Predicting pollution before it spreads
A final-year research project bridging machine learning, environmental science and smart-city infrastructure.
The Problem
Over 70% of India's surface water is contaminated. Traditional lab testing is slow, expensive and reactive — pollution is detected only after damage is done.
Our Approach
We combine IoT sensors, machine learning and explainable AI to predict contamination before it spreads.
The Outcome
Authorities and citizens get real-time risk scores, alerts and actionable insight — enabling proactive intervention.
7 critical water quality parameters
Every prediction is grounded in BIS IS 10500:2012 drinking water standards.
pH
Safe: 6.5 – 8.5
Dissolved Oxygen
Safe: > 5 mg/L
Turbidity
Safe: < 5 NTU
TDS
Safe: < 500 mg/L
Temperature
Safe: 15 – 30 °C
Conductivity
Safe: < 1500 µS/cm
BOD
Safe: < 3 mg/L
Case Study: Mandi Gobindgarh
India's "Steel Town" in Punjab — home to over 200 steel rolling mills and foundries discharging contaminants into surrounding water bodies. AquaAI was trained on data from 12 monitoring locations across the region.
- Industrial discharge from steel mills
- Heavy metal contamination
- Declining groundwater quality
- Need for real-time alerts