Intelligent Predictive Platform for Fire Hazard Detection in Photovoltaic Systems Using Hybrid AI-Based Models, Sensor Technologies, and Anomaly Analysis

Fabryka Bezpieczeństwa FABE Ltd. is implementing the project entitled "Intelligent Predictive Platform for Fire Hazard Detection in Photovoltaic Systems Using Hybrid AI-Based Models, Sensor Technologies, and Anomaly Analysis" (Grant Agreement No. FENG.01.01-IP.01-A19E/25), co-financed by the European Funds for a Modern Economy Programme 2021–2027.

The project is carried out in a consortium with Lodz University of Technology.

Project Tasks

1. Temperature characterization of photovoltaic system components and defects under varying environmental conditions and operating points (industrial research – Lodz University of Technology);
2. Development of physics-constrained and AI-based hybrid models for predictive diagnostics of PV+BESS systems using multimodal fusion of temperature, electrical, and electromagnetic signals: methodology development and proof-of-concept validation (industrial research – FABE);
3. Investigation of fiber-optic temperature sensors for photovoltaic installations (industrial research – Lodz University of Technology);
4. Development and proof-of-concept validation of a predictive platform model in Software-in-the-Loop (SiL) and Hardware-in-the-Loop (HiL) environments (industrial research – FABE);
5. Demonstration of a fiber-optic temperature monitoring system under operational conditions (experimental development – Lodz University of Technology);
6. Development of an AI-based predictive platform prototype in a relevant environment (Technology Readiness Level 6), integrating FBG (Fiber Bragg Grating) and DTS (Distributed Temperature Sensing) temperature measurement systems (experimental development – FABE).

Project Description

The rapid growth in the number of photovoltaic installations and battery energy storage systems (BESS) in Poland and worldwide has significantly increased the risk of fires caused by component degradation, electrical arcing, and overheating. Existing protection systems are predominantly reactive and do not provide reliable predictive capabilities. Consequently, the lack of effective early-warning tools results in property losses, increased risks to users, and higher maintenance costs.

To address this challenge and substantially improve the operational safety of photovoltaic installations, there is a need for an advanced system capable of detecting anomalies at a very early stage—well before they escalate into fire hazards—and predicting critical events.

The project will deliver the development and demonstration of an early-warning platform for photovoltaic systems, including PV installations integrated with battery energy storage systems. The platform will combine multimodal sensing—including fiber-optic FBG and DTS temperature sensors, electrical measurements, and electromagnetic field monitoring—with hybrid artificial intelligence models incorporating physics-based constraints.

The primary objective of the project is to develop a prototype achieving Technology Readiness Level (TRL) 6, making it suitable for pilot deployment in a relevant operational environment.

As part of the project, predictive models will be developed and the necessary laboratory equipment will be acquired to support research and validation activities.

The resulting platform will be intended for operators of photovoltaic farms and battery energy storage systems in Poland and international markets.

Total project cost: PLN 6,788,568.05

Total funding: PLN 5,748,513.14