Data Mining On Energy Efficiency Decay Rate Of Domestic Air Conditioners

The research on domestic air conditioner energy efficiency decay rate was significant for the air conditioner energy saving capacity in long term operation. The influence factors of energy efficiency decay rate was analyzed here, and the pattern recognition was used to research the domestic air conditioner operation patterns, then the operation parameters setting optimization and control method was discussed to improve the energy saving capacity in the air conditioner long term running.Large amount of data related to domestic air conditioner operation performance was collected through questionnaire surveys and experiment to analyze the influence factors of energy efficiency decay rate; besides, long running data of the air conditioners was the most direct reflection of their operation state in actual surroundings, so it was collected through long term online monitoring to analyze the air conditioners operation internal pattern. Since lots of redundancy data and outlier data exist, data mining method was applied to solve this problem.First of all, data mining theory and energy efficiency decay rate data characters were analyzed to propose an energy efficiency decay rate data mining system, it mainly including energy efficiency decay rate database and data mining system. Database was built for collecting data associated with energy efficiency decay rate in its life cycle. Data mining system used the modular approach to finish data mining tasks and meet the different needs about air conditioner designing and using.Secondly, the questionnaire survey data and experiment data were analyzed to explore the direct and indirect factors of energy efficiency decay rate. The clustering algorithm was applied to find out the regular performance changing pattern of 45 air conditioner testing samples under three modes, which were used for long time, cleared heat exchanger and after refilling refrigerant, respectively. It showed that air conditioner performance would raise after heat exchanger was cleared, and when the refrigerant was refilled some of the air conditioners had a better performance but some had worse, which might be related with the changing of optimum filling ratio. Besides, the association rule mining was used to analyze the influence factors of particulate fouling on heat exchangers, including operation environment, user habit and maintenance strategy. Results showed that particulate fouling was a process affected by multiple factors that different parameter combinations or the same parameters with different settings could lead to a similar result in fouling degree.Since the air conditioner performance would change in long term running, the optimized control method developed in designing phase might be ineffective one day, the long running data was analyzed here to figure out a long term energy saving control method. The clustering algorithm was used to research the air conditioner operation patterns, it was found that there were three modes in daily operation: Morning High Load mode, Afternoon High Load mode and Stable Low Frequency mode. In the Morning High Load or Afternoon High Load mode under refrigerating mode, the outdoor temperature peaked in the morning and afternoon respectively, at the same time the energy consumption reached the maximum and the Energy Efficiency Ratio(EER) reached the minimum. In the Stable Low Frequency mode, the outdoor temperature was not so high so the air conditioner came into steady state quickly after it staring and the operation parameters were relatively stable all the day. Furthermore, the relationship between the various operating parameters of domestic air conditioner was analyzed by using the statistical and association rules, and the results showed that there was a clear relationship among the parameters. Then the neural network algorithm was used to develop a performance prediction model with the air inlet parameters taken as input parameters. And this model had been proved to be effective in predicting air conditioner energy consumption and EER with the error range within 15%, it could be used to control air conditioner performance through inlet air parameters. Finally, an online performance optimal controlling system was presented based on long term operation data, it would be benefit for improving air conditioners’ long term energy efficiency by combining air conditioner operation recognition, performance prediction and parameters optimization.