工业设备的状态监测和预测性维护（Python）

摘要：# Common importsimport pandas as pdimport numpy as npfrom sklearn import preprocessingimport seaborn as snssns.set(color_codes=Tru

# Common importsimport pandas as pdimport numpy as npfrom sklearn import preprocessingimport seaborn as snssns.set(color_codes=True)import matplotlib.pyplot as plt%matplotlib inlineimport osfrom google.colab import drivedrive.mount('/content/drive')data_dir = '/content/drive/MyDrive/Machine life detection/2nd_test'merged_data = pd.DataFrame# Looping over all files from 12th Feb to 19th Febfor filename in os.listdir(data_dir):dataset=pd.read_csv(os.path.join(data_dir, filename), sep='\t')dataset_mean_abs = np.array(dataset.abs.mean)dataset_mean_abs = pd.DataFrame(dataset_mean_abs.reshape(1,4))dataset_mean_abs.index = [filename]merged_data = pd.concat([merged_data, dataset_mean_abs], ignore_index=False)# Renaming columnsmerged_data.columns = ['Bearing 1','Bearing 2','Bearing 3','Bearing 4']# Identifying index as dateTime formatmerged_data.index = pd.to_datetime(merged_data.index, format='%Y.%m.%d.%H.%M.%S')merged_data = merged_data.sort_indexmerged_data.to_csv('merged_dataset_BearingTest_2.csv')# Check - Begining of Datamerged_data.head

# Check - End of Datamerged_data.tail

#Check Total Pointsprint(f'Total Data Points {merged_data.shape[0] + 1}')Total Data Points 985# Visualising Dataax = merged_data.plot(figsize = (12,6), title="Vibration Data" , legend = True)ax.set(xlabel="Year-Month-Date", ylabel="Vibration/Acceleration(g)")plt.axvline(x='2004-02-19 06:12:39', linewidth=4, color='b', label ="Breakdown of Bearing 1")plt.text('2004-02-19 06:12:39',0.3,'Breakdown of Bearing 1',rotation=90, fontsize=14, color='b')

Zooming in Data for Bearing 1

fig = plt.figure# Divide the figure into a 1x2 grid, and give me the first sectionax1 = fig.add_subplot(121)# Divide the figure into a 1x2 grid, and give me the second sectionax2 = fig.add_subplot(122)healthy =merged_data['2004-02-12 11:02:39':'2004-02-12 23:52:39']healthy['Bearing 1'].plot(figsize = (12,6), title="Healthy State" , legend = True, ax=ax1)ax1.set(xlabel="Month-Date Time", ylabel="Vibration/Acceleration(g)")faulty =merged_data['2004-02-18 11:02:39':'2004-02-18 23:52:39']ax2 = faulty['Bearing 1'].plot(figsize = (12,6), title="Faulty State" , legend = True, ax= ax2)ax2.set(xlabel="Month-Date Time", ylabel="Vibration/Acceleration(g)")

Anamoly Detection using Prophet Package

pip install prophetfrom prophet import Prophethealthy_bearing1 = merged_data['2004-02-12 10:32:39':'2004-02-15 23:42:39']['Bearing 1']# Creating training dataframeprophet_healthy_train = pd.DataFrameprophet_healthy_train['ds'] = healthy_bearing1.indexprophet_healthy_train['y'] = healthy_bearing1.valuesprophet_healthy_train.head

m = Prophet(interval_width = 1)# Using the training data from "healthy part"m.fit(prophet_healthy_train)forecast = m.predict(prophet_healthy_train)forecast['fact'] = prophet_healthy_train['y'].reset_index(drop = True)print('Displaying Prophet plot')fig1 = m.plot(forecast)fig1 = healthy_bearing1.plot(figsize = (12,6), title="Fit of Training Data")fig1.set(xlabel="Month (MM)-Date(DD) Time", ylabel="Vibration/Acceleration(g)")

# Reduce data points (assuming prophet_faultydata is your full data)prophet_faultydata_subset = prophet_faultydata.resample('H').mean # Resample to hourly data# Prepare data for Prophet modelprophet_faulty_test = pd.DataFrameprophet_faulty_test['ds'] = prophet_faultydata_subset.indexprophet_faulty_test['y'] = prophet_faultydata_subset.values# Forecast and plot with reduced figure sizeforecast = m.predict(prophet_faulty_test)forecast['fact'] = prophet_faultydata_subset.reset_index(drop=True)print('Displaying Prophet plot')fig1 = m.plot(forecast, figsize=(4, 3)) # Reduce figure size# Plot actual faulty data on a separate figure (optional)fig2 = prophet_faultydata_subset.plot(figsize=(8, 4), title="Actual Faulty Data")fig2.set(xlabel="Month (MM)-Date(DD) Time", ylabel="Vibration/Acceleration(g)")fig2.text(12464.0, 0.06, 'Predicted',fontsize=10, color='r')fig2.text(12464.0,0.075,'Faulty Data', fontsize=10, color='r')fig2.text(12462.0,0.063,'Healthy', fontsize=10, color='r')