Intelligent Load Forecasting for Enhanced Energy Management in Smart Grids

Table of Contents

1. Overview
2. Objectives
3. Key Features
4. How it Works
5. Installation and Setup
6. How to Use
7. Dependencies
8. File Structure
9. Hosted Version

Overview

• This project is centered around load forecasting for electricity demand in a power grid.
• The data used for this project was sourced from the Delhi State Load Dispatch Centre.

Objective

• The main objective was to implement an LSTM model to generate short-term load forecasts, predicting demand 24 hours into the future.

Key Features

• Short-term load forecasting using LSTM model
• Data visualization using Streamlit
• High accuracy with an RSME (Root Mean Square Error) less than 1%

How it Works

• The LSTM model is trained using the dataset scraped from the Delhi State Load Dispatch Centre (https://www.delhisldc.org/Loaddata.aspx?mode=28/05/2024)
• The model learns to predict the electricity demand for the next 24 hours based on the previous 10 days.
• The predictions are then visualized using the Streamlit app.

Installation and Setup

1. Clone the repository:

   git clone https://github.com/CubeStar1/LoadPredictor.git
   cd LoadPredictor

2. Create a virtual environment:

   python -m venv venv

3. Activate the virtual environment:

   • On Windows:

     .\venv\Scripts\activate

   • On Unix or MacOS:

     source venv/bin/activate

4. Install dependencies:

   pip install -r requirements.txt
   

5. Run the Streamlit app:

    streamlit run app.py

6. Use the Jupyter Notebook to train the model and perform data analysis.

How to Use

• After setting up the project, you can use the Streamlit app to visualize the electricity demand and the model's predictions.
• You can also use the Jupyter Notebook to train the model and perform data analysis.

Dependencies

• Python
• Pandas
• NumPy
• Matplotlib
• Scikit-learn
• TensorFlow
• Keras
• Streamlit
• BeautifulSoup
• Jupyter Notebook

File Structure

• app.py: The Streamlit app for data visualization
• utilities/datasets/: The dataset used for training the LSTM model
• scripts/data-scraping.py: Used to scrape data from the Delhi State Load Dispatch Centre
• utilities/jupyter-notebook/: Jupyter Notebook used for data analysis and model training

Hosted Version

• The Streamlit app is hosted on and can be accessed here: https://loadpredictor.streamlit.app/