頁籤選單縮合
題 名 | 基於廣義迴歸神經網絡模型之分散式多臺嵌入式溫控節能系統=Distributed Multiple Embedded Temperature Control Energy-Saving Systems Based on General Regression Neural Network |
---|---|
作 者 | 朱力民; 湯久豎; 陳漢成; 蔡椿霳; 吳亦超; | 書刊名 | 臺東大學綠色科學學刊 |
卷 期 | 13:2 2023.11[民112.11] |
頁 次 | 頁91-103 |
分類號 | 446.73 |
關鍵詞 | 廣義迴歸神經網絡; 分散式; 嵌入式溫控節能系統; 權重值; 大數據資料庫; Distributed multiple embedded temperature control energy-saving systems; General regression neural network; Decentralized control; Database; |
語 文 | 中文(Chinese) |
中文摘要 | 本論文主要結合神經網路,開發「基於廣義迴歸神經網絡之分散式多台嵌入式溫控節能系統(Distributed Multiple Embedded Temperature Control Energy-Saving Systems based on General Regression Neural Network, DMETCESS-GRNN)」,讓整體系統運轉的電量消耗曲線收斂於透過 DMETCESS-GRNN 的電量消耗曲線。透過 DMETCESSGRNN,以最低整體總電量消耗讓多台嵌入式溫控節能裝置,以分散式運作方式,讓實際偵測溫度趨近於設定溫度,達到節能降溫的成效。本論文採用之廣義迴歸神經網絡模型(General Regression Neural Network, GRNN)將根據不同時間之環境因素,作權重值修正,來達到學習目的,並上傳至本論文建置之大數據資料庫,以作為下一週期 GRNN 的重要輸入,並再修正權重值,以分散式控制方式,讓多台嵌入式溫控節能裝置分別以不同風扇運轉功率,達到整體最低電量消耗。 |
英文摘要 | This paper proposed a Distributed Multiple Embedded Temperature Control EnergySaving Systems based on General Regression Neural Network (DMETCESS-GRNN) with neural networks. The power consumption curve of the overall system operation converges to the power consumption curve through DMETCESS-GRNN. In DMETCESS-GRNN, the multiple embedded temperature control energy-saving devices could achieve the energy saving and cooling effects with the lowest overall power consumption in a distributed manner to lead to actual detected temperature close to the set temperature. The General Regression Neural Network (GRNN) model used in this paper will modify the weight according to environmental factors at different times to achieve the learning purpose. The weight value was then uploaded to the big data database established in this paper as the important input of GRNN in next cycle. The weight value was thus revised with a decentralized control to allow multiple embedded temperature-controlled energy-saving devices to achieve the lowest overall power consumption with different fan powers. |
本系統中英文摘要資訊取自各篇刊載內容。