Solar-powered water pumping offers a sustainable water-supply option for rural communities with unreliable grid electricity, but conventional controllers react only to instantaneous tank-level conditions and cannot anticipate the mismatch between variable solar energy and changing water demand. This paper presents the design, implementation, and evaluation of an artificial intelligence (AI)-enabled solarpowered smart water pumping system with predictive energy and water-demand management. The prototype integrates a photovoltaic (PV) array with a maximum power point tracking (MPPT) pump controller, a submersible pump, water-level, flow, voltage and current sensors, an ESP32-based control unit, and GSM/IoT communication for remote monitoring and fault notification. A hybrid Convolutional Neural Network–Long Short-Term Memory (CNN-LSTM) model forecasts solar-energy availability and water demand over a 24-hour horizon, and the forecasts are used for predictive pump scheduling, while dry-run, level and voltage protection remain rule-based. Tests carried out in Kaduna State, Nigeria, showed sensor errors of at most 3.0%, communication success rates of 90.0–100%, and a peak PV output of about 771 W. The model achieved R² values of 0.91 for solar-energy prediction and 0.88 for waterdemand prediction. Compared with rule-based control, predictive control raised average water production from 3.42 to 3.92 m³/day (+14.6%), useful energy utilisation from 74.7% to 86.9%, and tank availability from 84.7% to 92.3%. The results demonstrate the feasibility of forecast-informed pumping in a unified, low-cost platform.

Abubakar Mohammed Baba, Ahmad Muhammad Bello, Zubairu Ibrahim, Lukman Salihu, Eeman Sani Abdu, "AI-Enabled Solar-Powered Smart Water Pumping System with CNN-LSTM-Based Predictive Energy and Water-Demand Management for Rural Communities" International Journal of Scientific Research and Engineering Development, V9(5): Page(1509-1515) September - October 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.