| Summary: | 碩士 === 國立臺灣科技大學 === 自動化及控制研究所 === 99 === A HVAC consumes heavy electricity in an intelligent building. It shows the importance of HVAC control for energy saving. This study investigates the performance of application of a Model Free Adaptive (MFA) controller to temperature control on a central air conditioning system. After collecting actual temperature data from a hotel in northern Taiwan, a simulated model of the HVAC load for restaurant is to built according to relationships between input and output. Then, both MFA and PID controllers are modeled and compared on the differences of time responses for steady state. In the first stage, a HVAC load model for restaurant is trained with Neural Network by using Matlab from collected 412 sets of data. The three inputs include control valve open, outside temperature, generated heat of restaurant, and the single output is the feedback temperature. The average trained error is around 1.055% between measured and simulated temperature. In second stage, simulated controller models with Fuzzy Control Theory are trained by using Matlab with inputs and outputs from both PID and MFA controller. By changing the setting parameters, the response time of steady state is measured for comparison. From the simulated results, the MFA controller has faster setting time of 7.78% than PID. It reduces action times of control valve and saves unnecessary energy wasting.
|
|---|
