Please use this identifier to cite or link to this item:
|Title:||Testing study of ejector based multi-evaporator refrigeration system||Authors:||Seah, Jian Min||Keywords:||DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation::Control engineering||Issue Date:||2014||Abstract:||Three parameters of the temperature from three testing chambers are essential in the shipping container and food processing applications. These are often required for freezing and maintain the freshness of the food. Based on the working principle of conventional Multi-Evaporators Refrigeration System (MERS) and Ejector based Multi-Evaporation Refrigeration System (EMERS), it found out that EMERS is much better compare to conventional MERS in terms of the pressure is able to recover. Hence, system efficiency can be improved. A novel ejector MERS is chosen to build up the experiment setup. The use of experiment setup is to operate the system at different working condition by doing experiment one by one. The experiment data consists of observing two sets of data, namely Ejector data and Ejector new data. The Ejector data is an original data based on past experiment and new data is created to compile all the new record data based on different working condition. To see the performance of both ejector data, learning machine method, namely Extreme Learning Machine (ELM) is introduced. Extreme Learning Machine is one of the supervised learning methods that can learn the input-output mapping from the training data. The benefits of this ELM are better performance, small training error, smallest norm weights and fast learning speed. This method is very suitable to apply to generate a proper model for the ejector-based MERS project by training it with numerous experiment data. Past experiment had drawn a conclusion that the original data from the ejector was not fully covered within the testing domain, which yields the problem of inaccurate prediction of the ELM model. By inserting more record data, it can be analysed from the experiment done in this final year project that the prediction of the ELM learned algorithm can enhanced by reducing the cross validation error, training error and testing error.||URI:||http://hdl.handle.net/10356/60192||Rights:||Nanyang Technological University||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Student Reports (FYP/IA/PA/PI)|
Updated on Jan 18, 2021
Updated on Jan 18, 2021
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.