The Design Of The Adaptive PID Controller In Air Conditioning System Based On Machine Vision

With the tremendous development of intelligent building, its wide spread impact on people’s daily life has been far-reaching. However, how to intelligently control the indoor temperature, satisfying people’s purpose of pursuing for comfort while saving energy is what human beings value most in the intelligent building field. Most of existing air conditioning control systems detect the current environment temperature mainly depends on temperature sensor. Once a large number of people pour into the room at some moment, if merely equipped with a temperature sensor, the control system requires temperature transmission delay, undoubtedly, several minutes to detect the rising of the environment temperature. Under such circumstance, the transmission delay delays the adjustment of the controller, even causes wide scale of temperature fluctuation.This research attempts to apply the machine vision to the air conditioning control system.We can detect the change in number of people rapidly by using machine vision technology. The definite advantage of machine vision technology lies in its fast speed, high efficiency and accurate identification.According to the exact number of people changed,the controller to the air conditioning system based on machine vision eliminats the definite temperature fluctuation resulting from the change of the numbers of people.The paper mainly aims at putting forward the framework for the realization of the system, realizing monitoring the change in number of people dynamically and designing the adaptive PID controller.After giving brief introduction of the main steps of configuring embedded development environment with OpenCV, the following three aspects will be discussed:Firstly, how to accurately count the number of the people in and out of the room by applying the machine vision technology. The first challenge is how to extract the dynamic background, thus, the foreground image can be extracted from the background subtraction. Afterwards, building the head model by using the features of circularity and Hu’s invariant moments, through which the target of person’s head can be identified and calculated according to the foreground image.Secondly, how to maintain the temperature steady in case of the fluctuation.The very start of the procedure is building an inner temperature model and applying quantitative approach to characterize the main factors that affect the room temperature. With the above mentioned model, the PID controller can easily eliminats the sinificant disturbance of the tempreature.At last, compared with the existing PID controller, the adaptive PID controller based on machine vision technology can effectively eliminate the wide range of temperature fluctuation caused by the changing number of people. It is needless to say that the latter controller is obliviously superior to the former regarding controlling capacity.