RobMulla / PIMachineLearning

LOAD FORECAST WITH OSISOFT PI USING PYTHON

Readme.txt

The scripts in this folder are used to: - Pull historic load from PI - Build and store machine learning models for the points - Predict and store forecast based on future weather conditions - Store results in PI under the name Tagname+.Forecast

Files:

• build_points.csv - A list of points for the BuildModels.py script to build for
• predict_points.csv - A list of points for the PredictandStore.py script to predict and store for
• BuildModels.py - Run this script to build models for the list of points in the build_points.csv file. Saves model file in the /models folder
• PredictandStore.py - Run this script to predict and store results for the points in the predict_points.csv file
• /models - Folder containing the model files
• /Code - Folder containing code for pulling PI tags, building model and storing reults
• LoadForecast.log - Logs the timestamp and results of the BuildModels.py and PredictandStore.py scripts. Use for debugging.

How to use:

• To Build New Models

  • Open the build_points.csv and add the point names you want to predict. Each point on its own line, no commas.
  • open a powershell of command prompt.
  • cd to the directory containing the scripts
  • run the script >> python BuildModels.py

• To Predict and Store Reults in PI

  • Open the predict_points.csv and add the point names you want to predict. Each point on its own line, no commas.
  • open a powershell of command prompt.
  • cd to the directory containing the scripts
  • run the script >> python PredictandStore.py

Questions? Contact Rob Mulla - [e-mail]