Please use this identifier to cite or link to this item:
|Title:||Research of modeling and optimization approaches for hybrid ejector-based air conditioning cycle||Authors:||Wang, Hao||Keywords:||DRNTU::Engineering::Electrical and electronic engineering||Issue Date:||2017||Source:||Wang, H. (2017). Research of modeling and optimization approaches for hybrid ejector-based air conditioning cycle. Doctoral thesis, Nanyang Technological University, Singapore.||Abstract:||This thesis presents the research outcomes on the performance evaluation of a developed hybrid ejector-based air-conditioning system (hybrid system for short), an artificial neural network modeling for hybrid system with feature selection approaches for input parameters determination and trained with extreme learning machine (ELM), and the development and realization of a novel optimization approach based on single layer feed-forward neural network (SLFN) model. Furthermore, the centralized optimization problems of hybrid system are formulated, the corresponding solution methodology is then developed. The details of these results are as follows: 1. Two ejectors with different mixing chamber diameters are applied separately for performance test and the system is operated under different working modes. The effect of generating, condensing and evaporating pressure on system performance are studied experimentally. The effect of ejector geometry parameters on ejector performance is also investigated. The performance comparison between two working modes is made and the results indicate that 1) the performance of proposed system is sensitive to the three pressures; 2) the coefficient of performance (COP) of hybrid ejector-based air-conditioning system is around 34% higher than that of conventional compressor based system which implies a potential energy saving ability of proposed hybrid system. 2. An artificial neural network (ANN) with single hidden layer architecture is applied for modeling of hybrid system, further, an optimization issues considering the requirement of cooling capacity, ambient environment and system operating conditions, is developed. The ANN model is then combined with an exhaustive search algorithm to locate the system optimal set points under variant cooling capacities requirement. Both simulation and experiment results indicate that the ANN trained with ELM possesses lesser deviation between actual value and predicted value than the ones with other learning algorithms. 3. The model-based centralized optimization for hybrid system is developed based on evolutionary operators integrated particle swarm optimization method, which aims at minimizing the total energy consumption under demanded cooling capacities, with regard of components energy consumption models and corresponding physical constraints, achieving the precise estimation of the operating conditions as well as the global optimal set point. The main contribution of this thesis is to propose two systematic approaches in optimizing the hybrid refrigeration cycle. In practice, these approaches can be adjusted accordingly to balance of performance and efficiency requirements. These approaches offer significant advantages versus traditional control and optimization methods.||URI:||http://hdl.handle.net/10356/72648||DOI:||10.32657/10356/72648||Fulltext Permission:||open||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Theses|
Updated on May 10, 2021
Updated on May 10, 2021
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.