Scholarlink

The purpose of this study is to use the neural network modeling approach in order to identify the operating parameters influencing the rotary desiccant wheel module, particularly during the adsorption (dehumidification) and regeneration (desorption) processes. In this work, Matlab-network toolbox was used to find new structure of the neural network between input, hidden, and output layer. The parameters used as an input networks are air temperature of the process Tpro,in, regeneration air temperature Treg,in, the humidity ratio of the process Wpro,in, and the regeneration air humidity ratio Wreg,in. The network output includes the outlet air temperature and the outlet air humidity ratio of the dehumidification and regeneration processes. The performance of the network model was evaluated using the experimental and the predicted values and by mean square error (MSE), and statistical coefficient of determination (R). The training, validation and test performance regression yields (R) values of 0.999, 0.999, and 0.997, respectively. Additionally, the sensitivity analysis indicated that all four working parameters have a significant effect, with 32% relative importance of the process air humidity ratio.

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

Sensitivity Analysis of Operating Parameters on Solid Desiccant Dehumidification and Regeneration Processes Using Artificial Neural Networks Approach

  • 章节

  • 发表日期：

    2025年

查看全文
  • Fatih Bouzeffour,
  • Walid Tanne

摘要

The purpose of this study is to use the neural network modeling approach in order to identify the operating parameters influencing the rotary desiccant wheel module, particularly during the adsorption (dehumidification) and regeneration (desorption) processes. In this work, Matlab-network toolbox was used to find new structure of the neural network between input, hidden, and output layer. The parameters used as an input networks are air temperature of the process Tpro,in, regeneration air temperature Treg,in, the humidity ratio of the process Wpro,in, and the regeneration air humidity ratio Wreg,in. The network output includes the outlet air temperature and the outlet air humidity ratio of the dehumidification and regeneration processes. The performance of the network model was evaluated using the experimental and the predicted values and by mean square error (MSE), and statistical coefficient of determination (R). The training, validation and test performance regression yields (R) values of 0.999, 0.999, and 0.997, respectively. Additionally, the sensitivity analysis indicated that all four working parameters have a significant effect, with 32% relative importance of the process air humidity ratio.