image: Solar plant (example picture) view more
Credit: Forschungszentrum Jülich/Sascha Kreklau
Using artificial intelligence (AI) to fully exploit the potential of solar plants is the main aim of the collaborative project Dig4morE, a cooperation between the Helmholtz Institute Erlangen-Nürnberg and photovoltaic companies SunSniffer, Aquila Capital, and Sunset Energietechnik. The project partners aim to develop a method that uses AI to identify measures of optimizing the plants in a quick and cost-effective manner. The process only requires monitoring data, which are generated during everyday operation. The German Federal Ministry for Economic Affairs and Energy (BMWi) will provide the project with funding worth more than € 2 million over a period of three years.
Using machine learning, researchers in the Dig4morE project aim to identify performance deficits or defects at an early stage. This is to be made possible by a new process that enables performance deficits to be read out directly in situ from the monitoring data of the individual modules. To develop the algorithms, Sunsniffer, Aquila Capital, and Sunset Energietechnik provide data from eleven of their solar parks, which are located throughout Europe.
The extensive analyses across the entire continent take account of the different operating conditions in the relevant climate zones. Various problems arise for the solar modules depending on the solar plant type and its environment. “In Hesse, which is situated in central Germany, different factors are at play compared to the west coast of Portugal, where strong winds cause the modules to vibrate,” explains Dr. Claudia Buerhop-Lutz from the Helmholtz Institute Erlangen-Nürnberg, an institute of Forschungszentrum Jülich. “The algorithms need to be trained in such a way that the various deficits can be distinguished on the basis of fundamental data such as electricity, voltage, and temperature.”
Towards the end of this year, the first results are set to be available and will be used to derive best practice examples and recommendations for action. Solar plant operators will then be able to use these to identify deficits and defects at an early stage, for example to budget for maintenance work such as cleaning measures.
An earlier study conducted by the Helmholtz Institute Erlangen-Nürnberg showed just how great the potential for optimization is. It revealed that around 8 % of European solar modules are not running at full capacity. “Alongside incorrectly set or defective modules, environmental influences such as dust, pollen, bird droppings, and tall-growing trees and grasses can lead to the plants supplying less electricity than they are capable of,” says Dr. Buerhop-Lutz.
Using state-of-the-art measuring technology, it is already possible to detect modules that are either defective or not being fully utilized by means of thermographic analyses, for example. However, this procedure is expensive and complex. Large-scale solar parks are typically analysed with the aid of drones in the sky. The use of AI measuring instruments, as proposed by the Dig4more project, aims to enable more cost-effective and comprehensive analyses.
Prof. Brabec, head of the High Throughput Methods in Photovoltaics department, stresses: “We regard the application of high-throughput measuring methods as being key to the sustainable operation of solar parks. The best possible yields and lifespans for solar fields can only be ensured through a combination of measurement technology, which can be used to quickly characterize large amounts of solar modules, and artificial intelligence.”