Smart on-line monitoring for nuclear power plants (SMART)
Lead Research Organisation:
Leeds Beckett University
Department Name: Built Environment and Engineering
Abstract
Nuclear power has great potential as a future global power source with a small carbon footprint. To realise this potential, safety (and also the public perception of safety) is of the utmost importance, and both existing and new design nuclear power plants strive to improve safety, maintain availability and reduce the cost of operation and maintenance. Moreover, plant life extensions and power updates push the demand for the new tools for diagnosing and prognosing the health of nuclear power plants. Monitoring the status of plants by diverse means has become a norm. Current approaches for diagnosis and prognosis, which rely heavily on operator judgement on the basis of online monitoring of key variables, are not always reliable. This project will bring together three UK Universities and an Indian nuclear power plant to directly address the modelling, validation and verification changes in developing online monitoring tools for nuclear power plant.
The project will use artificial intelligence tools, where mathematical algorithms that emulate biological intelligence are used to solve difficult modelling, decision making and classification problems. This will involve optimizing the number of inputs to the models, finding the minimum data requirement for accurate prediction of possible untoward events, and designing experiments to maximize the information content of the data. We will then use the optimised system to predict potential loss of coolant accidents and pinpoint their specific locations, after which we will progress to prediction of possible radioactive release for various accident scenarios, and, in order to facilitate emergency preparedness, the post release phase will be modelled to predict the dispersion pattern for the scenarios under consideration. Finally, all of the models will be validated, verified and integrated into a tool that can be used to monitor and act as an early warning device to prevent such scenarios from occurring.
The project will use artificial intelligence tools, where mathematical algorithms that emulate biological intelligence are used to solve difficult modelling, decision making and classification problems. This will involve optimizing the number of inputs to the models, finding the minimum data requirement for accurate prediction of possible untoward events, and designing experiments to maximize the information content of the data. We will then use the optimised system to predict potential loss of coolant accidents and pinpoint their specific locations, after which we will progress to prediction of possible radioactive release for various accident scenarios, and, in order to facilitate emergency preparedness, the post release phase will be modelled to predict the dispersion pattern for the scenarios under consideration. Finally, all of the models will be validated, verified and integrated into a tool that can be used to monitor and act as an early warning device to prevent such scenarios from occurring.
Planned Impact
The nature of the research is such that the outcomes of the project will be very relevant to the nuclear energy industry. The project aims to produce an effective tool to enhance the safety of nuclear power plants. The tool, software and underlying models that will form the end-result will be suitable for incorporation into the control room of power plant, or as separate modules for use to improve the safety of nuclear power plants. Some of the models to be developed could also be of use to the UK atomic energy and environmental authorities for monitoring the safety of the power plants and for predicting radioactive releases which could adversely affect humans and/or the environment.
In addition to the tool envisaged, the advanced signal processing method for verification and validation, the online modeling technique for the prediction of plume dispersion, the hybrid methods of neural network techniques and the signal processing methods to be used to develop the tool are innovative, and will be of interest to others in this field. Firstly, the interest would be in research and development, so would be expected to be primarily within the academic community. However, there would be some longer-term impact expected, as online modelling techniques, neural networks and signal processing can be used in many diverse applications, and the innovative techniques could be applied to many areas e.g. medical instrumentation.
A further impact is that it will help reduce radioactive releases to the environment with ensuing benefits to human health by managing the accidents better in nuclear plants. The methods to ensure the impact to both academics and industry will be:
1. Communications and engagement
The results will be presented at different conferences: Conference on Decision and Control, SAFEPROCESS, American Nuclear Society Winter Meeting, ICNESE, ESREL, and RAMS conferences, all of which are of relevance to the project. Those conference will attract the major players in the field, from both the academic and industrial world, and also those with safety-related interests.
A workshop will also be hosted at the University towards the end of the grant, to disseminate results to both academic and industrial colleagues.
