Case Study: AI for Innovation and Scalability in PV Site Management
Introduction
A startup called SOLAR SPY, which is now a thriving business, wanted to leverage AI and advanced data analytics to revolutionize processes, reduce costs, and introduce new solutions to the solar power sector, optimizing PV assets and increasing ROI.
Our Achievements:
- Developed a thermal image recognition system analyzing over 52,000 images.
- Improved the accuracy of detecting occupancy and activities within smart buildings.
- Received positive feedback for the system’s effectiveness in optimizing building management.
- Enhanced space usage monitoring with minimal privacy invasion.
Client Benefits:
- Provided metrics on occupancy rates and activity patterns.
- Enhanced energy efficiency through data-driven insights.
- Supported effective space management with minimal disruption to occupants.
The Problems:
Solar power site owners and managers face several struggles, including:
- Inconsistent results from drone inspections.
- Slow and often inaccurate analyses of thermal images.
- A lack of unified platforms for aggregating and analyzing data across all sites.
Our Solution:
The Solar Spy startup was initiated with a vision to leverage artificial intelligence (AI) and data analytics to analyze extensive image data (1500 Thermal and 1500 RGB images per 1MWp) and accurately pinpoint issues in solar power modules.
Our ambition extended beyond providing reliable and efficient assessments of solar power site health and performance; we aimed to offer a comprehensive platform for all-encompassing data management and actionable insights.
Path to Success:
- Collaborative R&D and Strategic Planning: Merging technological and PV domain expertise, our R&D team formulated a strategic plan, ensuring our solutions were precisely tailored to address the nuanced challenges within PV plants.
- Data Management and Model Definition: The R&D team meticulously crafted and managed data sets, utilizing domain knowledge to ensure model training was effectively targeted. Models were defined to solve specific issues, such as PV module detection on RGB images and issue identification on thermal images as well as module reference temperature calculation etc.
- Data Augmentation and AI Training: Utilizing advanced data augmentation methods, we built extensive data sets, enhancing our AI models’ robustness and accuracy, achieving a remarkable 99% accuracy in identifying and categorizing issues in solar power modules.
- Unified SaaS Platform Development: A comprehensive SaaS platform was developed to continuously monitor site performance and health, seamlessly integrating production data from all sites and providing a unified overview of site performance.
- Data Analytics and Proactive Machine Learning: We implemented robust data pipelines and analytics, coupled with real-time machine learning, to detect anomalies and ensure immediate issue identification and mitigation. Our ongoing developments in predictive analytics aim to further enhance proactive site management.
Results:
Solar Spy continues to grow, leveraging the robust foundation provided by Pragmile. Solar Spy addressed immediate challenges and set new industry benchmarks, ensuring that solar power site owners, managers, and O&Ms are equipped with reliable, efficient, and forward-thinking tools for optimal site management.
Conclusion:
Our success with Solar Spy shows how innovative solutions, developed with a clear understanding of industry challenges, can revolutionize standards and practices. While we at Pragmile continue to innovate and explore new horizons, Solar Spy, benefiting from its roots, is set to drive further advancements in the solar power industry, ensuring reliable, efficient, and proactive solar power site management.
Schedule a free consultation with
our AI and technology experts
Take advantage of the latest AI solutions, tailored to your company's needs. Book a consultation with AI solution architects at Pragmile and discover new opportunities in energy management.
Please, provide your business email to schedule a meeting
