; j < i; j++) {
cur.add(candidates[val]);
}
sum += candidates[val] * i;
backtracking(candidates, target, val + 1);
for (int j = 0; j < i; j++) {
cur.remove(cur.size() - 1);
}
sum -= candidates[val] * i;
}
}
/**
* 代码随想录回溯法
* 速度击败100%，内存击败73.19%
*
* @param candidates
* @param target
* @return
*/
public List<List<Integer>> combinationSum1(int[] candidates, int target) {
Arrays.sort(candidates);
backtracking1(candidates, target, 0);
return result;
}
/**
* 官方回溯法
*
* @param candidates
* @param target
* @param startIndex 当前candidates数组需要查看的数字的位置
*/
public void backtracking1(int[] candidates, int target, int startIndex) {
if (sum > target) {
return;
}
if (sum == target) {
ArrayList<Integer> list = new ArrayList<>(cur);
result.add(list);
return;
}
//定义边界，以用来剪枝
//
for (int i = startIndex; i <= candidates.length && sum + candidates[i] <= target; i++) {
sum += candidates[i];
cur.add(candidates[i]);
//这里不需要i+1,表示可以重复读取当前数
backtracking1(candidates, target, i);
sum -= candidates[i];
cur.remove(cur.size() - 1);
}
}
}

123
Leecode/src/main/java/com/markilue/leecode/backtrace/combinationSum3.java Normal file
View File

@ -0,0 +1,123 @@
package com.markilue.leecode.backtrace;
import org.junit.Test;
import java.util.ArrayList;
import java.util.List;
/**
* @BelongsProject: Leecode
* @BelongsPackage: com.markilue.leecode.backtrace
* @Author: markilue
* @CreateTime: 2022-10-13 10:21
* @Description: TODO 力扣216题 组合总和III:
* 找出所有相加之和为 n 的 k 个数的组合，且满足下列条件：
* 只使用数字1到9
* 每个数字 最多使用一次 
* 返回 所有可能的有效组合的列表 。该列表不能包含相同的组合两次，组合可以以任何顺序返回。
* @Version: 1.0
*/
public class combinationSum3 {
@Test
public void test() {
int k = 3, n = 7;
System.out.println(combinationSum3(k, n));
}
@Test
public void test1() {
int k = 3, n = 9;
System.out.println(combinationSum3(k, n));
}
/**
* 回溯算法：结果如果不剪枝，需要在9^k个解法中寻找
* 速度超过100%，内存超过87.1%
* @param k
* @param n
* @return
*/
public List<List<Integer>> combinationSum3(int k, int n) {
backtracking(k, n, 1);
return result;
}
List<List<Integer>> result = new ArrayList<>();
List<Integer> cur = new ArrayList<>();
int sum = 0;
public void backtracking(int k, int n, int val) {
if (cur.size() == k) {
if (sum == n) {
ArrayList<Integer> list = new ArrayList<>();
list.addAll(cur);
result.add(list);
}
return;
}
//定义边界，以用来剪枝
int threshold = 9;
if (n - sum < 9) {
threshold = n - sum;
}
for (int i = val; i <= threshold; i++) {
cur.add(i);
sum += i;
backtracking(k, n, i + 1);
cur.remove(cur.size() - 1);
sum -= i;
}
}
List<Integer> temp = new ArrayList<Integer>();
List<List<Integer>> ans = new ArrayList<List<Integer>>();
/**
* 官方自己枚举法:通过二进制的9位数，来模拟哪个数据被选中了，被选中了就加入temp，通过遍历所有的子集来获得全部的结果
* 结果需要2^9个解法中寻找
* @param k
* @param n
* @return
*/
public List<List<Integer>> combinationSum31(int k, int n) {
for (int mask = 0; mask < (1 << 9); ++mask) {
if (check(mask, k, n)) {
ans.add(new ArrayList<Integer>(temp));
}
}
return ans;
}
public boolean check(int mask, int k, int n) {
temp.clear();
for (int i = 0; i < 9; ++i) {
if (((1 << i) & mask) != 0) {
temp.add(i + 1);
}
}
if (temp.size() != k) {
return false;
}
int sum = 0;
for (int num : temp) {
sum += num;
}
return sum == n;
}
}

162
TensorFlow_eaxmple/Model_train_test/condition_monitoring/self_try/compare/DCConv.py
View File

@ -37,6 +37,7 @@ from model.CommonFunction.CommonFunction import *
from sklearn.model_selection import train_test_split from sklearn.model_selection import train_test_split
from tensorflow.keras.models import load_model, save_model from tensorflow.keras.models import load_model, save_model
from keras.callbacks import EarlyStopping from keras.callbacks import EarlyStopping
import random
'''超参数设置''' '''超参数设置'''
time_stamp = 120 time_stamp = 120
@ -196,7 +197,7 @@ def EWMA(data, K=K, namuda=namuda):
pass pass
def get_MSE(data, label, new_model): def get_MSE(data, label, new_model, isStandard: bool = True, isPlot: bool = True):
predicted_data = new_model.predict(data) predicted_data = new_model.predict(data)
temp = np.abs(predicted_data - label) temp = np.abs(predicted_data - label)
