Neuro-Regression Approach in Optimization of Predicted Enthalpy of a Refrigeration System with Two-Stage Compression and Intercooling
| Main Authors: | DIAMBU, AYDIN |
|---|---|
| Format: | Proceeding Journal |
| Bahasa: | eng |
| Terbitan: |
, 2020
|
| Subjects: | |
| Online Access: |
https://zenodo.org/record/5260052 |
Daftar Isi:
- Two-stage compression and intercooling are one of the enhancements made to refrigeration systems to increase efficiency and reduce the size of installations. When engineers design cooling machines, the goal is not just sizing but also to insure their optimum operation. Therefore, this leads to system optimization. Some design-optimization studies based on regression analysis are found to be misleading in a realistic point of view even if R2 values of the regression models are quite high. In order to overcome this drawback, the present study performed a hybrid approach based on multiple nonlinear regression artificial neural networks to measure the accuracy of predictions of objective functions for the enthalpy taking into consideration their realistic characters. For this end, R2training and R2testing check are performed for different models based on the data (Enthalpy values for vapor phase) retrieved from a previous publication on a Refrigeration System with Two-Stage and Intercooler. A subsequent boundedness check allowed to choose the appropriate models for the optimization process using Wolfram Mathematica. First, all the models were analyzed (R2 training and testing check). After that, the models having high R2training and R2testing values, were also checked whether the functions are bounded or not. It’s seen that only one model satisfies the three criteria mentioned above. Hence, this study showed how useful boundedness check is in design modelling and optimization of engineering systems.