A project website will also be launched, where a summary of progress will be maintained, and progress reports written at the end of each phase will be available to download.
The results will also be published on the journals that enjoy a wide circulation in the nuclear industry (principally IEEE Transactions on Nuclear Sciences, Reliability Engineering and Systems). But publications will also be written for journals dedicated to the fields of modeling, and of safety. At least one publication will be aimed at an open-access journal, to ensure a wide circulation.
Additionally, the PIs will be involved in public engagement via presenting at public open days hosted by the University, schools visits, Women's Engineering Society and popular science articles (e.g. New Scientist).
2. Collaboration
EDF energy will provide some practical advice, which could improve marketability. The PIs will look for potential partners by attending the conference, searching in the internet, publicizing the results through the project website.
3. Exploitation and application
The methods and tools to be developed will have the potential to be commercially exploitable. It addresses a problem experienced by industry, and can be best put to use by incorporation into commercial products.
4. Capability
The PI will be involved in impact activities, especially as they are likely to continue after the end of the award (e.g. commercialisation, Phase 4 UK-India Nuclear collaboration). However, she will ensure that both the RA and the student have some involvement in these activities, as part of their career development. All will be involved in producing reports and ensuring the website is up to date.
In addition to the tool envisaged, the advanced signal processing method for verification and validation, the online modeling technique for the prediction of plume dispersion, the hybrid methods of neural network techniques and the signal processing methods to be used to develop the tool are innovative, and will be of interest to others in this field. Firstly, the interest would be in research and development, so would be expected to be primarily within the academic community. However, there would be some longer-term impact expected, as online modelling techniques, neural networks and signal processing can be used in many diverse applications, and the innovative techniques could be applied to many areas e.g. medical instrumentation.
A further impact is that it will help reduce radioactive releases to the environment with ensuing benefits to human health by managing the accidents better in nuclear plants. The methods to ensure the impact to both academics and industry will be:
1. Communications and engagement
The results will be presented at different conferences: Conference on Decision and Control, SAFEPROCESS, American Nuclear Society Winter Meeting, ICNESE, ESREL, and RAMS conferences, all of which are of relevance to the project. Those conference will attract the major players in the field, from both the academic and industrial world, and also those with safety-related interests.
A workshop will also be hosted at the University towards the end of the grant, to disseminate results to both academic and industrial colleagues.
A project website will also be launched, where a summary of progress will be maintained, and progress reports written at the end of each phase will be available to download.
The results will also be published on the journals that enjoy a wide circulation in the nuclear industry (principally IEEE Transactions on Nuclear Sciences, Reliability Engineering and Systems). But publications will also be written for journals dedicated to the fields of modeling, and of safety. At least one publication will be aimed at an open-access journal, to ensure a wide circulation.
Additionally, the PIs will be involved in public engagement via presenting at public open days hosted by the University, schools visits, Women's Engineering Society and popular science articles (e.g. New Scientist).
2. Collaboration
EDF energy will provide some practical advice, which could improve marketability. The PIs will look for potential partners by attending the conference, searching in the internet, publicizing the results through the project website.
3. Exploitation and application
The methods and tools to be developed will have the potential to be commercially exploitable. It addresses a problem experienced by industry, and can be best put to use by incorporation into commercial products.
4. Capability
The PI will be involved in impact activities, especially as they are likely to continue after the end of the award (e.g. commercialisation, Phase 4 UK-India Nuclear collaboration). However, she will ensure that both the RA and the student have some involvement in these activities, as part of their career development. All will be involved in producing reports and ensuring the website is up to date.
