oussamaftaimia/cellular-intelligence-renewable-energy

Physarum-inspired machine learning framework to design optimal renewable energy networks using Tokyo metro data for training and Namibia for testing. Combines bio-inspired intelligence with geospatial datasets. Cellular Intelligence for Renewable Energy Networks This repository implements a Physarum-inspired approach to optimize energy infrastructure layout by mimicking the behavior of Physarum polycephalum, a slime mold known for forming efficient transport networks. The system learns from real urban and energy data to simulate network growth and optimization. -Project Highlights Bio-Inspired Optimization: Based on the slime mold's ability to connect nutrient sources using minimal cost paths. Training Phase: Tokyo railway system is used as a reference (replicating Physarum’s successful reproduction of Tokyo’s network). Testing Phase: Applied to Namibia's solar potential and settlements to evaluate renewable energy layout planning. Scalable Framework: Designed to extend to any geographic location with elevation, demand, and energy resource data. -Datasets 1. tokyo-station20180330_Dataset.csv Contains coordinates and metadata for Tokyo metro stations. Used as nodes in training the graph structure. 2. tokyo_rail_edges.csv Represents edges and distances between Tokyo stations. Basis for graph topology that mimics Physarum’s pathfinding. 3. namibia_dre_atlas_settlements.csv Settlement locations, population, demand, PV potential, and proximity to infrastructure. Used to test the Physarum-inspired energy distribution model. -Potential Machine Learning Models Linear Regression: Estimate energy demand or capacity needs. K-Means Clustering: Group demand centers or source locations. Decision Trees / Random Forests: Classify optimal vs suboptimal routes. Support Vector Machines (SVM): For terrain-aware or resource classification. Reinforcement Learning (RL): Learn dynamic energy routing policies. Graph Neural Networks (GNNs): Predict edge activations or simulate network evolution. Deep Q-Learning: Advanced RL for policy optimization across regions. These models can be integrated at different stages to enhance prediction, routing, or system robustness.