Statistical Forecasting and Battery Energy Storage System (BESS) Optimization

This repository presents a comprehensive project that integrates statistical forecasting of energy prices with an optimization model for Battery Energy Storage Systems (BESS). The aim is to demonstrate how data-driven forecasting techniques and decision algorithms can be combined to enhance energy market operations and maximize profitability from storage assets.

Note: The dataset used in this analysis is confidential and cannot be shared publicly. Therefore, this repository contains only the code and methodology, not the original data files.

---

Project Overview

Energy markets are increasingly influenced by the rise of renewables, storage, and dynamic pricing mechanisms. This project demonstrates a simplified yet powerful workflow for:

1. Forecasting short-term energy prices using statistical learning.
2. Optimizing BESS operation based on predicted prices to maximize economic returns.

The workflow combines machine learning, time-series analysis, and operational strategy modeling, making it an excellent example for analysts and researchers working in the energy domain.

---

Part 1: Energy Price Forecasting

The first module focuses on building a predictive model for energy prices using historical market data.

Methodology

• Data Processing: Cleans and prepares historical price and demand data.
• Feature Engineering: Creates lagged price features and integrates renewable generation data.
• Modeling Technique: Uses a Random Forest Regressor to predict next-period prices.
• Performance Visualization: Compares predicted and actual prices over time.

Example Workflow

from sklearn.ensemble import RandomForestRegressor
from sklearn.model_selection import train_test_split
import pandas as pd
import matplotlib.pyplot as plt

1. Load and preprocess historical market data.
2. Create lagged and derived features.
3. Train and test a forecasting model.
4. Visualize predicted vs. actual price trends.

The results help determine when market prices are expected to rise or fall — a crucial input for storage operation strategies.

---

Part 2: Battery Energy Storage System (BESS) Optimization

The second module develops a simplified optimization strategy for BESS operation based on the forecasted prices.

Objective

To maximize profit by charging the battery when prices are low and discharging when prices are high.

Decision Logic

• Charge when price < 90% of average price.
• Discharge when price > 110% of average price.
• Hold otherwise.

Implementation

profit, decisions = optimize_bess(hourly_prices, capacity, initial_charge_level, charge_eff, discharge_eff)

Parameters

Parameter	Description	Example
hourly_prices	List of 24-hour price data	[100, 95, 90, ...]
capacity	Battery capacity (kWh)	100
charge_efficiency	Energy retained during charging	0.95
discharge_efficiency	Energy retained during discharging	0.95

Output

• Total Profit: Numerical value representing cumulative gains.
• Hourly Decisions: “Charge”, “Discharge”, or “Hold” for each hour.

---

Combined Use Case

The forecasted prices from the statistical model can be fed into the BESS optimization function to evaluate profit potential based on predicted market trends — demonstrating the synergy between predictive analytics and operational decision-making.

---

Technologies Used

• Python 3.10+
• Pandas – Data manipulation
• Scikit-learn – Forecasting model (Random Forest)
• Matplotlib – Visualization
• Statistics – Mean-based decision rules

---

How to Run

1. Clone the Repository

   git clone https://github.com/yourusername/Statistical-Forecasting-and-BESS.git
   cd Statistical-Forecasting-and-BESS

2. Install Required Packages

   pip install -r requirements.txt

3. Run the Notebook

   jupyter notebook "Statistical Forecasting and BESS.ipynb"

---

Example Visualization

The notebook produces time-series plots comparing:

• Actual vs. Predicted Energy Prices
• Optimal Charge/Discharge Schedule for BESS

---

Future Enhancements

• Integrate LSTM or ARIMA models for more accurate forecasting.
• Introduce market constraints and degradation costs in BESS optimization.
• Expand to multi-day and multi-market operations.
• Deploy a dashboard interface for interactive forecasting and simulation.

---