Summary: | 碩士 === 國立臺北科技大學 === 冷凍與低溫科技研究所 === 92 === The function of ice-storing system in Taiwan is not as good as expecting that bad controlling is the main problem. Load forecasting or other forecasting methods are not used to manage chiller and ice-storing amount is the main reason of not-effective controlling. The main aim of HVAC load forecasting is help operators to reach the expect storage by measuring tomorrow’s ice-store capacity.
Many methods such as time series method, regression analysis, and neural networks are adopted in load forecasting in recent years. And high accuracy is obtained by using neural networks which is a worthy used method.
For ice-storing system, the demand of operators can be satisfied to get forecast of tomorrow’s ice-store capacity(for one day). The essay brings the data of weather prediction to be the influencing factors of HVAC load forecasting. We can get more precise HVAC load forecasting by exact weather condition which can be offered by accurate weather prediction. In this experiment, 80~90% accuracy of HVAC load forecasting can be reached to use neural networks by the data of weather prediction to be the influencing factors.
|