Study And Application Of Performance Simulation Refrigerated Display Cabinet Based On SVM Method

With the increase of the complexity of refrigeration system, modeling of refrigeration system with high accuracy and high efficiency has become a research hotspot in the refrigeration field. Based on the reaearch and modification of SVM (support vector machine, SVM) algorithm which is a new generation of machine learning using the statistical learning theory, this paper focuses on the research of the modelling, experiments and performance prediction of refrigerated display cabinet system. The main works are summarized as follows:Aiming at some defects on modeling, e.g. deficiency of the model generality, high skill on mdelling, and low efficiency on dealing with small sample size problem, the SVM machine learning algorithm with high generalization performance and good performance on dealing with small sample size problem is proposed.Based on the proposed SVM, the performance prediction model of frost growth on evaporator is built. The predicted results are found to be in good agreement with the experimental data, with mean relative error 1.82% for the total heat flux, 2.65% for the frost mass concentration, and 5.15% for the frost thickness. Then, a sensitivity analysis of the frost growth model is used to investigate the effects of the operating condition parameters that influence frost growth. Finally, the total heat flux prediction model is selected as an example to investigate the models'roughness by adding Gauss noise in the input vectors and output targets of the training set, respectively and together. The results show that the presented model is very suited to the frost growth prediction with high accuracy and good robust against noise.With the refrigerated display cabinet experiment table, the experiments of refrigerated display cabinet in water cooling condition and air cooling condition are achieved respectively. The key parameters which influence the performance of display cabinets are confired by the thermodynamic principleand experimental results. For the refrigerant circle of display cabinet in water cooling condition, it is verified by theories and experments that the refrigeration electrical energy consumption REC is represnted by saturated vapour temperature T_es, display cabinet inlet and outlet temperature T_in, T_out and refrigerant mass flow q_m with one certain condensing pressure. For the air circle (air curtain system) of display cabinet in water cooling condition, it is verified by theories and experments that the refrigeration electrical energy consumption REC is represnted by saturated vapour temperature T_es, indoor ambient temperature T_a , relative humidity h_a , air supply temperature T_s , air return temperature T_r , mesh diameter D and angle of air supply outletθwith one certain condensing pressure. For the display cabinet in air cooling condition, it is verified by theories and experments that the refrigeration electrical energy consumption REC is represnted by saturated vapour temperature T_es, condensing air inlet temperature T_c in, compressor speed N , display cabinet inlet and outlet temperature T_in, T_out and average M-packages temperature T_M with variable condensing condition or variable compressor speed.Moreover, the SVM is modified owing to the low accurarry of original SVM for complicated refrigeration system and its some key parameters (such as penalty coefficient, insensitive loss coefficient and kernel parameter) hard to be set in the practical application. Based on the modelling of frost growth on evaporator, the adaptive support vector machine (ASVM) method is introduced to the field of intelligent modeling of refrigerated display cabinets and used to construct a highly precise mathematical model of their energy consumption. A model for a variable speed open vertical display cabinet was constructed using preprocessing techniques for measured data, including the elimination of outlying data points by use of an exponential weighted moving average (EWMA). Using dynamic loss coefficient adjustment, the adaptation of the SVM for use in this application was achieved. From there, the object function for energy use per unit of display area TEC / TDA was constructed and solved using the ASVM method. When compared to the results achieved using a back-propagation neural network (BPNN) model, the ASVM model for the refrigerated display cabinet was characterized by its simple structure, fast convergence speed and high prediction accuracy (whin 5%). The EWMA-ASVM model also has better noise rejection properties than that of original SVM model. It was revealed by the theoretical analysis and experimental results that it is feasible to model of the display cabinet built using the EWMA-ASVM method with TEC / TDA decreased by 20.7%.Aiming to the complicated function and non-linear relationship among the parameters during the optimization of refrigerated display cabinets, and the time consuming and hard sledding problems of traditional design methods, the method for predicting the performance of refrigerated display cabinets which is based on a modified two-fluid (MTF) model and an adaptive support vector machine (ASVM) algorithm is presented. A MTF model (physical model) was built for open vertical refrigerated display cabinets, and then an ASVM algorithm (machine learning algorithm) was built. To verify the quantity of air leakage from the cabinet during operation, an important performance factor of display cabinets, an MTF model was built. After the training and validation data sets were constructed from the output of the MTF model, the problem was solved using an ASVM algorithm. After validation using experimental data, the TEC / TDA of the improved display cabinet with the optimal controlled parameters achieved from the strategy were found to be reduced by 39.2% and 19.3% respectively resulting in a significant reduction in cabinet air leakage. Moreover, from the optimal controlled parameters achieved from the modeling and expeiments, some guidelines of refrigerated display cabinets controlling can be generalized.