| Work Detail |
An international research team has created a digital twin that is said to enable analysis of different aerial monitoring scenarios for photovoltaic power plants. The new tool is said to reduce the risk associated with real-world experimentation and help identify the most effective strategies for improving monitoring of photovoltaic systems.
An international research team has created a digital twin (DT) platform to test autonomous aerial monitoring of large-scale photovoltaic power plants. DTs are virtual representations of physical systems that allow operators and researchers to evaluate scenarios without the risk associated with real-world experimentation.
“The novel digital twin-based solution called Digital-PV has been developed for the simulation and managed execution of autonomous aerial monitoring of photovoltaic power plants,” the researchers explain. “It provides a virtual test platform for autonomous flights and missions, including boundary detection, trajectory planning and fault detection, together with data generation capabilities to develop data-driven monitoring and inspection models.”
The DT was based on Unreal Engine (UE), created by the American video game company Epic Games to develop game environments. The scientists created a 4 km2 area with small terrain variations and natural elements, and adjusted atmospheric and lighting parameters to emulate real-world lighting.
“To add plant components such as photovoltaic panels and wind turbines to the level, we acquired the necessary assets and 3D models, including a wind turbine and a photovoltaic panel prototype, from 3D model marketplaces,” the group adds. “UE’s built-in static mesh editor was used to create custom meshes of bird droppings of various shapes and sizes, which were used as glitch characters.”
The company used Microsoft’s AirSim simulator to simulate an aerial robot in the simulated environment. It was set up to record RGB images, identify bird droppings, descend to the fault site, and capture an image of the likely faulty panel surface. These images can then be used to train artificial intelligence (AI) fault detection models.
“One of the challenges when using AI techniques is acquiring a significant amount of annotated data. This is because publicly available datasets are quite limited,” the academics say. “We demonstrate the potential of Digital-PV in generating datasets for developing monitoring models. For this purpose, bird droppings were used as an example, although other defects are planned to be included in the future.”
In total, the aerial robot took images of 2,469 simulated PV samples with bird droppings. It used an encoder-decoder-based fully convolutional network (FCN) architecture, adapted from the VGG16 model, for fault detection. About 85% of the images were used for model training, 10% for validation during the training process, and 5% for testing.
The scientists found that the average accuracy for training and validation was 98.31% and 97.93%, respectively, which they say shows that the trained model can accurately locate birds landing on PV modules with an average accuracy of 95.2% for the test data.
DTs can also be used to evaluate AI boundary extraction, i.e. to identify the PV plant in relation to the environment. The team also demonstrated that they can be used to test trajectory planning models. For boundary extraction, they used classical image processing (CIP) and a deep learning (DP)-based model, while for trajectory planning they used algorithms from the scientific literature. However, quantitative results of these tests were not presented.
“By evaluating their performance and effectiveness, we gain valuable insights into their capabilities and potential real-world performance,” the scientists concluded. “This analysis contributes to the advancement of AI-based solutions for aerial monitoring of photovoltaic plants, enabling more efficient inspection and maintenance of these critical renewable energy infrastructures.”
The digital twin was presented in “ Digital-PV: A digital twin-based platform for autonomous aerial monitoring of large-scale photovoltaic power plants ,” published in Energy Conversion and Management . It was proposed by scientists from Amirkabir University of Technology in Iran, the University of Isfahan, Concordia University in Canada, the Norwegian University of Science and Technology (NTNU), and Albert Ludwigs University in Freiburg, Germany. |
