Please use this identifier to cite or link to this item:
https://accedacris.ulpgc.es/handle/10553/137932
| Title: | ANN model presented in "ANN-based surrogate model for the structural evaluation of jacket support structures for offshore wind turbines" | Authors: | Quevedo Reina, Román Álamo Meneses, Guillermo Manuel Aznárez González, Juan José |
UNESCO Clasification: | Investigación | Issue Date: | 2025 | Publisher: | Zenodo | Description: | <p>This repository contains the best of the models developed in the scientific article "ANN-based surrogate model for the structural evaluation of jacket support structures for offshore wind turbines" by Román Quevedo-Reina, Guillermo M. Álamo, Juan J. Aznárez. (https://doi.org/10.1016/j.oceaneng.2024.119984)</p>
<p>This is an enssemble model of 20 artificial neural networks with 74 neurons in the input layer, 5 hidden layers with 250 neurons per hidden layer, and 26 neuron in the output layer.</p> <div>The model is stored in the file "model.mat" and can be used by the included "class_model_classification" class.</div> <div> </div> <div><strong>Model Contents</strong></div> <div> </div> <div>The "model.mat" file contains a Matlab variable named "model", which has the following properties:</div> <ul> <li>dlnet: A collection of 20 artificial neural networks (ANNs) generated in the study.</li> <li>name_input: Names of the input variables for the model (described in the scientific article).</li> <li>name_output: Names of the partial checks predicted by the model (indicated in the scientific article and detailed in http://dx.doi.org/10.1016/j.oceaneng.2024.117802).</li> <li>mean_input: Mean values of each input variable, obtained from the training dataset. Used for normalization.</li> <li>std_input: Standard deviation of each input variable, obtained from the training dataset. Used for normalization.</li> </ul> <div><strong>How to Use the Model</strong></div> <div> </div> <div><em>Prerequisites</em></div> <div>To use the model, you must have the Deep Learning Toolbox installed in Matlab.</div> <div> </div> <div><em>Evaluation Method</em></div> <div>The model can be evaluated using the "eval" method included in the "class_model_classification". This method provides the following options:</div> <div> </div> <ul> <li>output: Specifies how the global feasibility is computed from the partial checks. <ul> <li>"global": Global feasibility is computed as the conjunction of independent events. (Default)</li> <li>"global_min": Global feasibility is equal to the minimum of the partial checks.</li> <li>"global_geomean": Global feasibility is the geometric mean of the partial checks.</li> <li>"partial": Returns the partial checks predictions.</li> </ul> </li> <li>ensemble: Specifies how the individual ANN predictions should be aggregated (more details in the scientific article). The options are: <ul> <li>"mean": Mean of the predictions from the individual ANNs.</li> <li>"voting": Proportion of individual ANNs whose predictions are greater than 0.5. (Default)</li> <li>"bayes": The combined probability of feasibility is updated by the evidences given by individual ANNs, following the Bayesian inference.</li> <li>"none": Returns the prediction of individual ANNs.</li> </ul> </li> <li>ensembledvar: Specifies how the individual ANN predictions are agregated on partial checks predictions or on global feasibility prediction: <ul> <li>"partial": Ensembled is made on partial checks predictions. (Default)</li> <li>"global": Ensembled is made on global feasibility predictions.</li> </ul> </li> <li>IDnet: Specifies which ANNs to use for the evaluation. The options are: <ul> <li>"all": Uses all ANNs embedded in the model. (Default)</li> <li>any number: Uses only the ANN at the specified position.</li> <li>any vector: Uses only the ANNs at the positions specified in the vector.</li> </ul> </li> </ul> <div> </div> <div><em>Input Data Format</em></div> <div>Prepare your input as a matrix where:</div> <ul> <li>Each row represents a sample.</li> <li>The matrix must have 74 columns, corresponding to the 74 input variables, arranged in the same order as specified in name_input.</li> </ul> <div> </div> <div><em>Example</em></div> <div>Below is an example of how to load the model and use the eval method:</div> <div> </div> <div> </div> <div>% Load the model</div> <div>load('model.mat');</div> <div> </div> <div>% Prepare your input data (using case example)</div> <div>load("case_example.mat"); % Load case example</div> <div> inputData=example{:,:}; % Extract matrix of data</div> <div> </div> <div>% Evaluate the model (default)</div> <div>output = model.eval(inputData);</div> <div> </div> <div>% Evaluate the model (specific options)</div> <div>output = model.eval(inputData,'output','partial','IDnet',1:10);</div> |
URI: | https://accedacris.ulpgc.es/handle/10553/137932 | Other Identifiers: | https://doi.org/10.5281/zenodo.15132439 oai:zenodo.org:15132439 |
Rights: | info:eu-repo/semantics/openAccess Creative Commons Attribution 4.0 International https://creativecommons.org/licenses/by/4.0/legalcode |
| Appears in Collections: | Datasets ULPGC |
Items in accedaCRIS are protected by copyright, with all rights reserved, unless otherwise indicated.