@ -209,24 +210,31 @@ def get_MSE(data, label, new_model):
# mse=np.mean((predicted_data-label)**2,axis=1) # mse=np.mean((predicted_data-label)**2,axis=1)
print("mse", mse) print("mse", mse)
if isStandard:
dims, = mse.shape
mean = np.mean(mse)
std = np.sqrt(np.var(mse))
max = mean + 3 * std
print("max:", max)
# min = mean-3*std
max = np.broadcast_to(max, shape=[dims, ])
# min = np.broadcast_to(min,shape=[dims,])
mean = np.broadcast_to(mean, shape=[dims, ])
if isPlot:
plt.figure(random.randint(1,9))
plt.plot(max)
plt.plot(mse)
plt.plot(mean)
# plt.plot(min)
plt.show()
else:
if isPlot:
plt.figure(random.randint(1, 9))
plt.plot(mse)
# plt.plot(min)
plt.show()
return mse
dims, = mse.shape
mean = np.mean(mse)
std = np.sqrt(np.var(mse))
max = mean + 3 * std
# min = mean-3*std
max = np.broadcast_to(max, shape=[dims, ])
# min = np.broadcast_to(min,shape=[dims,])
mean = np.broadcast_to(mean, shape=[dims, ])
# plt.plot(max)
# plt.plot(mse)
# plt.plot(mean)
# # plt.plot(min)
# plt.show()
#
#
return mse, mean, max return mse, mean, max
# pass # pass
@ -340,20 +348,20 @@ def showResult(step_two_model: Joint_Monitoring, test_data, isPlot: bool = False
return total_result return total_result
def GRU_Model(): def DCConv_Model():
input = tf.keras.Input(shape=[time_stamp, feature_num]) input = tf.keras.Input(shape=[time_stamp, feature_num])
input = tf.cast(input, tf.float32) input = tf.cast(input, tf.float32)
LSTM = tf.keras.layers.Conv1D(10, 3, padding="causal",dilation_rate=2)(input) LSTM = tf.keras.layers.Conv1D(10, 3, padding="causal", dilation_rate=2)(input)
LSTM = tf.keras.layers.Conv1D(20, 3, padding="causal",dilation_rate=4)(LSTM) LSTM = tf.keras.layers.Conv1D(20, 3, padding="causal", dilation_rate=4)(LSTM)
LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal",dilation_rate=8)(LSTM) LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal", dilation_rate=8)(LSTM)
LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal",dilation_rate=16)(LSTM) LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal", dilation_rate=16)(LSTM)
LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal",dilation_rate=32)(LSTM) LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal", dilation_rate=32)(LSTM)
LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal",dilation_rate=64)(LSTM) LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal", dilation_rate=64)(LSTM)
LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal",dilation_rate=128)(LSTM) LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal", dilation_rate=128)(LSTM)
# LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal",dilation_rate=2)(LSTM) # LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal",dilation_rate=2)(LSTM)
LSTM = LSTM[:, -1, :]
# bn = tf.keras.layers.BatchNormalization()(LSTM) # bn = tf.keras.layers.BatchNormalization()(LSTM)
d1 = tf.keras.layers.Dense(20)(LSTM) d1 = tf.keras.layers.Dense(20)(LSTM)
@ -365,6 +373,44 @@ def GRU_Model():
pass pass
# healthy_data是健康数据,用于确定阈值，all_data是完整的数据,用于模型出结果
def getResult(model: tf.keras.Model, healthy_data, healthy_label, unhealthy_data, unhealthy_label, isPlot: bool = False,
isSave: bool = False):
# TODO 计算MSE确定阈值
mse, mean, max = get_MSE(healthy_data, healthy_label, model)
# 误报率的计算
total, = mse.shape
faultNum = 0
faultList = []
for i in range(total):
if (mse[i] > max[i]):
faultNum += 1
faultList.append(mse[i])
fault_rate = faultNum / total
print("误报率:", fault_rate)
# 漏报率计算
missNum = 0
missList = []
mse1 = get_MSE(unhealthy_data, unhealthy_label, model, isStandard=False)
all, = mse1.shape
flag=True
for i in range(all):
if (mse1[i] < max[0] and flag) :
missNum += 1
missList.append(mse1[i])
elif(mse1[i]>max[0]):
flag=False
print("all:",all)
miss_rate = missNum / all
print("漏报率:", miss_rate)
pass
if __name__ == '__main__': if __name__ == '__main__':
total_data = loadData.execute(N=feature_num, file_name=file_name) total_data = loadData.execute(N=feature_num, file_name=file_name)
