以智慧代理者理論建立可調適環境之研究：智慧住宅外殼設計為例

The Study of Applying an Agent-based Theory to Build Adaptive Environment: A Smart House Envelope Design as an Example

資訊科技的進展使得智慧住宅成為重要的研究方向。雖然感測器、資訊通訊與網路科技的發展，已經啟發並且創造了新的智慧裝置，但是如何滿足使用者的需求仍是空間智慧化的重要課題。本研究提議以“環境代理者”作為動態反映使用者需求與環境對應的機制，並以模糊理論建立判斷與反應的控制條件，主動的推論與偵測不同的情境，作出最佳化的決定，並且經由學習，調適其機能以符合情境變化的不同需求。本研究以智慧住宅的外殼為例分析智慧化的可行性，並且以三項相關理論作為整合的關鍵要素包括：（1）智慧代理者、（2）涵構覺察、（3）模糊理論。藉由智慧化建築外殼的原型測試，評估分析使用者與環境對應關係，建立最基本的感知條件，本研究發現模糊邏輯的推論與類神經模糊的學習，使得環境代理者具備以使用者的經驗為導向的涵構覺察機能。

A smart house design has been an important research paradigm relying on information and current communication technology development. Although making use of environmental sensors, information, communication and internet technologies have inspired the creation of smart equipment; the question of 'How to satisfy occupants’ needs' is still an important issue for making living space smart. An intelligent agent, as a knowledge representation tool for computing and communication functions, is an autonomous software program of artificial intelligence. This paper proposes to build adaptive environmental agents, which are able to actively detect different situations and make optimum decisions relying on fuzzy-logic, inferring and adapting their functions to meet environmental variations and different occupants’ needs, following neuro-fuzzy learning. This research explored potentials of a smart house envelope and proposes three key elements for system integration: (1) intelligent agent, (2) context-awareness, and (3) fuzzy theory. This research discovers that correspondent relationships and experimental transmissions between occupants and environments make environmental agents’ user experience-oriented, with context-awareness abilities based on the smart envelope prototype’s construction, evaluation and analysis.