An adaptive control method for room air conditioners based on application scene identification and user preference prediction.

Room air conditioners are widely used to control indoor air parameters to user preferred values for thermal comfort, but the existing control methods might be uncomfortable due to changeable user preferences or be highcost due to physiological sensors. The purpose of this study is to develop an adaptive control method for room air conditioners at a low cost. The basic idea is to adopt data mining of operating parameters instead of monitoring by physiological sensors, and the key technology is the control of compressor frequency and indoor unit fan speed based on the application scene of the room air conditioner and the user preferred values of indoor air parameters. During the use of the room air conditioner, the application scene is identified by comparing the probabilities of the room air conditioner being in the sleep scene, work scene, or leisure scene, and the user preferred values are predicted by correcting the group preferred values of users in the same city with the setting records of the user. To ensure the reliability of the control method, the accuracy of application scene identification, user preference prediction, and adaptive control is validated by the data collected from the room air conditioners used in the cities of Shanghai, Guangzhou, Dalian, Wuhan, Chongqing, and Haikou. It is shown that the accuracy of application scene identification, user preferred air temperature prediction and user preferred air
velocity prediction is 79 %, 88 %, and 94 %, respectively; indoor air temperatures can be controlled within ±0.5 ◦C of the set values.