Smart on-line monitoring for nuclear power plants (SMART)
Lead Research Organisation:
University of Portsmouth
Department Name: School of Engineering
Abstract
Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.
Publications
Becerra V
(2020)
Estimation of Radioactivity Release Activity Using Non-Linear Kalman Filter-Based Estimation Techniques
in Energies
Patelli E
(2017)
Simulation methods for system reliability using the survival signature
in Reliability Engineering & System Safety
Tian X
(2018)
A study on the robustness of neural network models for predicting the break size in LOCA
in Progress in Nuclear Energy
Tolo S
(2019)
Robust on-line diagnosis tool for the early accident detection in nuclear power plants
in Reliability Engineering & System Safety
| Description | The diagnosis of loss of coolant accidents (LOCA) in nuclear reactors has attracted a great deal of attention in condition monitoring of nuclear power plants because the health of cooling system is crucial to the stability of the nuclear reactor. Multi-layer perceptron (MLP) neural networks have commonly been applied to LOCA diagnosis. The data used for training these models consists of a number of time series, each for a different break size, with the transient behavior of different measurable variables in the coolant system of the reactor following a LOCA accident. It is important to select an architecture for the neural network that delivers robust results, in that the predicted break size is deemed to be accurate even for break size values not included in the training data sets. During the work carried out in this research grant, a method for measuring the robustness of diagnostic models for predicting the break size in loss of coolant accidents has been developed. A robustness metric is proposed based on the leave-one-out approach and the mean squared error resulting from a diagnostics model. Using this metric it becomes possible to compare the robustness of different diagnostic models. Given data obtained from a high fidelity simulation of the coolant system of a nuclear reactor, several different diagnostic models have been obtained and their properties compared. These models include a fully connected multi-layer perceptron with one and two hidden layers, a multi-layer perceptron with one hidden layer pruned using the optimal brain surgeon algorithm, a group method of data handling (GMDH) neural network, and an ANFIS neuro-fuzzy system. Moreover, given that the data sets employed for training the neural network models correspond to a sparse set of values of the break size, it has been noticed that the ability of the neural network models to predict break sizes that are outside the training set is often not very good. Considering the sparsity of the available transient data, an approach has been developed that permits to generate additional data sets which, when used for training the neural networks along with the original data, allow to significantly increase the accuracy and robustness of neural network models even when predicted break sizes which are not present in the original transient data sets. Moreover, a mechanism is being developed to combine the outputs of different models, where the output of the combined model produces exhibits higher robustness and accuracy than the outputs of the individual models. Furthermore, a novel approach to tackle the challenges associated with the on-line diagnostics of loss of coolant accidents and the limitations of the current state of the art was developed and implemented in the form of a software tool. Based on the combination of a set of artificial neural network architectures through the use of Bayesian statistics, this tool allows to robustly absorb different sources of uncertainty without requiring their explicit characterization. The tool provides the quantification of the output confidence bounds but also enhances of the model response accuracy. The implemented methodology allows to relax the need for model selection as well as to limit the demand for user-defined analysis parameters. Finally, the estimation of radioactivity release following an accident in a nuclear power plant is crucial due to its short and long-term impacts on the surrounding population and the environment. In the case of any accidental release, the activity needs to be estimated quickly and reliably to effectively plan a rapid emergency response and design an appropriate evacuation strategy. The accurate prediction of incurred dose rate during normal or accident scenario is another important aspect. We have proposed the use of three different non-linear estimation techniques, extended Kalman filter, unscented Kalman filter, and cubature Kalman filter in order to estimate release activity and to improve the prediction of dose rates. Radionuclide release rate, average wind speed, and height of release are estimated using data from the dose rate monitors that are collected in the proximity of the release point. Further, the estimates are employed to improve the prediction of dose rates. |
| Exploitation Route | This project involves collaboration with the Bhabha Atomic Research Centre (BARC) in Mumbai, India. The work done has been presented to and discussed with collaborators at BARC. The robustness measure that has been developed, the new neural network architectures that have been evaluated, the new method to generate new data sets for break levels that are not present in the original data, and the approach for model combination are very likely to be used by BARC, as the robustness achieved with these approaches is measurably better than what they have achieved with their existing diagnostic models. Moreover, once published, the approaches that have been developed for the diagnostic of loss of coolant faults are likely to be used by other organisations. |