total_data = normalization(data=total_data) total_data = normalization(data=total_data)
@ -375,7 +421,7 @@ if __name__ == '__main__':
is_Healthy=False) is_Healthy=False)
#### TODO 第一步训练 #### TODO 第一步训练
# 单次测试 # 单次测试
model = GRU_Model() model = DCConv_Model()
model.compile(optimizer=tf.optimizers.Adam(0.01), loss=tf.losses.mse) model.compile(optimizer=tf.optimizers.Adam(0.01), loss=tf.losses.mse)
model.summary() model.summary()
@ -389,49 +435,25 @@ if __name__ == '__main__':
mode='min', mode='min',
period=1) period=1)
history = model.fit(train_data_healthy[:30000, :, :], train_label1_healthy[:30000, :], epochs=20, # history = model.fit(train_data_healthy[:30000, :, :], train_label1_healthy[:30000, :], epochs=20,
batch_size=32, validation_split=0.2, shuffle=False, verbose=1, # batch_size=32, validation_split=0.2, shuffle=False, verbose=1,
callbacks=[checkpoint, early_stop]) # callbacks=[checkpoint, early_stop])
model.save(save_name) # model.save(save_name)
## TODO testing ## TODO testing
test_data, test_label = get_training_data(total_data[:300455, :]) # test_data, test_label = get_training_data(total_data[:healthy_date, :])
# newModel = tf.keras.models.load_model(save_name)
# mse, mean, max = get_MSE(test_data, test_label, new_model=newModel)
healthy_size, _, _ = train_data_healthy.shape
unhealthy_size, _, _ = train_data_unhealthy.shape
all_data, _, _ = get_training_data_overlapping(
total_data[healthy_size - 2 * unhealthy_size:unhealthy_date, :], is_Healthy=True)
newModel = tf.keras.models.load_model(save_name) newModel = tf.keras.models.load_model(save_name)
mse, mean, max = get_MSE(test_data, test_label, new_model=newModel) getResult(newModel, healthy_data=train_data_healthy[healthy_size - 2 * unhealthy_size:, :],
healthy_label=train_label1_healthy[healthy_size - 2 * unhealthy_size:, :],
test_data, test_label = get_training_data(total_data[20000:, :]) unhealthy_data=train_data_unhealthy, unhealthy_label=train_label1_unhealthy)
predicted_data = newModel.predict(test_data) # mse, mean, max = get_MSE(train_data_healthy[healthy_size - 2 * unhealthy_size:, :],
rows, cols = predicted_data.shape # train_label1_healthy[healthy_size - 2 * unhealthy_size:, :], new_model=newModel)
temp = np.abs(predicted_data - test_label)
temp1 = (temp - np.broadcast_to(np.mean(temp, axis=0), shape=predicted_data.shape))
temp2 = np.broadcast_to(np.sqrt(np.var(temp, axis=0)), shape=predicted_data.shape)
temp3 = temp1 / temp2
mse = np.sum((temp1 / temp2) ** 2, axis=1)
plt.plot(mse)
plt.plot(mean)
plt.plot(max)
plt.show()
data = pd.DataFrame(mse).ewm(span=3).mean()
print(data)
data = np.array(data)
index, _ = data.shape
for i in range(2396):
if data[i, 0] > 5:
data[i, 0] = data[i - 1, :]
print(data)
mean = data[2000:2396, :].mean()
std = data[2000:2396, :].std()
mean = np.broadcast_to(mean, shape=[500, ])
std = np.broadcast_to(std, shape=[500, ])
plt.plot(data[2000:2396, :])
plt.plot(mean)
plt.plot(mean + 3 * std)
plt.plot(mean - 3 * std)
plt.show()
pass pass

439
TensorFlow_eaxmple/Model_train_test/condition_monitoring/self_try/compare/GRU-CNN.py
View File

@ -6,5 +6,440 @@
@Author : dingjiawen @Author : dingjiawen
@Date : 2022/10/11 18:52 @Date : 2022/10/11 18:52
@Usage : 对比实验，与JointNet相同深度,进行预测 @Usage : 对比实验，与JointNet相同深度,进行预测
@Desc : @Desc : CNN-GRU
''' '''
import tensorflow as tf
import tensorflow.keras
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from model.DepthwiseCon1D.DepthwiseConv1D import DepthwiseConv1D
from model.Dynamic_channelAttention.Dynamic_channelAttention import DynamicChannelAttention
from condition_monitoring.data_deal import loadData
from model.Joint_Monitoring.Joint_Monitoring3 import Joint_Monitoring
from model.CommonFunction.CommonFunction import *
from sklearn.model_selection import train_test_split
from tensorflow.keras.models import load_model, save_model
from keras.callbacks import EarlyStopping
import random
'''超参数设置'''
time_stamp = 120
feature_num = 10
batch_size = 16
learning_rate = 0.001
