The Simulation of Chiller-Running Behavior and Operation-Sequence Optimal Using Neural Network

• Main Authors: Yi-Lang Ku, 古義榔
• Other Authors: Po-Chuan Huang
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/71643302401286539137

碩士 === 國立臺北科技大學 === 冷凍與低溫科技研究所 === 92 === The main function of chiller provides the chilled water to execute air cooling and dehumidifieded for air-conditioning system that it maintains the product on the good yield. The power consumption of chiller-running is very large. Therfore, the innovation ideas are to search the simulation method of chiller system, to reduce the chiller -running cost and to reduce dependence on the professional engineers. This paper used the Principal Component Analysis (PCA) and the Covariance Matrix Analysis to sieve out the parmeters of chiller-running which they could impact the running performance and then setuped the very clear mondel of nural network. It was proven that we could use very clear model to meet the running-behaver of chillers according to the training and test result of Artificial Nural Network (ANN). We could also show that it had very good and reliable simulation status. Further, we applied the Artificial Neural Network with Optimization Design method to search the Optimal Chiller Sequence (OCS) and the Optimal Chiller Load Distribution (OCLD). We obtained the very good coefficient of performance (COP) enhamcement for chiller-running system. In a word, the simulation status indicated the result regardless of the single chiller or the whole chiller system, ANN with optimization design model provided not only the satisfying performance but provided convenient, fast and the most effective way to chiller system.