Krismadinata, Leo Raju, Pandikumar Maniraj
This article provides a detailed comparison of the economics and technology of two different types of Solar Photovoltaic Water Pump Systems (SPVWPS); one type is a traditional SPVWPS and the other is an AI based Predictive Solar Pump System for banana production in the semi-arid regions of Theni district of Tamil Nadu, India. In this study the new AI system is made up of a high-fidelity sensor array (weather, soil moisture, evapotranspiration), a Cloud-Edge Computing Architecture made possible by Texas Instruments TMS320F28379D DSP for real time motor control and an STM32H743 Microprocessor for coordinating all of the components of the system. An LSTM Network Based Predictive Algorithm uses historical data on the banana crop to predict when the crop needs water, allowing for optimal scheduling of irrigation, providing water at the precise time it is needed. For the eight-month trial period of the 5 HP Prototype, results indicated that the AI system provided 33.7 percent savings in water use, and a 19.2 percent increase in banana yield (tons/hectare) over the traditional timer based SPVWPS used as a reference point for this research. Economic evaluations indicated the AI system would provide a 0.93-year payback (11.2 months) for the additional costs associated with implementing the AI system; these savings are derived from reduced water usage and increased yields. These results validate the ability of intelligent resource utilization in precision agriculture to produce positive environmental impacts. © 2026 The Authors.
Universitas Negeri Padang, Centre for Energy and Power Electronics Research (CEPER), Padang, 25131, Indonesia; Sri Sivasubramaniya Nadar College of Engineering, Department of Electrical and Electronics Engineering, Chennai, 603110, India; Saveetha School of Engineering, Department of Electrical Power and Energy Conversion, Simats, Chennai, 602105, India