EPOCH = 101
model_name = "DCNN-GRU"
'''EWMA超参数'''
K = 18
namuda = 0.01
'''保存名称'''
save_name = "./model/{0}_timestamp{1}_feature{2}.h5".format(model_name,
time_stamp,
feature_num,
batch_size,
EPOCH)
save_step_two_name = "../hard_model/two_weight/{0}_timestamp{1}_feature{2}_weight_epoch14/weight".format(model_name,
time_stamp,
feature_num,
batch_size,
EPOCH)
# save_name = "../model/joint/{0}_timestamp{1}_feature{2}.h5".format(model_name,
# time_stamp,
# feature_num,
# batch_size,
# EPOCH)
# save_step_two_name = "../model/joint_two/{0}_timestamp{1}_feature{2}.h5".format(model_name,
# time_stamp,
# feature_num,
# batch_size,
# EPOCH)
'''文件名'''
file_name = "G:\data\SCADA数据\jb4q_8_delete_total_zero.csv"
'''
文件说明：jb4q_8_delete_total_zero.csv是删除了只删除了全是0的列的文件
文件从0:415548行均是正常值(2019/7.30 00:00:00 - 2019/9/18 11:14:00)
从415549:432153行均是异常值(2019/9/18 11:21:01 - 2021/1/18 00:00:00)
'''
'''文件参数'''
# 最后正常的时间点
healthy_date = 415548
# 最后异常的时间点
unhealthy_date = 432153
# 异常容忍程度
unhealthy_patience = 5
def remove(data, time_stamp=time_stamp):
rows, cols = data.shape
print("remove_data.shape:", data.shape)
num = int(rows / time_stamp)
return data[:num * time_stamp, :]
pass
# 不重叠采样
def get_training_data(data, time_stamp: int = time_stamp):
removed_data = remove(data=data)
rows, cols = removed_data.shape
print("removed_data.shape:", data.shape)
print("removed_data:", removed_data)
train_data = np.reshape(removed_data, [-1, time_stamp, cols])
print("train_data:", train_data)
batchs, time_stamp, cols = train_data.shape
for i in range(1, batchs):
each_label = np.expand_dims(train_data[i, 0, :], axis=0)
if i == 1:
train_label = each_label
else:
train_label = np.concatenate([train_label, each_label], axis=0)
print("train_data.shape:", train_data.shape)
print("train_label.shape", train_label.shape)
return train_data[:-1, :], train_label
# 重叠采样
def get_training_data_overlapping(data, time_stamp: int = time_stamp, is_Healthy: bool = True):
rows, cols = data.shape
train_data = np.empty(shape=[rows - time_stamp - 1, time_stamp, cols])
train_label = np.empty(shape=[rows - time_stamp - 1, cols])
for i in range(rows):
if i + time_stamp >= rows:
break
if i + time_stamp < rows - 1:
train_data[i] = data[i:i + time_stamp]
train_label[i] = data[i + time_stamp]
print("重叠采样以后：")
print("data:", train_data) # (300334,120,10)
print("label:", train_label) # (300334,10)
if is_Healthy:
train_label2 = np.ones(shape=[train_label.shape[0]])
else:
train_label2 = np.zeros(shape=[train_label.shape[0]])
print("label2:", train_label2)
return train_data, train_label, train_label2
# 归一化
def normalization(data):
rows, cols = data.shape
print("归一化之前:", data)
print(data.shape)
print("======================")
# 归一化
max = np.max(data, axis=0)
max = np.broadcast_to(max, [rows, cols])
min = np.min(data, axis=0)
min = np.broadcast_to(min, [rows, cols])
data = (data - min) / (max - min)
print("归一化之后:", data)
print(data.shape)
return data
# 正则化
def Regularization(data):
rows, cols = data.shape
print("正则化之前:", data)
print(data.shape)
print("======================")
# 正则化
mean = np.mean(data, axis=0)
mean = np.broadcast_to(mean, shape=[rows, cols])
dst = np.sqrt(np.var(data, axis=0))
dst = np.broadcast_to(dst, shape=[rows, cols])
data = (data - mean) / dst
print("正则化之后:", data)
print(data.shape)
return data
pass
def EWMA(data, K=K, namuda=namuda):
# t是啥暂时未知
t = 0
mid = np.mean(data, axis=0)
standard = np.sqrt(np.var(data, axis=0))
UCL = mid + K * standard * np.sqrt(namuda / (2 - namuda) * (1 - (1 - namuda) ** 2 * t))
LCL = mid - K * standard * np.sqrt(namuda / (2 - namuda) * (1 - (1 - namuda) ** 2 * t))
return mid, UCL, LCL
pass
def get_MSE(data, label, new_model, isStandard: bool = True, isPlot: bool = True):
predicted_data = new_model.predict(data)
temp = np.abs(predicted_data - label)
temp1 = (temp - np.broadcast_to(np.mean(temp, axis=0), shape=predicted_data.shape))
temp2 = np.broadcast_to(np.sqrt(np.var(temp, axis=0)), shape=predicted_data.shape)