Publications
G. Colantuono
(2017)
Principal-component-based detection algorithm for fault detection
Oparaji U
(2017)
Robust artificial neural network for reliability and sensitivity analyses of complex non-linear systems.
in Neural networks : the official journal of the International Neural Network Society
| Description | Our partner Bhabha Atomic Research Centre has problems in predicting big breaks in the pipes of nuclear power plant. We have achieved so far to predict the big break size of the pipes accurately. Artificial intelligence methods have been applied to monitor the safety of nuclear power plants (NPPs). One major safety issue of an NPP is the loss of a coolant accident (LOCA) which is caused by the occurrence of a large break in the inlet headers (IH) of a nuclear reactor. Researchers have generated transient data using RELAP5 for detecting the break sizes based on artificial neural networks (NNs). Detecting large breaks accurately can be challenging because it is inherently difficult to detect the abnormal patterns of transient data of large breaks due to the transient data changing so rapidly that they change in almost real-time. This project proposes a methodology for constructing an optimal NN based on the transient data of inlet headers to identify large breaks at high accuracy. The methodology is composed of 1) training, validation and testing of NNs with different number of hidden nodes; 2) selection of the optimal NN; and 3) robustness testing of the optimal NN against noisy data. The results show that the optimal NN obtained has an accuracy of 98.5% in detecting break sizes.The optimal NN is robust against noisy data with an average case accuracy of 97.5%±0.25% on a noisy dataset of unseen break sizes. Neural networks can be trained on transient datasets of a NPP to detect LOCA of the NPP. However, the transient datasets exhibit big data characteristics and designing an optimised neural network by exhaustive training all possible neural network architectures on big data can be very time-consuming. This project proposes a neural network design methodology in three stages to detect the break sizes of the IHs of a NPP. In stage one, an optimised 1-hidden layer multilayer perceptron (MLP) is obtained. In stage two, a number of 2-hidden layer MLP architectures are determined; then, an optimised 2-hidden layer MLP is obtained . In stage three, the break sizes not present in the transient dataset are generated using linear interpolation method. The results show that the proposed methodology outperformed the MLP of the previous work. The length of data sets for neural network training has great effect on the performance. This project proposes a methodology based on short-time Fourier transform to find a training set of minimum size for NNs in detecting LOCA in NPPs. The results show that a minimum training set of the transient dataset accounts for 32.3% of the entire transient dataset. Additionally, the optimised network trained on the minimum training set has a much better performance than the optimised network trained on 50% of the transient dataset. The performance of the neural networks critically depends on their architectures and it is too costly to ?nd an optimal network architecture by exhaustively training all the architectures. In this project, a constraint-based genetic algorithm (GA) is used to ?nd optimal 2-hidden layer network architectures for detecting LOCA of a NPP. The GA uses a proposed constraint satisfaction algorithm called random walk heuristic to create an initial population of neural network architectures of high performance. The results showed that for LOCA detection, the GA-optimised network outperformed all the other approaches in terms of generalization performance. Finding the best performance of NNs is challenging. This project proposes a constraint-based random search algorithm for optimizing neural network (NN) architectures and ensemble construction in three stages for detecting the break size of an IH of a NPP. An optimised 2-hidden layer network, an optimised 3-hidden layer network and an optimised 4-hidden layer network are combined into a neural network ensemble (NNE) using a weighted meaning approach. The results show that the NNE outperformed the individual optimised neural networks in detecting the break size of an IH. |
| Exploitation Route | In May 2017 and June 2018, Bhabha Atomic Research Centre researcher came to Leeds Beckett and implemented the prediction model of this big break size to their nuclear safety system. The published result could be used by other users. |
| Sectors | Aerospace, Defence and Marine,Chemicals,Construction,Energy,Environment,Manufacturing, including Industrial Biotechology,Pharmaceuticals and Medical Biotechnology |
