Summary: | 碩士 === 大同大學 === 機械工程研究所 === 90 === In air-conditioned room, the main factors that affect the human comfort are air velocity and temperature. To describe the relation among these state variables all over the air-conditioned room, the TS fuzzy models of real system are built by using data clustering algorithms. To increase the accuracy of models, the number and center of clusters are automatically and quickly adjusted according to certain criteria we proposed. That is, a fast and rough clustering is first performed by K-means algorithm. Then a clusters split/merge algorithm is applied which can automatically find suitable cluster centers for fuzzy clustering with Fuzzy c-means algorithm. Also, to demonstrate the feasibility of the clusters split/merge algorithm, the built fuzzy model of air-conditioned room is applied in various control approaches.
|