- 學位論文
類神經網路於冷卻水塔模型建立之應用
Application of Artifical Neural Network for Modeling of Mechanical Cooling Tower
摘要
目前中央空調系統主流的冷卻水塔變頻控制方法，主要是利用固定冷卻水出水溫度，或是加裝外氣濕球溫度計，讓設定溫度可依照外氣濕球溫度的變化即時調整設定值，此效果雖然可以降低冷卻水塔在系統部分負載需求時的風扇耗電量，但是並非最節能的控制方法，由於冷卻水塔運作除了空氣的循環還包含了冷卻水的循環，若能充分掌握冷卻水塔風量與冷卻水量在降載的同時對於冷卻水塔散熱量的變化，將對冷卻水迴路最佳化的控制有莫大的幫助。 因此本研究主要是探討冷卻水塔在上述風量或水量參數改變狀況下，應用類神經網路，建立冷卻水塔散熱模型。由於冷卻水塔熱平衡方程式推導出冷卻水塔散熱量與另外四項參數間之關係，所以本實驗系統，利用調變運轉中之冷卻水塔風量及水量等參數，收集其相關運轉數據。最後再利用Matlab軟體內建之類神經網路工具，將輸入變數設定為冷卻水出水塔溫度、外氣濕球溫度、冷卻水塔風扇風量、冷卻水塔運轉水量；以及輸出目標值為冷卻水塔散熱量，建立冷卻水塔模型。 以實測值與模擬值之比較來驗證冷卻水塔模型，依建模方式可獲得平均誤差率0.72~2.13%；且R2值0.97~0.99。其結果呈現出由類神經網路建立之模型，可靠度非常高，若再搭配最佳化演算法，即可應用於空調系統冷卻水迴路最佳化運轉策略。
並列摘要
Currently, the Cooling Tower Variable Frequency Control Method is the mainstream for Central Air Conditioning Systems, which uses fixed cooling water temperature or outside air Wet Ball Thermometer to instantly adjust the value of the set temperature according to the temperature fluctuation of the outside air Wet Ball. Although this method can reduce the power consumption of the Fan in the Cooling Tower when a load is required by the system, it’s not an ideal energy conservation control method. In fact, the cooling water circulation is also part of the operation for the Cooling Tower in addition to the air circulation. Sufficient information over the change of the heat-radiating capacity for the Cooling Tower when derating the air flow and the cooling water of the Cooling Tower would bring a positive contribution to the optimized control of the cooling water loop. Driven by this, the main purpose of this Research will be focusing on the application of Artificial Neural Network to configure the Heat-radiating Model for the Cooling Tower when the parameters of the aforesaid air flow or cooling water are changing. To do so, the Cooling Tower Heat Balance Formula is used to establish the relationship between the heat-radiating capacity of the Cooling Tower and four other parameters. On this basis, the air flow and cooling water parameters of the Cooling Tower under the modulating process are utilized by the experimental system to collect relevant operating figures. Finally the Artificial Neural Network tool in the Matlab Software will be employed to set the input parameters as the Cooling water temperature, outside air wet ball temperature, air flow of Cooling Tower, and the cooling water; further, the output target value is also set as the heat-radiating capacity of the Cooling Tower for developing the Cooling Tower model. In addition, the tested values and the simulated values are compared to verify the Cooling Tower model. Through the model development method, the mean error of 0.72~2.13% is obtained, with R2 value as 0.97~0.99 accordingly. The result indicated that pretty high reliability can be achieved for the model configured by Artificial Neural Network. With the support of optimized computation method, it can be applied to the Optimization Operation Strategies for the cooling water loop of the Air Conditioning System.