Scholarlink

<p>The effective management of water resources is challenging, given the complexity of environmental systems, the scarcity of data, and non-linear interactions, plus the uncertainty arising from factors such as climate change and human activities. Accurate river flow prediction plays an essential role in decision-making for sustainable water use. Models that use historical data and statistical techniques are used to analyze patterns, relationships, and trends in the data, and the area of study focuses on river flow prediction and its relevance in various sectors. The work addresses the Group Method of Data Handling (GMDH) model using the Nondominated Sorting Genetic Algorithm II (NSGA-II) algorithm for optimization and for exploring the set of results for the Pareto frontier, and it was found that both models met expectations. Despite the limitations encountered, the models proved to be very effective, maintaining values constantly below 0.50 for the Root Mean Square Error metric. To better understand the performance of the GMDH, the complexities are presented in terms of time and space. Time complexity refers to the time needed to train and predict with the model. In contrast, space complexity extends to storing the parameters of the polynomial functions or kernels that are used in the model. Finally, the analysis of the results underscores the effectiveness of the multiobjective approach in delivering balanced solutions concerning prediction accuracy and model complexity.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multi-criteria polynomial neural networks for hydrological time series modeling

  • 期刊论文

  • 发表日期：

    2025年5月20日

查看全文
  • Danilo P. M. Souza,
  • Gabriel E. Carlos,
  • Eliane da S. Christo,
  • Alfeu Dias Martinho,
  • Leonardo Goliatt

摘要

The effective management of water resources is challenging, given the complexity of environmental systems, the scarcity of data, and non-linear interactions, plus the uncertainty arising from factors such as climate change and human activities. Accurate river flow prediction plays an essential role in decision-making for sustainable water use. Models that use historical data and statistical techniques are used to analyze patterns, relationships, and trends in the data, and the area of study focuses on river flow prediction and its relevance in various sectors. The work addresses the Group Method of Data Handling (GMDH) model using the Nondominated Sorting Genetic Algorithm II (NSGA-II) algorithm for optimization and for exploring the set of results for the Pareto frontier, and it was found that both models met expectations. Despite the limitations encountered, the models proved to be very effective, maintaining values constantly below 0.50 for the Root Mean Square Error metric. To better understand the performance of the GMDH, the complexities are presented in terms of time and space. Time complexity refers to the time needed to train and predict with the model. In contrast, space complexity extends to storing the parameters of the polynomial functions or kernels that are used in the model. Finally, the analysis of the results underscores the effectiveness of the multiobjective approach in delivering balanced solutions concerning prediction accuracy and model complexity.