This repository contains a model and data set for the project PJM Energy Load Profile
The purpose of this project is to solve load profile classification task. Load profile term is related to loads in energy grid. Profile is 24 values of energy load related to each hour of the day. Winter and summer hourly load profiles are quite different. At winter we can see morning and evening load peaks caused by electric lightning systems. At summer we can see middle-of-the-day peaks caused by air conditioners. Fall load profile may look different: sometimes it is similar to summer profile, sometimes - to winter, it dependns on weather. The task solved in this project is to build and train a neural network that can distinguish different load profiles from one another and identify abnormal ones among ordinary profiles.
Project repository contains the following directories: *data: data directory, *model: directory which contains model file, *code: directory which contains Jupiter notebook with Python code.
Data stored in SCV format. Model created in Keras with TensorFlow. Python code contains the parts for training and validation data sets preparation, model creation, model train and valudation.
Jupiter notebook contains all necessary code to install all the required software. Just run the notebook cells and see the output.