基于预训练BERT的时间序列异常检测（Python）

摘要：import stumpy as spimport pandas as pdimport numpy as npimport scipy.signal as signalimport matplotlib.pyplot as pltfrom ts_transf

import stumpy as spimport pandas as pdimport numpy as npimport scipy.signal as signalimport matplotlib.pyplot as pltfrom ts_transformers.models.pattern_matching import matrix_profile, plot, plot_matrix_profile, generate_origin, generate_seasonal, generate_shapelet, generate_trend

generate pattern-wise anomaly

ShapeletSeasonalTrend# original serieslength = 50000period = 3e-1lim = 1000x, y = generate_origin(length, period)plot(y)

x, y = generate_shapelet(length, period)plot(y)t = int(np.ceil(1/period * length / lim))query = y[0: t]idxs = matrix_profile(query, y, top_k=3)plot_matrix_profile(y, len(query), idxs, 3000)

x, y = generate_seasonal(length, period)plot(y)query = y[0: t]idxs = matrix_profile(query, y, top_k=3)plot_matrix_profile(y, len(query), idxs, 3000)

x, y = generate_trend(length, period)plot(y, lim = 5000)query = y[0: 3*t]idxs = matrix_profile(query, y, top_k=3)plot_matrix_profile(y, len(query), idxs, 5000)

知乎学术咨询：

https://www.zhihu.com/consult/people/792359672131756032?isMe=1

担任《Mechanical System and Signal Processing》《中国电机工程学报》等期刊审稿专家，擅长领域：信号滤波/降噪，机器学习/深度学习，时间序列预分析/预测，设备故障诊断/缺陷检测/异常检测。

分割线分割线分割线

一种一维时间序列信号变化/事件/异常检测方法(MATLAB)

完整数据，代码和参考文献可通过知乎学术咨询获得

几种基于机器学习的时间序列异常检测方法（Python，Jupyter Notebook文件）

需要安装pyod模块：

pip install pyod

Using Residual PlotCART (Isolation Forest)Forescating (Prophet)Clustering (KMeans)Autoencoders

所需模块如下：

import pandas as pdimport numpy as npimport matplotlib.pyplot as pltimport seaborn as snsfrom fbprophet import Prophetfrom time import timefrom datetime import datetimefrom datetime import timedeltafrom pandas.plotting import register_matplotlib_convertersfrom mpl_toolkits.mplot3d import Axes3Dimport statsmodelsfrom statsmodels.tsa.seasonal import seasonal_decomposefrom statsmodels.tsa.stattools import acf, pacffrom statsmodels.tsa.statespace.sarimax import SARIMAXfrom sklearn.preprocessing import StandardScalerfrom sklearn.decomposition import PCAfrom sklearn.cluster import KMeansfrom sklearn.covariance import EllipticEnvelopefrom sklearn.ensemble import IsolationForestimport warningswarnings.filterwarnings('ignore')