Evaluating Neural Networks for real-time wastewater forecasting and infrastructure security.
As climate change intensifies extreme rainfall, traditional sewer models are becoming too rigid and costly to maintain. This project introduces a systematic evaluation protocol for Machine Learning in Combined Sewer Systems (CSS). We benchmark Neural Network architectures not just on accuracy, but on their ability to handle peak events and survive network outages. By comparing "Global" powerhouse models with "Local" IoT-ready alternatives, we provide a blueprint for reliable, decentralized, and cost-efficient urban water management.
Access paper