temp3 = temp1 / temp2
mse = np.sum((temp1 / temp2) ** 2, axis=1)
print("z:", mse)
print(mse.shape)
# mse=np.mean((predicted_data-label)**2,axis=1)
print("mse", mse)
if isStandard:
dims, = mse.shape
mean = np.mean(mse)
std = np.sqrt(np.var(mse))
max = mean + 3 * std
print("max:", max)
# min = mean-3*std
max = np.broadcast_to(max, shape=[dims, ])
# min = np.broadcast_to(min,shape=[dims,])
mean = np.broadcast_to(mean, shape=[dims, ])
if isPlot:
plt.figure(random.randint(1,9))
plt.plot(max)
plt.plot(mse)
plt.plot(mean)
# plt.plot(min)
plt.show()
else:
if isPlot:
plt.figure(random.randint(1, 9))
plt.plot(mse)
# plt.plot(min)
plt.show()
return mse
return mse, mean, max
# pass
def condition_monitoring_model():
input = tf.keras.Input(shape=[time_stamp, feature_num])
conv1 = tf.keras.layers.Conv1D(filters=256, kernel_size=1)(input)
GRU1 = tf.keras.layers.GRU(128, return_sequences=False)(conv1)
d1 = tf.keras.layers.Dense(300)(GRU1)
output = tf.keras.layers.Dense(10)(d1)
model = tf.keras.Model(inputs=input, outputs=output)
return model
# trian_data:(300455,120,10)
# trian_label1:(300455,10)
# trian_label2:(300455,)
def shuffle(train_data, train_label1, train_label2, is_split: bool = False, split_size: float = 0.2):
(train_data, test_data, train_label1, test_label1, train_label2, test_label2) = train_test_split(train_data,
train_label1,
train_label2,
test_size=split_size,
shuffle=True,
random_state=100)
if is_split:
return train_data, train_label1, train_label2, test_data, test_label1, test_label2
train_data = np.concatenate([train_data, test_data], axis=0)
train_label1 = np.concatenate([train_label1, test_label1], axis=0)
train_label2 = np.concatenate([train_label2, test_label2], axis=0)
# print(train_data.shape)
# print(train_label1.shape)
# print(train_label2.shape)
# print(train_data.shape)
return train_data, train_label1, train_label2
pass
def split_test_data(healthy_data, healthy_label1, healthy_label2, unhealthy_data, unhealthy_label1, unhealthy_label2,
split_size: float = 0.2, shuffle: bool = True):
data = np.concatenate([healthy_data, unhealthy_data], axis=0)
label1 = np.concatenate([healthy_label1, unhealthy_label1], axis=0)
label2 = np.concatenate([healthy_label2, unhealthy_label2], axis=0)
(train_data, test_data, train_label1, test_label1, train_label2, test_label2) = train_test_split(data,
label1,
label2,
test_size=split_size,
shuffle=shuffle,
random_state=100)
# print(train_data.shape)
# print(train_label1.shape)
# print(train_label2.shape)
# print(train_data.shape)
return train_data, train_label1, train_label2, test_data, test_label1, test_label2
pass
def test(step_one_model, step_two_model, test_data, test_label1, test_label2):
history_loss = []
history_val_loss = []
val_loss, val_accuracy = step_two_model.get_val_loss(val_data=test_data, val_label1=test_label1,
val_label2=test_label2,
is_first_time=False, step_one_model=step_one_model)
history_val_loss.append(val_loss)
print("val_accuracy:", val_accuracy)
print("val_loss:", val_loss)
def showResult(step_two_model: Joint_Monitoring, test_data, isPlot: bool = False):
# 获取模型的所有参数的个数
# step_two_model.count_params()
total_result = []
size, length, dims = test_data.shape
for epoch in range(0, size - batch_size + 1, batch_size):
each_test_data = test_data[epoch:epoch + batch_size, :, :]
_, _, _, output4 = step_two_model.call(each_test_data, is_first_time=False)
total_result.append(output4)
total_result = np.reshape(total_result, [total_result.__len__(), -1])
total_result = np.reshape(total_result, [-1, ])
if isPlot:
plt.scatter(list(range(total_result.shape[0])), total_result, c='black', s=10)
# 画出 y=1 这条水平线
plt.axhline(0.5, c='red', label='Failure threshold')
# 箭头指向上面的水平线
# plt.arrow(35000, 0.9, 33000, 0.75, head_width=0.02, head_length=0.1, shape="full", fc='red', ec='red',
# alpha=0.9, overhang=0.5)
# plt.text(35000, 0.9, "Truth Fault", fontsize=10, color='black', verticalalignment='top')
plt.axvline(test_data.shape[0] * 2 / 3, c='blue', ls='-.')
