SolarRECLAIM
Multimodal Recoverable-Loss Attribution Platform for Photovoltaic Plant Performance Management
European PV operators lose an estimated €3,000–6,000 per MWp per year to avoidable performance losses from soiling, module degradation, hot spots, micro-cracks, PID and other faults. Today, drone-based RGB/IR inspections provide occasional panel-level defect reports, while SCADA/EMS systems provide continuous electrical performance data at string or inverter level; these workflows are not linked.
As a result, operators cannot reliably determine which visual anomalies are actually causing recoverable energy loss, how much value could be recovered through cleaning or repair, or which intervention should be prioritised. This creates a growing market need as EU PV capacity expands and more plants enter the 5–10 year degradation phase where performance losses and warranty-related decisions become more critical.
SolarRECLAIM will combine visual time-series analysis, multimodal data fusion and economic loss attribution in a single SaaS platform. Machine-learning models will detect and track soiling, hot spots, micro-cracks, PID and delamination in repeated RGB and IR imagery, while correcting for seasonal and environmental variation.
The core innovation is a fusion engine that links image-derived anomaly features with EMS/SCADA time-series data, irradiance, module temperature and environmental signals against an IEC 61724-1 performance baseline. A multi-task AI model will classify defects and estimate the recoverable loss in kWh and €/MW, distinguishing physically recoverable losses from general underperformance.
The platform will convert these estimates into ranked work orders by comparing expected recovered value with cleaning, repair and mobilisation costs. It will support fixed sensors, drones or hybrid data acquisition, and will use federated learning, secure aggregation, differential privacy, explainability and audit logging to enable EU-sovereign, GDPR- and AI Act-ready deployment across multiple operators and sites.
The consortium will deploy and configure hybrid inspection infrastructure at pilot PV plants in Slovenia and Croatia, combining fixed RGB/thermal cameras, drone-based validation where needed, EMS/SCADA data, irradiance, temperature and environmental measurements. Enertec will provide access to operational PV assets, EMS data streams and O&M expertise, while AlphaWave will lead AI, software, data architecture and SaaS platform development.
The partners will build a labelled multimodal dataset linking repeated RGB/IR panel observations with time-matched operational data across several inspection cycles, sites, seasons and PV technologies. They will develop and train visual time-series models, multimodal fusion models, recoverable-loss regression models, confidence calibration, explainability functions and maintenance prioritisation logic.
Pilot validation will benchmark SolarRECLAIM against drone-inspection-only and SCADA-only workflows, with targets including recoverable-loss attribution within ±15% of measured post-intervention yield recovery, per-class defect detection of at least 90% where data quality permits, and recoverable-loss identification of at least 70%. The project will also deliver the federated learning and EU-compliance framework, joint IP arrangements, SaaS pricing model and go-to-market plan for Slovenia, Croatia and later EU expansion.
.png)
