Decay Rate Prediction Of Energy Efficiency Of Domestic Air Conditioners Based On BP Neural Network

Domestic air conditioner(DAC) energy efficiency evaluation is mainly based on the performance of the new machine in the specific standards of inside and outside conditions. However, DAC operation performance could be changed with the different standards between indoor and outdoor conditions, and various operative habits. Moreover, lots of factors such as compressor abrasion, refrigerant leakage, mixing of lube oil, corrosion of evaporator and condenser etc, could lead to performance degradation of the overall system over time.The long-term performance of DAC would be affected by the comprehensive influence of a lot of factors. In this paper, the BP neural network prediction model was built with the quantization and residual rate of long-term performance as the input and output parameters, respectively. In order to complete the learning and testing of RAC long-term performance prediction model, 22 room air conditioners were used for performance testing, with 85% as the training samples and 15% as model validation samples. In addition, Factors and their weights affecting long-term performance of DAC can be analyzed based on the model learning.The results showed that the DAC’s long-term performance evaluation mathematical model was convergence, in which 4 retention rates of LTP under 4 specific operation conditions and quantified RAC’s LTP values were set as input and output parameters respectively. The relative error range between fitted values and measured values for 19 groups of independent variables training BP neural network model was-6.024%~4.807%. Compared to 3 groups of independent variables validating BP neural network model, the relative error range between predicted values and measured values was-2.823%~6.094%. Analyzing the decision weights of BP neural network model which has finished its model learning, the time-weighted retention rates of high temperature cooling performance(0.187), rated cooling performance(0.203), low temperature heating performance(0.312), rated heating performance(0.298) could be calculate, respectively.To explore the relationship between the actual operation performance of air conditioner and the environment parameters such as the indoor and outdoor conditions and the user habits, an DAC on-line monitoring system was built to accomplish the real-time data acquisition and feed-back. Based on user background research, lots of data were analyzed and mined. The conclusions are summarized as follows:In Guangzhou, the probability of user turning on the air conditioner raised as the average outdoor air temperature increased, and the probability could reach upwards of 94.95% when the average outdoor air temperature come to 33℃. And the using time also increased as the average outdoor air temperature increased. The using time could reach upwards of 15 h when the average outdoor air temperature come to 33 ℃. In addition to the outdoor air temperature, other parameters affected the air condition using time were as follows: building area, air condition type, floor level and income of the family.For the inverter air conditioner, as the outdoor air temperature increased, the cooling capacity declined after reaching a peak, the consumed power increased at first and then remained invariant, and the energy efficiency ratio(EER) remained invariant at first and then decreased slowly.