plt.xlabel("time")
plt.ylabel("confience")
plt.text(total_result.shape[0] * 4 / 5, 0.6, "Fault", fontsize=10, color='black', verticalalignment='top',
horizontalalignment='center',
bbox={'facecolor': 'grey',
'pad': 10})
plt.text(total_result.shape[0] * 1 / 3, 0.4, "Norm", fontsize=10, color='black', verticalalignment='top',
horizontalalignment='center',
bbox={'facecolor': 'grey',
'pad': 10})
plt.grid()
# plt.ylim(0, 1)
# plt.xlim(-50, 1300)
# plt.legend("", loc='upper left')
plt.show()
return total_result
def CNN_GRU_Model():
input = tf.keras.Input(shape=[time_stamp, feature_num])
input = tf.cast(input, tf.float32)
LSTM = tf.keras.layers.Conv1D(10, 3, padding="causal", dilation_rate=2)(input)
LSTM = tf.keras.layers.Conv1D(20, 3, padding="causal", dilation_rate=4)(LSTM)
LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal", dilation_rate=8)(LSTM)
LSTM = tf.keras.layers.Conv1D(40, 3, padding="causal", dilation_rate=8)(LSTM)
LSTM = tf.keras.layers.GRU(units=40, return_sequences=True)(LSTM)
LSTM = tf.keras.layers.GRU(units=40, return_sequences=True)(LSTM)
LSTM = tf.keras.layers.GRU(units=40, return_sequences=True)(LSTM)
LSTM = tf.keras.layers.GRU(units=40, return_sequences=False)(LSTM)
# bn = tf.keras.layers.BatchNormalization()(LSTM)
d1 = tf.keras.layers.Dense(20)(LSTM)
# bn = tf.keras.layers.BatchNormalization()(d1)
output = tf.keras.layers.Dense(10, name='output')(d1)
model = tf.keras.Model(inputs=input, outputs=output)
return model
pass
# healthy_data是健康数据,用于确定阈值，all_data是完整的数据,用于模型出结果
def getResult(model: tf.keras.Model, healthy_data, healthy_label, unhealthy_data, unhealthy_label, isPlot: bool = False,
isSave: bool = False):
# TODO 计算MSE确定阈值
mse, mean, max = get_MSE(healthy_data, healthy_label, model)
# 误报率的计算
total, = mse.shape
faultNum = 0
faultList = []
faultNum=mse[mse[:]>max[0]].__len__()
# for i in range(total):
# if (mse[i] > max[i]):
# faultNum += 1
# faultList.append(mse[i])
fault_rate = faultNum / total
print("误报率:", fault_rate)
# 漏报率计算
missNum = 0
missList = []
mse1 = get_MSE(unhealthy_data, unhealthy_label, model, isStandard=False)
all, = mse1.shape
flag=True
for i in range(all):
if (mse1[i] < max[0] and flag) :
missNum += 1
missList.append(mse1[i])
elif(mse1[i]>max[0]):
print(i)
flag=False
print("all:",all)
miss_rate = missNum / all
print("漏报率:", miss_rate)
pass
if __name__ == '__main__':
total_data = loadData.execute(N=feature_num, file_name=file_name)
total_data = normalization(data=total_data)
train_data_healthy, train_label1_healthy, train_label2_healthy = get_training_data_overlapping(
total_data[:healthy_date, :], is_Healthy=True)
train_data_unhealthy, train_label1_unhealthy, train_label2_unhealthy = get_training_data_overlapping(
total_data[healthy_date - time_stamp + unhealthy_patience:unhealthy_date, :],
is_Healthy=False)
#### TODO 第一步训练
# 单次测试
model = CNN_GRU_Model()
model.compile(optimizer=tf.optimizers.Adam(0.01), loss=tf.losses.mse)
model.summary()
early_stop = EarlyStopping(monitor='val_loss', min_delta=0.0001, patience=3, mode='min', verbose=1)
checkpoint = tf.keras.callbacks.ModelCheckpoint(
filepath=save_name,
monitor='val_loss',
verbose=1,
save_best_only=True,
mode='min',
period=1)
# history = model.fit(train_data_healthy[:30000, :, :], train_label1_healthy[:30000, :], epochs=20,
# batch_size=32, validation_split=0.2, shuffle=False, verbose=1,
# callbacks=[checkpoint, early_stop])
# model.save(save_name)
## TODO testing
# test_data, test_label = get_training_data(total_data[:healthy_date, :])
# newModel = tf.keras.models.load_model(save_name)
# mse, mean, max = get_MSE(test_data, test_label, new_model=newModel)
healthy_size, _, _ = train_data_healthy.shape
