numpy 波谷检测算法返回波峰数据

我已经做了一个算法来检测一个波从波谷开始到下一个通过所花费的时间，以计算所有单独波的持续时间，但是通过函数不断返回一些波峰。

算法

import pandas as pd
import numpy as np
import peakutils

# Read the data from the CSV file
df = pd.read_csv('test.csv')

# Convert the first column to datetime format
df['Column1'] = pd.to_datetime(df['Column1'])

# Convert the second column to numeric type
df['Column2'] = df['Column2'].astype(int)

def through(arr, n, num, i, j):

    # If num is smaller than the element
    # on the left (if exists)
    if (i >= 0 and arr[i] < num):
        return False

    # If num is smaller than the element
    # on the right (if exists)
    if (j < n and arr[j] < num):
        return False
    return True

# Function that returns true if num is
# smaller than both arr[i] and arr[j]
def isTrough(arr, n, num, i, j):

    # If num is greater than the element
    # on the left (if exists)
    if (i >= 0 and arr[i] < num):
        return False

    # If num is greater than the element
    # on the right (if exists)
    if (j < n and arr[j] < num):
        return False
    return True

def printPeaksTroughs(arr, n):

    print("Peaks : ", end = "")

    # For every element
    for i in range(n):

        # If the current element is a peak
        if (through(arr, n, arr[i], i - 1, i + 1)):
            # print(arr[i], end = " ")
            peaks_info = np.vstack((arr[i],arr2[i])).T
            print(peaks_info)
    print()

    print("Troughs : ", end = "")

    # For every element
    for i in range(n):

        # If the current element is a trough
        if (isTrough(arr, n, arr[i], i - 1, i + 1)):
            print(arr[i], end = " ")

# Driver code
arr = df['Column2']
arr2=df['Column1']

# arr = [5, 10, 5, 7, 4, 3, 5]
# arr2 = [1,2,3,4,5,6,7]
n = len(arr)

printPeaksTroughs(arr, n)

这是我得到的结果

[[87 Timestamp('2023-03-14 14:20:08')]]
[[86 Timestamp('2023-03-14 14:22:23')]]        
[[86 Timestamp('2023-03-14 14:23:30')]]        
[[86 Timestamp('2023-03-14 14:24:38')]]        
[[262 Timestamp('2023-03-14 14:34:46')]]       
[[262 Timestamp('2023-03-14 14:35:54')]]       
[[91 Timestamp('2023-03-14 14:56:09')]]        
[[262 Timestamp('2023-03-14 15:07:25')]]       
[[262 Timestamp('2023-03-14 15:08:32')]]       
[[262 Timestamp('2023-03-14 15:09:40')]]       
[[89 Timestamp('2023-03-14 15:31:03')]]        
[[86 Timestamp('2023-03-14 15:35:33')]]
[[86 Timestamp('2023-03-14 15:36:41')]]
[[86 Timestamp('2023-03-14 15:37:49')]]
[[262 Timestamp('2023-03-14 15:49:04')]]
[[95 Timestamp('2023-03-14 16:07:05')]]
[[262 Timestamp('2023-03-14 16:17:13')]]