| Description | The findings have been widely used in Radiological Field, Environmental Study, Textile,, Fertilizer, Water Treatment, Oil and Gas Industries, Nuclear, Automobile ,Robotics, Stock Market, Mathematics, Software, Ship Modelling, Civil Engineering, Computational Fluid Dynamics, Sports, Mechanical System, Power Industry, Agriculture, Finance, Computer Science, Chemical and Process Industry, Human Reliability, Offshore, Maritime , Geotechnics, Mechanical Engineering, Power Grid, Prognostic and Health Management, Life Cycle Assessment ,Methane Leakage, Ocean Engineering, Health, Amplifiers, etc. |
| First Year Of Impact | 2022 |
| Sector | Aerospace, Defence and Marine,Agriculture, Food and Drink,Chemicals,Digital/Communication/Information Technologies (including Software),Electronics,Energy,Environment,Financial Services, and Management Consultancy,Healthcare,Manufacturing, including Industrial Biotechology,Pharmaceuticals and Medical Biotechnology,Transport,Other |
| Impact Types | Cultural,Societal,Economic,Policy & public services |
| Description | Fault tolerant control for increased safety and security of nuclear power plants |
| Amount | £236,000 (GBP) |
| Funding ID | EP/R021961/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 10/2018 |
| End | 07/2021 |
| Description | PhD studentship |
| Amount | £60,000 (GBP) |
| Organisation | Leeds Beckett University |
| Sector | Academic/University |
| Country | United Kingdom |
| Start | 01/2017 |
| End | 03/2019 |
| Title | A Constraint-Based Random Search Algorithm for Optimizing Neural Network Architectures and Ensemble Construction in Detecting Loss of Coolant Accidents in Nuclear Power Plants |
| Description | One major accident of a nuclear power plant (NPP) is the loss of a coolant accident (LOCA) which is caused by a large break in an inlet header (IH) of a nuclear reactor. This work proposes a constraint-based random search algorithm for optimizing neural network (NN) architectures and ensemble construction in three stages for detecting the break size of an IH of a NPP. In stage one, a number of 2-hidden layer, 3-hidden layer and 4-hiddden layer network architectures are created using a proposed constraint satisfaction algorithm. Then, an optimised 2-hidden layer network, an optimised 3-hidden layer network and an optimised 4-hidden layer network are chosen from these architectures by training and testing them on a transient dataset of IHs and a linear interpolation dataset. In stage two, the optimised 2-hidden layer network, the optimised 3-hidden layer network and the optimised 4-hidden layer network are trained and tested iteratively 200 times on the transient dataset to further improve their performance. In stage three, the optimised 2-hidden layer network, the optimised 3-hidden layer network and the optimised 4-hidden layer network are combined into a neural network ensemble (NNE) using a weighted meaning approach. The results show that the NNE outperformed the individual optimised neural networks in detecting the break size of an IH. |
| Type Of Material | Improvements to research infrastructure |
| Year Produced | 2018 |
| Provided To Others? | Yes |
| Impact | The work is being used at the Bhabha Atomic Research Centre in India for enhancing their fault diagnosis tool for nuclear reactors. |
| URL | https://ieeexplore.ieee.org/document/8648616 |
| Title | A constraint-based genetic algorithm for optimizing neural network architectures for detection of loss of coolant accidents of nuclear power plants |
| Description | Monitoring the safety of nuclear power plants (NPPs) is a crucial task in nuclear energy industry. One type of severe accidents of a NPP is the loss of coolant accident (LOCA) which can be caused by a large break in the inlet headers (IHs) of the primary heat transport (PHT) of a nuclear reactor. Nowadays, neural networks with 2-hidden layers trained on nuclear simulation transient datasets have been used as a dominant methodology to detect LOCA. However, the performance of the neural networks critically depend on their architectures and it is too costly to ?nd an optimal network architecture by exhaustively training all the architectures. In this project, a constraint-based genetic algorithm (GA) is used to ?nd optimal 2-hidden layer network architectures for detecting LOCA of a NPP. The GA uses a proposed constraint satisfaction algorithm called random walk heuristic to create an initial population of neural network architectures of high performance. At each generation, the GA population is split into a sub-population of inputs and a sub-population of 2-hidden layer architectures to breed offspring from each sub-population independently in order to generate a wide