Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
645355 | Applied Thermal Engineering | 2015 | 45 Pages |
Abstract
The current study is an attempt to address the investigation of the CCHP (combined cooling, heating and power) system when 10 input variables were chosen to analyze 10 most important objective output parameters. Moreover, ANN (artificial neural network) was successfully applied on the tri-generation system on account of its capability to predict responses with great confidence. The results of sensitivity analysis were considered as foundation for selecting the most suitable and potent input parameters of the supposed cycle. Furthermore, the best ANN topology was attained based on the least amount of MSE and number of iterations. Consequently, the trainlm (Levenberg-Marquardt) training approach with 10-9-10 configuration has been exploited for ANN modeling in order to give the best output correspondence. The maximum MREÂ =Â 1.75% (mean relative error) and minimum R2Â =Â 0.984 represents the reliability and outperformance of the developed ANN over common conventional thermodynamic analysis carried out by EES (engineering equation solver) software.
Related Topics
Physical Sciences and Engineering
Chemical Engineering
Fluid Flow and Transfer Processes
Authors
Hadi Taghavifar, Simin Anvari, Rahim Khoshbakhti Saray, Shahram Khalilarya, Samad Jafarmadar, Hamid Taghavifar,