Hyperparameter Optimization of a Convolutional Neural Network Model for Pipe Burst Location in Water Distribution Networks
The current paper presents a hyper parameterization optimization process for a convolutional neural network (CNN) applied to pipe burst locations in water distribution networks (WDN). The hyper parameterization process of the CNN includes the early stopping termination criteria, dataset size, datase...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-03-01
|
| Series: | Journal of Imaging |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2313-433X/9/3/68 |
| _version_ | 1797610938320289792 |
|---|---|
| author | André Antunes Bruno Ferreira Nuno Marques Nelson Carriço |
| author_facet | André Antunes Bruno Ferreira Nuno Marques Nelson Carriço |
| author_sort | André Antunes |
| collection | DOAJ |
| description | The current paper presents a hyper parameterization optimization process for a convolutional neural network (CNN) applied to pipe burst locations in water distribution networks (WDN). The hyper parameterization process of the CNN includes the early stopping termination criteria, dataset size, dataset normalization, training set batch size, optimizer learning rate regularization, and model structure. The study was applied using a case study of a real WDN. Obtained results indicate that the ideal model parameters consist of a CNN with a convolutional 1D layer (using 32 filters, a kernel size of 3 and strides equal to 1) for a maximum of 5000 epochs using a total of 250 datasets (using data normalization between 0 and 1 and tolerance equal to max noise) and a batch size of 500 samples per epoch step, optimized with Adam using learning rate regularization. This model was evaluated for distinct measurement noise levels and pipe burst locations. Results indicate that the parameterized model can provide a pipe burst search area with more or less dispersion depending on both the proximity of pressure sensors to the burst or the noise measurement level. |
| first_indexed | 2024-03-11T06:20:58Z |
| format | Article |
| id | doaj.art-3883a6d54aea4218b3f375557655736f |
| institution | Directory Open Access Journal |
| issn | 2313-433X |
| language | English |
| last_indexed | 2024-03-11T06:20:58Z |
| publishDate | 2023-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Journal of Imaging |
| spelling | doaj.art-3883a6d54aea4218b3f375557655736f2023-11-17T11:55:16ZengMDPI AGJournal of Imaging2313-433X2023-03-01936810.3390/jimaging9030068Hyperparameter Optimization of a Convolutional Neural Network Model for Pipe Burst Location in Water Distribution NetworksAndré Antunes0Bruno Ferreira1Nuno Marques2Nelson Carriço3Sustain.RD, Escola Superior de Tecnologia de Setúbal, Instituto Politécnico de Setúbal, 2914-508 Setúbal, PortugalINCITE, Escola Superior de Tecnologia do Barreiro, Instituto Politécnico de Setúbal, 2839-001 Lavradio, PortugalNOVA LINCS, Department of Computer Science, Faculdade de Ciências e Tecnologia, Universidade NOVA de Lisboa, 2829-516 Caparica, PortugalINCITE, Escola Superior de Tecnologia do Barreiro, Instituto Politécnico de Setúbal, 2839-001 Lavradio, PortugalThe current paper presents a hyper parameterization optimization process for a convolutional neural network (CNN) applied to pipe burst locations in water distribution networks (WDN). The hyper parameterization process of the CNN includes the early stopping termination criteria, dataset size, dataset normalization, training set batch size, optimizer learning rate regularization, and model structure. The study was applied using a case study of a real WDN. Obtained results indicate that the ideal model parameters consist of a CNN with a convolutional 1D layer (using 32 filters, a kernel size of 3 and strides equal to 1) for a maximum of 5000 epochs using a total of 250 datasets (using data normalization between 0 and 1 and tolerance equal to max noise) and a batch size of 500 samples per epoch step, optimized with Adam using learning rate regularization. This model was evaluated for distinct measurement noise levels and pipe burst locations. Results indicate that the parameterized model can provide a pipe burst search area with more or less dispersion depending on both the proximity of pressure sensors to the burst or the noise measurement level.https://www.mdpi.com/2313-433X/9/3/68convolutional neural networksdeep learninghydraulic modelhyper parameterizationpipe burst location |
| spellingShingle | André Antunes Bruno Ferreira Nuno Marques Nelson Carriço Hyperparameter Optimization of a Convolutional Neural Network Model for Pipe Burst Location in Water Distribution Networks Journal of Imaging convolutional neural networks deep learning hydraulic model hyper parameterization pipe burst location |
| title | Hyperparameter Optimization of a Convolutional Neural Network Model for Pipe Burst Location in Water Distribution Networks |
| title_full | Hyperparameter Optimization of a Convolutional Neural Network Model for Pipe Burst Location in Water Distribution Networks |
| title_fullStr | Hyperparameter Optimization of a Convolutional Neural Network Model for Pipe Burst Location in Water Distribution Networks |
| title_full_unstemmed | Hyperparameter Optimization of a Convolutional Neural Network Model for Pipe Burst Location in Water Distribution Networks |
| title_short | Hyperparameter Optimization of a Convolutional Neural Network Model for Pipe Burst Location in Water Distribution Networks |
| title_sort | hyperparameter optimization of a convolutional neural network model for pipe burst location in water distribution networks |
| topic | convolutional neural networks deep learning hydraulic model hyper parameterization pipe burst location |
| url | https://www.mdpi.com/2313-433X/9/3/68 |
| work_keys_str_mv | AT andreantunes hyperparameteroptimizationofaconvolutionalneuralnetworkmodelforpipeburstlocationinwaterdistributionnetworks AT brunoferreira hyperparameteroptimizationofaconvolutionalneuralnetworkmodelforpipeburstlocationinwaterdistributionnetworks AT nunomarques hyperparameteroptimizationofaconvolutionalneuralnetworkmodelforpipeburstlocationinwaterdistributionnetworks AT nelsoncarrico hyperparameteroptimizationofaconvolutionalneuralnetworkmodelforpipeburstlocationinwaterdistributionnetworks |