variety of network architectures. During the breeding of 2-hidden layer architectures, a constraint-based nearest neighbour search algorithm is proposed to ?nd the nearest neighbours of the offspring population generated by mutation. The performance of the GA in detecting LOCA was evaluated using a break size dataset generated by the RELAP-3D nuclear simulator and the skillcraft dataset from the UCI machine learning repository. The performance of the GA was compared with that of a random search, that of an exhaustive search and that of a RBF kernel support vector regression(SVR). The results showed that for LOCA detection, the GA-optimised network outperformed all the other approaches in terms of generalization performance. For the skillcraft dataset, the GA-optimised network has a similar performance to the RBF kernel SVR and outperformed the other approaches. |
| Type Of Material | Improvements to research infrastructure |
| Year Produced | 2018 |
| Provided To Others? | Yes |
| Impact | After using this model, the break size prediction is more accurate. This work is being used at the Bhabha Atomic Research Centre in India for enhancing their fault diagnosis tool for nuclear reactors. |
| URL | https://www.sciencedirect.com/science/article/abs/pii/S0925231218310750 |
| Title | Selecting a Minimum Training Set for Neural Networks using Short-Time Fourier Transform in Detecting Loss of Coolant Accidents in Nuclear Power Plants |
| Description | In safety monitoring of nuclear power plants (NPPs), neural networks (NNs) are trained on transient datasets of NPPs to detect loss of coolant accidents (LOCA) in NPPs. This project proposes a methodology based on short-time Fourier transform (STFT) to find a training set of minimum size for NNs in detecting LOCA in NPPs. The results show that a minimum training set of the transient dataset accounts for 32.3% of the entire transient dataset. Additionally, the optimised network trained on the minimum training set has a much better performance than the optimised network trained on 50% of the transient dataset and the optimised networks trained on smaller subsets of the minimum training set |
| Type Of Material | Computer model/algorithm |
| Year Produced | 2018 |
| Provided To Others? | Yes |
| Impact | The work is being used at the Bhabha Atomic Research Centre in India for enhancing their fault diagnosis tool for nuclear reactors. |
| URL | https://csce.ucmss.com/cr/books/2018/LFS/CSREA2018/ICA3130.pdf |
| Description | Smart on-line monitoring for nuclear power plants |
| Organisation | Bhabbha Atomic Research Centre |
| Country | India |
| Sector | Public |
| PI Contribution | 1. Regular video conference meeting 2. Improve the model performance of BARC 3. provide better model to update BARC safety system |
| Collaborator Contribution | 1. Provide data for model building 2. explanation of their test facility 3. regular video conference meeting |
| Impact | 1. Better break size prediction model has been obtained from this collaboration 2.Signal processing method to be used to find the minimum data 3. Minimum data could be used to build accurate neural network model |
| Start Year | 2015 |
| Description | 2017 Summer School (Leeds Beckett) |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | This is a summer school, which is funded by Horizon 2020.There are large group of people. The purpose of this presentation in the summer school is to articulate the importance of the EPSRC research of " Smart online monitoring of nuclear power plants". The result has been demonstarted. |
| Year(s) Of Engagement Activity | 2017 |
| Description | 2018 Summer School |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | This is a summer school, which is funded by Horizon 2020.There are large group of people. The purpose of this presentation in the summer school is to articulate the importance of the EPSRC research of " Smart online monitoring of nuclear power plants". The result has been demonstrated. |
| Year(s) Of Engagement Activity | 2018 |
| Description | Final Project workshop |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Professional Practitioners |
| Results and Impact | 20 researchers attended the workshop, which sparked questions and discussion afterwards |
| Year(s) Of Engagement Activity | 2008 |
| Description | Research feature magazine |
| Form Of Engagement Activity | A magazine, newsletter or online publication |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Public/other audiences |
| Results and Impact | The activity is trying to inform general audiences about our work . |
| Year(s) Of Engagement Activity | 2018 |
| URL | https://researchfeatures.com/wp-content/uploads/2017/11/Professor-Jiamei-Deng-Leeds-Becket-Universit... |