| Sectors | Energy Environment |
| Description | The findings from this project are being used at least by the Bhabha Atomic Research Centre in Mumbai, India. The main use of this research relates to improvements in robustness and accuracy of models used for predicting the characteristics of critical faults in nuclear reactors. As the work has already been published, it is possible that impact will take place elsewhere as well, however it is still too early to ascertain such additional impact. Work on the the estimation of radionuclide release activity was presented by Professor Becerra at the workshop "Current Advances in Risk Assessment for Radiological releases during Nuclear Emergencies" (CARE-2019), which took place in the Indira Gandhi Centre for Atomic Research in Chennai, India, on August 5-7 2019. Discussions are ongoing for a future EPSRC research proposal in collaboration with IGCAR on this topic. An institutional report that summarises this project has been published by the Bhabha Atomic Research Centre. The report can be accessed through this URL: https://inis.iaea.org/search/search.aspx?orig_q=RN:50071217 |
| First Year Of Impact | 2018 |
| Sector | Energy |
| Impact Types | Societal Economic |
| Description | Fault tolerant control for increased safety and security of nuclear power plants |
| Amount | £292,834 (GBP) |
| Funding ID | EP/R022062/1 |
| Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
| Sector | Public |
| Country | United Kingdom |
| Start | 12/2018 |
| End | 05/2020 |
| Description | Model based state estimation for post release phase monitoring around nuclear power plants |
| Amount | £10,000 (GBP) |
| Organisation | University of Portsmouth |
| Sector | Academic/University |
| Country | United Kingdom |
| Start | 01/2019 |
| End | 07/2019 |
| Title | Methods to study the robustness of neural network models for predicting the break size in LOCA |
| Description | The diagnosis of loss of coolant accidents in nuclear reactors has attracted a great deal of attention in condition monitoring of nuclear power plants given that the health of the cooling system is crucial to the nuclear reactor's stable operation. Many different types of neural networks have commonly been applied to loss of coolant accident diagnosis. It is important to select a suitable architecture for the neural network that delivers robust results, in that the predicted break size is deemed to be accurate even for break sizes that are not included in the training data sets. A robustness metric was proposed and applied to compare the robustness of different diagnostic models. The data used for training these models consists of a number of time-series data sets, each for a different break size, with the transient behaviuor of different measurable variables in the coolant system of a nuclear reactor, following a simulated loss of coolant accident in a high-fidelity simulator. Given the simulation data for different break sizes, four different neural network architectures are investigated and their properties are compared and discussed. These models include a fully-connected multilayer perceptron with one hidden layer, a multilayer perceptron with one hidden layer that is pruned using the optimal brain surgeon algorithm, a fully-connected multilayer perceptron with two hidden layers, and a group method of data handling neural network. An interpolation pre-processing method is investigated and shown to be effective to further improve the capability of neural networks for robustly predicting the break size of a loss of coolant accident. Both linear interpolation and cubic spline interpolation are studied as alternatives for the pre-processing approach. The performance of models developed with and without interpolation pre-processing are compared with the previously proposed robustness metric. Finally, a combined diagnostic model is proposed based on three different architectures to obtain high prediction accuracy and good robustness. |
| Type Of Material | Improvements to research infrastructure |
| Year Produced | 2018 |
| Provided To Others? | Yes |
| Impact | The work is being used at the Bhaba Atomic Research Centre in India to improve their fault diagnosis tools for nuclear reactors. |
| URL | https://www.sciencedirect.com/science/article/pii/S0149197018301707 |
| Title | Robust on-line diagnosis methods for the early accident detection in nuclear power plants |
| Description | Any loss of coolant accident mitigation strategy is necessarily bound by the promptness of the break detection as well as the accuracy of its diagnosis. The availability of on-line monitoring tools is then crucial for enhancing safety of nuclear facilities. The requirements of robustness and short latency implied by the necessity for fast and effective actions are undermined by the challenges associated with break prediction during transients. A novel approach was developed to tackle the challenges associated with the on-line diagnostics of loss of coolant accidents and the limitations of the current state of the art. Based on the combination of a set of artificial neural network architectures through the use of Bayesian statistics, it allows to robustly absorb different sources of uncertainty without requiring their explicit characterization in input. It provides the quantification of the output confidence bounds but also enhances of the model response accuracy. The implemented methodology allows to relax the need for model selection as well as to limit the demand for user-defined analysis parameters. A numerical case-study entailing a 220 MWe heavy-water reactor is analysed in order to test the efficiency of the developed computational tool. |
