Artificial Intelligence in Renewable Energy Systems

Lead Research Organisation: University of Southampton
Department Name: Faculty of Engineering & the Environment

Abstract

Over the last two decades, sustainable energy technologies have become a critical part and a major contributor to the global energy supply mix especially in the electricity sector. This is driven by many factors including: (i) our desire to use sustainable resources to reduce pollution emanating from the current use of fossil fuels and (ii) to provide a pathway to achieve national and internationally agreed emission reductions coupled with increasing energy security through local resource utilisation. The renewable energy industry has matured, with huge investments being ploughed into it globally. Such major investments covers solar and wind energy technologies deployed in arrays and farms at the multi-MW scale. It is clearly important for conversion devices used in these arrays and farms to be operating optimally to maximise energy output and hence economic return on investments. This research aims to develop an accurate and low-cost analysis system that is capable of predicting fault progression in solar photovoltaic modules in farms (Infra-Red) and will extend the methodology and analysis to individual turbine blades in wind farms (LIDAR). The impacts of such faults will then be related to farm characteristics and outputs. The study will have in-field (solar farms) studies including various imaging techniques using drones and robotics as well as the development of AI models to investigate and provide alerts on faults, model their impacts on energy yields and related these to operations and maintenance schedules for such farms.

Key challenges to be addressed include:
(1) Establish appropriate approaches for optimally surveying and imaging of photovoltaic farms. This will included drones and ground vehicle surveys. Challenges such as the height and speed of travel required to achieve accuracy of the survey and provide data on real a hotspots.
(2) Understanding of thermographic theory and the analysis to help with interpretation of the survey results and other diagnostics needed to support accuracy.
(3) How the surveys' outcomes can be related to individual farm characteristics and other data output such as those from the inverters and sensors.
(4) Establish the best approach of AI including statistical analysis to provide a streamline analysis of imaging and the data captured through surveys and this can be linked to optimised farm outputs and operation.

This project will have access to around 100MW of solar wind farms some of which are located around Southampton. The resulting methods will be tested on real-world solar farms and will have national and global applicability applications.

People

ORCID iD

Alois Klink (Student)

Publications

10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/N509747/1 30/09/2016 29/09/2021
2106189 Studentship EP/N509747/1 30/09/2018 29/09/2021 Alois Klink
EP/R513325/1 30/09/2018 29/09/2023
2106189 Studentship EP/R513325/1 30/09/2018 29/09/2021 Alois Klink
 
Title aloisklink/flirextractor v1.0.0: An efficient GPLv3 Python package for extracting temperature data from FLIR IRT images 
Description An efficient GPLv3-licensed Python package for extracting temperature data from FLIR IRT images. DOI: doi:10.5258/SOTON/D1133 There is an existing Python package for extracting temperature values from raw IRT images, see nationaldronesau/FlirImageExtractor. However, it has some issues that I didn't like: - Most importantly, it is forked from the UNLICENSED Nervengift/read_thermal.py, so until Nervengift/read_thermal.py#4 is answered, this package cannot be legally used. - Secondly, it is quite inefficient, as it runs a new exiftool process for each image, and it converts the temperature for each pixel, instead of using numpy's vectorized math. See the README.md for more details, including how to install, use the software, and run tests. Many thanks to Glenn J. Tattersall, for their gtatters/Thermimage R package. This work uses an image and adapts part of gtatters/Thermimage under the GPLv3.0 License. This work was supported by the Engineering and Physical Sciences Research Council [Doctoral Training Partnership Grant EP/R513325/1]. 
Type Of Material Computer model/algorithm 
Year Produced 2019 
Provided To Others? Yes  
Impact Currently, as far as I'm aware, that have been no other notable impacts from this model, other than this award. However, this package has had more that 8500 downloads as of 2021-03-11, and researchers have already asked for help with using this package. 
URL https://github.com/aloisklink/flirextractor
 
Description Centre of Machine Intelligence Showcase 2019 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Industry/Business
Results and Impact I presented a poster at the Centre of Machine Intelligence Showcase 2019.

This event contained members of industry/business, students, researchers, politicians, and members of charities.
Year(s) Of Engagement Activity 2019
URL https://www.cmi.ecs.soton.ac.uk/machine-intelligence-showcase-2019
 
Description PhD presentation to Southampton Masters students 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Local
Primary Audience Postgraduate students
Results and Impact I gave a presentation to postgraduate master students in the University of Southampton about the work in this award. The purpose of this talk was to both teach them about artificial intelligence and about doing PhD research.
Year(s) Of Engagement Activity 2019,2020