工業技術研究院

技術簡介

本技術開發基於增強式深度學習(DRL)的人工智慧節電控制軟體，克服商用場域之多目標與控制策略易發散的技術瓶頸，在幾乎不增加硬體與施工成本下，達成整店8~14％節電成效。本技術於控制條件極為嚴苛之食品零售店場域進行技術示範與驗證，獲得穩定且有效之節能效益。

Abstract

This technology develops artificial intelligence energy saving control software based on deep reinforcement learning (DRL), which overcomes the technical bottleneck of multi-objective and control strategy divergence in commercial fields, and achieves 8％ ~ 14％ power saving for the entire store without increasing hardware and construction costs. This technology has been demonstrated and verified in food retail stores with extremely stringent control conditions, achieving stable and effective energy-saving benefits.

技術規格

1.可應用於多系統間交互影響時的複雜情境，透過及時掌握環境變化、設備之製冷製熱、其他設備運轉情況，自動生成最佳目標之設備運轉參數組合。 2.運用強化式深度學習演算法，對空調、冷凍櫃、冷藏櫃，進行最佳化控制，可將店面控制於最佳的設備運轉條件以最大化節能成效。

Technical Specification

1. It can be applied to complex situations where multiple systems interact with each other. By timely grasping environmental changes, cooling and heating of equipment, and the operation of other equipment, it can automatically generate the best target equipment operation parameter combination. 2. By using deep reinforcement learning algorithms to optimize the control of air conditioners, freezers, and refrigerators, stores can be controlled under the best equipment operating conditions to maximize energy saving results.

技術特色

本系統使用深度增強式學習做為基礎之新一代能源管理系統關鍵技術，整合食品零售業場域中各類耗能設備，並可自動進行最佳化運轉控制，使店舖兼具節能、環境舒適性和商品品質。

應用範圍

人工智慧、能源管理系統

接受技術者具備基礎建議（設備）

無

接受技術者具備基礎建議（專業）

C＋＋ / C sharp / python程式能力

聯絡資訊