| Type Of Material | Improvements to research infrastructure |
| Year Produced | 2019 |
| 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://www.sciencedirect.com/science/article/pii/S0951832018304253 |
| Description | Collaborations under the SMART project |
| Organisation | Bhabbha Atomic Research Centre |
| Country | India |
| Sector | Public |
| PI Contribution | Each partner has different roles in the project, with the BARC partner providing data and expertise, and the UK University partners performing research. |
| Collaborator Contribution | University of Liverpool provided expertise on probabilistic neural network models and related analysis, as well as nuclear engineering Leeds Beckett University provided expertise on fault detection and isolation, as well as neural network modelling. Bhabha Atomic Research Centre provided expertise on nuclear engineering as well as nuclear reactor data that was used during the project. |
| Impact | Two journal papers, one in Applied Nuclear Energy and the other one in Reliability Engineering. See the publications section. |
| Start Year | 2016 |
| Description | Collaborations under the SMART project |
| Organisation | Leeds Beckett University |
| Country | United Kingdom |
| Sector | Academic/University |
| PI Contribution | Each partner has different roles in the project, with the BARC partner providing data and expertise, and the UK University partners performing research. |
| Collaborator Contribution | University of Liverpool provided expertise on probabilistic neural network models and related analysis, as well as nuclear engineering Leeds Beckett University provided expertise on fault detection and isolation, as well as neural network modelling. Bhabha Atomic Research Centre provided expertise on nuclear engineering as well as nuclear reactor data that was used during the project. |
| Impact | Two journal papers, one in Applied Nuclear Energy and the other one in Reliability Engineering. See the publications section. |
| Start Year | 2016 |
| Description | Collaborations under the SMART project |
| Organisation | University of Liverpool |
| Country | United Kingdom |
| Sector | Academic/University |
| PI Contribution | Each partner has different roles in the project, with the BARC partner providing data and expertise, and the UK University partners performing research. |
| Collaborator Contribution | University of Liverpool provided expertise on probabilistic neural network models and related analysis, as well as nuclear engineering Leeds Beckett University provided expertise on fault detection and isolation, as well as neural network modelling. Bhabha Atomic Research Centre provided expertise on nuclear engineering as well as nuclear reactor data that was used during the project. |
| Impact | Two journal papers, one in Applied Nuclear Energy and the other one in Reliability Engineering. See the publications section. |
| Start Year | 2016 |
| Title | Robust on-line diagnosis tool for the early accident detection in nuclear power plants |
| Description | This software tool implements a novel approach to tackle the challenges associated with the on-line diagnostics of loss of coolant accidents and the limitations of the current state of the art. The proposed strategy relies on the combination of more artificial neural network architectures through the use of Bayesian statistics, allowing to robustly absorb different sources of uncertainty without requiring their explicit characterization in input. As a result, it provides not only the quantification of the output confidence bounds but also the enhancement of the model response accuracy on the basis of the credibility of each individual architecture along the output domain. In addition to these advantages, the implemented methodology allows to relax the need for model selection as well as to limit the demand for user-defined analysis parameters, still guaranteeing the real-time feasibility of the computation. |
| Type Of Technology | Software |
| Year Produced | 2018 |
| Open Source License? | Yes |
| Impact | None yet |
| Description | Impact workshop: SMART online monitoring for nuclear power plants |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Other audiences |
| Results and Impact | The workshop was organised at Leeds Beckett University on 18 September 2018. The workshop involved presentations by the project partners including a representative from the Bhabha Atomic Research Centre, India. |
| Year(s) Of Engagement Activity | 2018 |
| Description | Seminar |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | National |
| Primary Audience | Industry/Business |
| Results and Impact | Professor Becerra gave a presentation on the estimation of radionuclide release activity using an Unscented Kalman Filter at the workshop: Current Advances in Risk Assessment for Radiological releases during Nuclear Emergencies (CARE-2019), IGCAR, Chennai, India, August 2019. |
| Year(s) Of Engagement Activity | 2019 |
| Description | Seminar by Professor Victor Becerra at Bhabha Atomic Research Centre, India. |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Professional Practitioners |
| Results and Impact | Seminar entitled "Computational Intelligence Methods for Anomaly Detection" |
| Year(s) Of Engagement Activity | 2016 |