Researchers at the National Institute of Technology (NIT) Rourkela have developed an Artificial Intelligence (AI)-powered autonomous system that can monitor, detect faults and recommend cleaning of solar panels with minimal human intervention, a breakthrough that could significantly improve the efficiency of India’s rapidly expanding solar power infrastructure.
The research team has secured an Indian patent for the innovation, titled ‘Federated Learning based Autonomous System and Method for Monitoring and Cleaning Solar Plant’, which combines AI, federated learning, edge computing and predictive maintenance into a single integrated platform.
The innovation has been developed by Prof. Arun Kumar, Assistant Professor, in collaboration with Prof. Bibhudatta Sahoo, Professor, and research graduates Dr. Lopamudra Hota and Dr. Biraja Prasad Nayak from the Department of Computer Science and Engineering at NIT Rourkela.
Addressing a critical challenge
India’s rapidly growing solar energy sector faces a persistent operational challenge as dust, bird droppings, industrial pollutants and other debris accumulate on solar panels, reducing electricity generation by as much as 40%, particularly in arid and dusty regions.
Conventional cleaning methods are labour-intensive, consume large quantities of water and are generally carried out on fixed schedules rather than based on the actual condition of the panels.
The NIT Rourkela team aims to overcome these limitations through an AI-driven system that continuously monitors panel performance, detects faults and recommends cleaning only when required, improving operational efficiency while conserving water.
Privacy-first AI architecture
A distinguishing feature of the patented technology is its use of Federated Learning (FL), an AI approach that enables multiple systems to learn collaboratively without sharing raw operational data.
Unlike conventional AI platforms that transmit sensitive data to a central server, the NIT Rourkela system exchanges only encrypted model parameters, enhancing data privacy while reducing bandwidth requirements and cybersecurity risks. The architecture also improves scalability for deployment across large solar installations.
The technology has so far been validated through simulations and has reached Technology Readiness Level (TRL)-3, demonstrating proof of concept under controlled experimental conditions.
Lower costs, smarter operations
According to the research team, the integrated platform offers real-time edge intelligence, autonomous fault detection, selective need-based cleaning, predictive maintenance and reduced water consumption, resulting in lower maintenance costs.
NIT Rourkela Assistant Professor Prof Arun Kumar said the patented system combines federated learning, edge computing, artificial intelligence, autonomous cleaning and predictive maintenance into a unified platform.
“Notable features, including real-time edge intelligence, autonomous fault detection, selective need-based cleaning, reduced water consumption and lower maintenance costs, make it a one-of-its-kind system,” he said.
Targeting large-scale deployment
The researchers believe the technology can be deployed across utility-scale solar parks, floating solar farms, rooftop photovoltaic systems, industrial solar installations, smart city energy infrastructure, defence facilities and remote off-grid renewable energy projects.
NIT Rourkela Professor Prof Bibhudatta Sahoo said existing market solutions remain expensive and offer limited intelligence.
“Our developed system integrates advanced AI capabilities for autonomous operation. Once scaled for field implementation, the technology is expected to deliver superior performance and features at approximately 10% of the cost of existing systems,” he said.
Roadmap for commercialization
The next phase of development will focus on building a hardware prototype integrated with an Internet of Things (IoT)-based sensing infrastructure and validating the technology through pilot deployments.
The research team is also looking to collaborate with government agencies and industry partners for field implementation and technology transfer. Future upgrades are expected to include drone-assisted inspections, multi-agent collaborative cleaning and predictive energy yield forecasting.
The innovation aligns with the Government of India’s National Solar Mission and the country’s Net Zero goals by enabling intelligent, autonomous and privacy-preserving maintenance of solar energy assets, potentially improving the productivity and sustainability of solar power plants in India and global markets.