unhealthy_size, _, _ = train_data_unhealthy.shape
all_data, _, _ = get_training_data_overlapping(
total_data[healthy_size - 2 * unhealthy_size:unhealthy_date, :], is_Healthy=True)
newModel = tf.keras.models.load_model(save_name)
getResult(newModel, healthy_data=train_data_healthy[healthy_size - 2 * unhealthy_size:, :],
healthy_label=train_label1_healthy[healthy_size - 2 * unhealthy_size:, :],
unhealthy_data=train_data_unhealthy, unhealthy_label=train_label1_unhealthy)
# mse, mean, max = get_MSE(train_data_healthy[healthy_size - 2 * unhealthy_size:, :],
# train_label1_healthy[healthy_size - 2 * unhealthy_size:, :], new_model=newModel)
pass

81
TensorFlow_eaxmple/Model_train_test/condition_monitoring/self_try/compare/RNet.py
View File

@ -17,15 +17,12 @@ import matplotlib.pyplot as plt
from model.DepthwiseCon1D.DepthwiseConv1D import DepthwiseConv1D from model.DepthwiseCon1D.DepthwiseConv1D import DepthwiseConv1D
from model.Dynamic_channelAttention.Dynamic_channelAttention import DynamicChannelAttention from model.Dynamic_channelAttention.Dynamic_channelAttention import DynamicChannelAttention
from condition_monitoring.data_deal import loadData from condition_monitoring.data_deal import loadData
from model.Joint_Monitoring.Joint_Monitoring3 import Joint_Monitoring from model.Joint_Monitoring.compare.RNet import Joint_Monitoring
from model.CommonFunction.CommonFunction import * from model.CommonFunction.CommonFunction import *
from sklearn.model_selection import train_test_split from sklearn.model_selection import train_test_split
from tensorflow.keras.models import load_model, save_model from tensorflow.keras.models import load_model, save_model
'''超参数设置''' '''超参数设置'''
time_stamp = 120 time_stamp = 120
feature_num = 10 feature_num = 10
@ -38,16 +35,16 @@ K = 18
namuda = 0.01 namuda = 0.01
'''保存名称''' '''保存名称'''
save_name = "../hard_model/weight/{0}_timestamp{1}_feature{2}_weight_epoch8/weight".format(model_name, save_name = "./model/weight/{0}_timestamp{1}_feature{2}_weight/weight".format(model_name,
time_stamp,
feature_num,
batch_size,
EPOCH)
save_step_two_name = "./model/two_weight/{0}_timestamp{1}_feature{2}_weight/weight".format(model_name,
time_stamp, time_stamp,
feature_num, feature_num,
batch_size, batch_size,
EPOCH) EPOCH)
save_step_two_name = "../hard_model/two_weight/{0}_timestamp{1}_feature{2}_weight_epoch14/weight".format(model_name,
time_stamp,
feature_num,
batch_size,
EPOCH)
# save_name = "../model/joint/{0}_timestamp{1}_feature{2}.h5".format(model_name, # save_name = "../model/joint/{0}_timestamp{1}_feature{2}.h5".format(model_name,
# time_stamp, # time_stamp,
@ -536,8 +533,10 @@ if __name__ == '__main__':
is_Healthy=False) is_Healthy=False)
#### TODO 第一步训练 #### TODO 第一步训练
# 单次测试 # 单次测试
# train_step_one(train_data=train_data_healthy[:32, :, :], train_label1=train_label1_healthy[:32, :],train_label2=train_label2_healthy[:32, ]) # train_step_one(train_data=train_data_healthy[:256, :, :], train_label1=train_label1_healthy[:256, :],
# train_step_one(train_data=train_data_healthy, train_label1=train_label1_healthy, train_label2=train_label2_healthy) # train_label2=train_label2_healthy[:256, ])
train_step_one(train_data=train_data_healthy, train_label1=train_label1_healthy, train_label2=train_label2_healthy)
