Research On Room Temperature Control Algorithm In Household Heat Metering System

Household room temperature control in the household heat metering system is prerequisite for household heat metering and charging. The household room temperature control is also the main means of the household heat metering system to save energy. Adopting an effective control mode of room temperature can not only save a lot of energy, but also provide comfortable living environment for users and at the same time reduce the heating cost of users. However, the constant temperature control modes adopted at present are not good enough to meet the above requirements perferctly, and it is necessary to put forward a control scheme capable of meeting requirements of the household heat metering system.Based on the related control theories, a predictive control strategy for room temperature is proposed in this thesis. Using a neural network algorithm to predict time necessary to increase (reduce) temperature of heating rooms by a certain temperature, the strategy realizes constant temperature control by a fuzzy controller. Firstly, By analyzing the temperature variation in heating rooms, this thesis establishes mathematic models of heating in heating rooms and then analyzes various environment factors affecting the temperature variation in heating rooms. Secondly, this thesis establishes BP network predictive models, studies BP algorithm deeply and optimizes it. Considering the limitation of resources in microcontroller, microcomputer is used to complete BP network training which needs much computation and time. Only the predictive procedures run on the microcontroller for reducing operation. At the same time, the fuzzy controller which controls temperature of heating rooms is also designed. The algorithm on the microcontroller is implemented finally.In this thesis, Embedded Workbench for ARM is selected as software development environment, and the microcontroller STM32 is used as the hardware platform for the predictive algorithm. Test results show that the predictive algorithm achieves intended purpose and promise application values with its short consuming-time and high prediction accuracy.