Intelligent data analytics to optimise energy utilisation in industrial processing
Lead Research Organisation:
University of Nottingham
Department Name: Faculty of Engineering
Abstract
This Resilient Decarbonised Fuel Energy Systems PhD project will investigate the use of sensor measurements and machine learning to predict and optimise either the energy production or energy utilisation for a range of industrial relevant case studies. The PhD project will be sponsored by the industrial partner Intelligent Plant, who focus on the analysis and visualisation of industrial data. Although the project will focus on different case studies. The aim is to determine suitable data collection and analysis strategies that can be applied to a variety of different industrial systems. The first and second case studies will focus on energy production systems and the third and fourth on the energy efficiency of industrial processes. An objective of this project is to develop appropriate data analysis and visualisation methods, which contribute to the UK's, net zero ambition
People |
ORCID iD |
| Nicholas Watson (Primary Supervisor) | |
| Khivishta Boodhoo (Student) |
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/S022996/1 | 30/09/2019 | 30/03/2028 | |||
| 2284079 | Studentship | EP/S022996/1 | 30/09/2019 | 29/09/2023 | Khivishta Boodhoo |
| Description | This work helps the maintenance team to perform maintenance on offshore wind turbines without disrupting the power output of the wind turbine when doing so. By using artificial intelligence as a tool to develop a model capable of predicting power outputs, wind turbines can be turned off when the power output is minimum during a certain period for maintenance. This helps optimisation of the power output of offshore wind turbines. Future case studies will be conducted to use the same framework to optimise their energy usage or production. |
| Exploitation Route | The framework being developed to apply to different case studies in systems that produce or use energy can be used in different scenarios to optimise the energy being used or produced. This framework would help in saving and optimising energy in multiple scenarios. |
| Sectors | Energy,Environment |
| Description | Intelligent data analytics to optimise energy utilisation in industrial processing |
| Amount | £75,036 (GBP) |
| Funding ID | 2284079 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 09/2019 |
| End | 09/2023 |
| Title | MLP |
| Description | This deep learning neural network model was developed and deployed to make power predictions for offshore wind turbines. Cross-validation, grid search and hyperparameter tuning were used to produce the best model. |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2022 |
| Provided To Others? | No |
| Impact | The deep learning model developed can outperform others on a large number of datasets. This can be successfully used whenever very large amount of data is available. |
| Title | SCADA and MET dataset |
| Description | This dataset is obtained from the SCADA of the ORE catapult wind turbine and consists of 30 million data points. This dataset was further joined with the MET dataset from the weather office containing around 27 million data points. These datasets were also available in an already compressed format on the Company Intelligent Plant servers who is also sponsoring this project. The framework developed to analyse such a large amount of data and relevant steps to preprocess the datasets which were prone to sensor errors, shutdowns and wrong recordings have been applied starting from loading such a massive dataset, visualisation of potential problems in the dataset, replacing missing values, removing data points with brake conditions, abnormal pitch angle and negative values of power. The next step in the framework was to identify the outliers using a specific method of clustering and afterwards removing an excessive number of repetitive values. |
| Type Of Material | Data analysis technique |
| Year Produced | 2022 |
| Provided To Others? | No |
| Impact | The framework being developed could be applied to any SCADA datasets to clean and preprocess the data before any modelling is undertaken for offshore wind turbines around the world. |
| Title | An application to predict low power output for wind turbines |
| Description | This application is still being developed and will be available on the Intelligent Plant industrial app store. The app will inform the maintenance team of wind turbines when is the most suitable time to go and perform maintenance on the wind turbine without disrupting the power output of the wind turbine. This is being done by predicting low power outputs in four hours blocks to allow the maintenance team enough time to do maintenance when the power output of the wind turbine is predicted to be low. This will optimise the overall wind turbine performance. |
| Type Of Technology | Webtool/Application |
| Year Produced | 2022 |
| Impact | The app will be sold to operators of wind turbines to plan maintenance for offshore wind turbines. |
| Description | Presentation |
| Form Of Engagement Activity | A formal working group, expert panel or dialogue |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Industry/Business |
| Results and Impact | The PowerPoint presentation was a summarised version of the research methodology and results that I had obtained during the year. The audience was composed of different individuals working at Intelligent Plant (CEO, graphic designers, engineers, marketing teams etc...). This presentation enabled the individuals to understand and develop a recorded presentation with subtitles from the live presentation and use the latter to market the app being developed at Intelligent Plant. |
| Year(s) Of Engagement Activity | 2020 |