正如你所看到的，它有时会拾取值262，这是数据集中的最高值，同一算法上的数据图也会检测到这些峰值

此处绿色箭头表示直通，红色箭头表示峰
我想从第一至第二的数据通过被设置为第一波然后通过在第一波结束被视为第二波的开始例如

这是书面形式的数据，因为我不能上传csv文件。这些只是第一个几个峰值

Column1,Column2
2023-03-14 14:00:59.0,195.80
2023-03-14 14:02:06.0,174.20
2023-03-14 14:03:14.0,156.76
2023-03-14 14:04:21.0,142.36
2023-03-14 14:05:29.0,131.00
2023-03-14 14:06:37.0,122.00
2023-03-14 14:07:44.0,114.91
2023-03-14 14:08:52.0,109.18
2023-03-14 14:10:00.0,104.56
2023-03-14 14:11:07.0,100.74
2023-03-14 14:12:15.0,97.93
2023-03-14 14:13:22.0,95.45
2023-03-14 14:14:30.0,93.43
2023-03-14 14:15:37.0,91.85
2023-03-14 14:16:45.0,90.73
2023-03-14 14:17:53.0,89.49
2023-03-14 14:19:00.0,88.59
2023-03-14 14:20:08.0,87.91
2023-03-14 14:21:15.0,87.13
2023-03-14 14:22:23.0,86.68
2023-03-14 14:23:30.0,86.23
2023-03-14 14:24:38.0,86.23
2023-03-14 14:25:45.0,108.61
2023-03-14 14:26:53.0,142.70
2023-03-14 14:28:01.0,175.89
2023-03-14 14:29:08.0,203.79
2023-03-14 14:30:16.0,225.84
2023-03-14 14:31:23.0,241.25
2023-03-14 14:32:31.0,253.29
2023-03-14 14:33:39.0,262.18
2023-03-14 14:34:46.0,262.29
2023-03-14 14:35:54.0,262.29
2023-03-14 14:37:01.0,262.29
2023-03-14 14:38:09.0,260.83
2023-03-14 14:39:16.0,235.51
2023-03-14 14:40:24.0,208.85
2023-03-14 14:41:31.0,185.45
2023-03-14 14:42:39.0,166.33
2023-03-14 14:43:46.0,150.35
2023-03-14 14:44:54.0,137.41
2023-03-14 14:46:01.0,127.06
2023-03-14 14:47:09.0,118.96
2023-03-14 14:48:17.0,112.55
2023-03-14 14:49:24.0,107.15
2023-03-14 14:50:32.0,103.10
2023-03-14 14:51:39.0,99.61
2023-03-14 14:52:47.0,96.80
2023-03-14 14:53:54.0,94.55
2023-03-14 14:55:02.0,92.75
2023-03-14 14:56:09.0,91.18
2023-03-14 14:57:17.0,97.70
2023-03-14 14:58:24.0,127.06
2023-03-14 14:59:32.0,161.04
2023-03-14 15:00:39.0,190.85
2023-03-14 15:01:47.0,214.81
2023-03-14 15:02:55.0,233.38
2023-03-14 15:04:02.0,247.21
2023-03-14 15:05:10.0,256.66
2023-03-14 15:06:17.0,262.29
2023-03-14 15:07:25.0,262.29
2023-03-14 15:08:32.0,262.29
2023-03-14 15:09:40.0,262.29
2023-03-14 15:10:47.0,262.29
2023-03-14 15:11:55.0,246.31
2023-03-14 15:13:02.0,219.65
2023-03-14 15:14:10.0,194.56
2023-03-14 15:15:17.0,173.53
2023-03-14 15:16:25.0,156.43
2023-03-14 15:17:33.0,142.03
2023-03-14 15:18:40.0,130.78
2023-03-14 15:19:48.0,121.89
2023-03-14 15:20:55.0,114.80
2023-03-14 15:22:03.0,109.18
2023-03-14 15:23:10.0,104.68
2023-03-14 15:24:18.0,101.19
2023-03-14 15:25:25.0,98.26
2023-03-14 15:26:33.0,95.90
2023-03-14 15:27:41.0,93.88
2023-03-14 15:28:48.0,92.41
2023-03-14 15:29:56.0,91.06
2023-03-14 15:31:03.0,89.94
2023-03-14 15:32:11.0,89.04
2023-03-14 15:33:18.0,88.03
2023-03-14 15:34:26.0,87.35
2023-03-14 15:35:33.0,86.79
2023-03-14 15:36:41.0,86.34
2023-03-14 15:37:49.0,86.34
2023-03-14 15:38:56.0,108.39
2023-03-14 15:40:04.0,142.59
2023-03-14 15:41:11.0,175.33
2023-03-14 15:42:19.0,203.00
2023-03-14 15:43:26.0,224.94
2023-03-14 15:44:34.0,240.91
2023-03-14 15:45:41.0,252.39
2023-03-14 15:46:49.0,260.71
2023-03-14 15:47:56.0,262.29
2023-03-14 15:49:04.0,262.29
2023-03-14 15:50:11.0,262.29
2023-03-14 15:51:19.0,259.14
2023-03-14 15:52:26.0,233.60
2023-03-14 15:53:34.0,207.39
2023-03-14 15:54:41.0,183.99
2023-03-14 15:55:49.0,164.98
2023-03-14 15:56:57.0,149.00
2023-03-14 15:58:04.0,136.06
2023-03-14 15:59:12.0,125.94
2023-03-14 16:00:19.0,117.84
2023-03-14 16:01:27.0,111.43
2023-03-14 16:02:35.0,106.25
2023-03-14 16:03:42.0,102.31
2023-03-14 16:04:50.0,98.94
2023-03-14 16:05:57.0,96.35
2023-03-14 16:07:05.0,95.34

如果数据是平稳的（波峰和波谷的大小相差不大）。例如，对于trough，我们检查left上的三个值是否大于当前值，以及right上的三个值是否大于或等于。我使用List comprehension而不是循环，因为它快了很多倍。写下索引并将它们替换到loc中以获取值。
这是考虑到如果你的索引有一个整数值（0，1，2…）

import pandas as pd
import matplotlib.pyplot as plt

df['Column1'] = pd.to_datetime(df['Column1'])

period = 3

dn = [i for i in range(period, len(df) - period - 1) if
      (df.loc[i, 'Column2'] < df.loc[i - period:i - 1, 'Column2']).all() == True
      and (df.loc[i, 'Column2'] <= df.loc[i + 1:i + period, 'Column2']).all() == True]

up = [i for i in range(period, len(df) - period - 1) if
      (df.loc[i, 'Column2'] > df.loc[i - period:i - 1, 'Column2']).all() == True
      and (df.loc[i, 'Column2'] >= df.loc[i + 1:i + period, 'Column2']).all() == True]

fig, ax = plt.subplots()
ax.plot(df['Column1'], df['Column2'])
ax.plot(df.loc[dn, 'Column1'], df.loc[dn, 'Column2'], 'o', color='green', markersize=5)
ax.plot(df.loc[up, 'Column1'], df.loc[up, 'Column2'], 'o', color='red', markersize=5)
fig.autofmt_xdate()
plt.show()

并获得波谷之间的差值：

df.loc[dn, 'Column1'].diff()

如果你想从值开始下降时开始计算峰值。你需要在左边取大于或等于的条件，在右边取更多：

up = [i for i in range(period, len(df) - period - 1) if
      (df.loc[i, 'Column2'] >= df.loc[i - period:i - 1, 'Column2']).all() == True
      and (df.loc[i, 'Column2'] > df.loc[i + 1:i + period, 'Column2']).all() == True]

更新11.04.2023

dn = [i for i in range(period, len(df) - period - 1) if
      (df.loc[i, 'Column2'] <= df.loc[i - period:i - 1, 'Column2']).all() == True
      and (df.loc[i, 'Column2'] < df.loc[i + 1:i + period, 'Column2']).all() == True]

arr = df.loc[dn, ['Column2', 'Column1']].values
print(arr)

输出

[[86.23 Timestamp('2023-03-14 14:24:38')]
 [91.18 Timestamp('2023-03-14 14:56:09')]
 [86.34 Timestamp('2023-03-14 15:37:49')]]

numpy 波谷检测算法返回波峰数据

1条答案

相关问题

热门标签

最新问答