# 导入第一步已经训练好的模型,一个继续训练，一个只输出结果 # 导入第一步已经训练好的模型,一个继续训练，一个只输出结果
# step_one_model = Joint_Monitoring() # step_one_model = Joint_Monitoring()
@ -546,34 +545,34 @@ if __name__ == '__main__':
# step_two_model = Joint_Monitoring() # step_two_model = Joint_Monitoring()
# step_two_model.load_weights(save_name) # step_two_model.load_weights(save_name)
#### TODO 第二步训练 # #### TODO 第二步训练
### healthy_data.shape: (300333,120,10) # ### healthy_data.shape: (300333,120,10)
### unhealthy_data.shape: (16594,10) # ### unhealthy_data.shape: (16594,10)
healthy_size, _, _ = train_data_healthy.shape # healthy_size, _, _ = train_data_healthy.shape
unhealthy_size, _, _ = train_data_unhealthy.shape # unhealthy_size, _, _ = train_data_unhealthy.shape
# train_data, train_label1, train_label2, test_data, test_label1, test_label2 = split_test_data( # # train_data, train_label1, train_label2, test_data, test_label1, test_label2 = split_test_data(
# healthy_data=train_data_healthy[healthy_size - 2 * unhealthy_size:, :, :], # # healthy_data=train_data_healthy[healthy_size - 2 * unhealthy_size:, :, :],
# healthy_label1=train_label1_healthy[healthy_size - 2 * unhealthy_size:, :], # # healthy_label1=train_label1_healthy[healthy_size - 2 * unhealthy_size:, :],
# healthy_label2=train_label2_healthy[healthy_size - 2 * unhealthy_size:, ], unhealthy_data=train_data_unhealthy, # # healthy_label2=train_label2_healthy[healthy_size - 2 * unhealthy_size:, ], unhealthy_data=train_data_unhealthy,
# unhealthy_label1=train_label1_unhealthy, unhealthy_label2=train_label2_unhealthy) # # unhealthy_label1=train_label1_unhealthy, unhealthy_label2=train_label2_unhealthy)
# train_step_two(step_one_model=step_one_model, step_two_model=step_two_model, # # train_step_two(step_one_model=step_one_model, step_two_model=step_two_model,
# train_data=train_data, # # train_data=train_data,
# train_label1=train_label1, train_label2=np.expand_dims(train_label2, axis=-1)) # # train_label1=train_label1, train_label2=np.expand_dims(train_label2, axis=-1))
#
### TODO 测试测试集 # ### TODO 测试测试集
step_one_model = Joint_Monitoring() # step_one_model = Joint_Monitoring()
step_one_model.load_weights(save_name) # step_one_model.load_weights(save_name)
step_two_model = Joint_Monitoring() # step_two_model = Joint_Monitoring()
step_two_model.load_weights(save_step_two_name) # step_two_model.load_weights(save_step_two_name)
# test(step_one_model=step_one_model, step_two_model=step_two_model, test_data=test_data, test_label1=test_label1, # # test(step_one_model=step_one_model, step_two_model=step_two_model, test_data=test_data, test_label1=test_label1,
# test_label2=np.expand_dims(test_label2, axis=-1)) # # test_label2=np.expand_dims(test_label2, axis=-1))
#
###TODO 展示全部的结果 # ###TODO 展示全部的结果
all_data, _, _ = get_training_data_overlapping( # all_data, _, _ = get_training_data_overlapping(
total_data[healthy_size - 2 * unhealthy_size:unhealthy_date, :], is_Healthy=True) # total_data[healthy_size - 2 * unhealthy_size:unhealthy_date, :], is_Healthy=True)
# all_data = np.concatenate([]) # # all_data = np.concatenate([])
# 单次测试 # # 单次测试
# showResult(step_two_model, test_data=all_data[:32], isPlot=True) # # showResult(step_two_model, test_data=all_data[:32], isPlot=True)
showResult(step_two_model, test_data=all_data, isPlot=True) # showResult(step_two_model, test_data=all_data, isPlot=True)
pass pass