普及家庭環境下之情境感知節能系統

Context-Aware Energy Saving System in a Pervasive Home Environment

葉慧文(Hui-Wen Yeh)

指導教授 ： 傅立成

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2011年)

https://doi.org/10.6342/NTU.2011.01987

摘要

近年來，由於能源過度使用所造成的能源危機及全球暖化問題，「節能」儼然已成為ㄧ個值得我們思考如何建立一個永續發展世界的重要議題。大多的節能研究著重於技術導向的解決方法，少數使用活動辨識技術來輔助節能系統，但他們仍然忽略了使用者的感受，因此提供的節能服務無法滿足使用者需求。
智慧家庭下的使用者活動辨識技術已逐漸成熟，其做法大多為在環境中佈建各類感測器以收集環境資訊以及使用者與環境之互動資料，例如：室內溫度、室內照度、使用者動靜及環境中感測器之觸發狀態，並建立活動辨識模型。現有的行為辨識節能系統多半只考慮當下與該活動相關性密切的感測資料，未考慮到背景已開啟之電器，這些被忽略的電器可能也與該活動相關且具大量能源消耗，而在多人活動辨識的情況下，辨識率往往較單人活動辨識低，此外，他們往往忽視環境中使用者的感受，可能導致節能效果不佳或更多不必要的能源消耗。
本研究的主要貢獻有以下三點：第一，將家電的能源消耗量納入推論考量，找出執行某活動所觸發的確切能源消耗，並透過推論技術得知該活動與某些家電耗電量的關係。第二，因為我們是過群體生活，為了能進行多人活動辨識，我們根據區域性來進行資訊匯集(data aggregation)以有效簡化多人辨識時所需資料關連(data association)的複雜度，如此亦能較準確推論該區域內的所有活動以及相對應的耗電量。第三，不只是單純的行為辨識，我們根據廣受接受的標準化舒適度評量指標來全面性地衡量環境中使用者的舒適程度，結合前述之能源相關行為辨識結果，在兼顧舒適度以及節能效應前提下提供適當的節能服務。

關鍵字

普及家庭環境 ； 行為辨識 ； 節能 ； 群體活動 ； 動態貝氏網路 ； 使用者舒適度 ； 情境感知

並列摘要

In the recent years, energy saving has become an important issue due to energy crisis and global warming caused by overused energy consumption. Therefore, it is worthy of concern for us to think about how to create a sustainable world. Most of the prior works on energy saving focused more on technology-oriented solutions whereas few works exploit activity recognition to assist energy saving system. Even taking human activity into consideration, most of them ignore user feeling. For this reason, the energy saving services these systems provided often cannot meet user need.
The techniques of activity recognition in a smart home have been more mature than ever. The researchers often deploy many kinds of sensors to collect environmental information and the interactions between users and their environment, e.g. indoor temperature, indoor illumination, users’ motion and states, to build activity recognition models. Now the present home energy saving systems based on activity recognition merely take those appliances switched on due to the onset of an activity, yet often ignoring those appliances which are turned on and indirectly or implicitly related to the activity (referred to as background appliances). The usage of these implicit appliances might be one of the main factors that cause the power consumption. Moreover, the accuracy of multi-user activity recognition is often lower than the one with single-user, which makes energy saving in multi-user environment more difficult. And the most important issue about energy saving is that most of prior works or systems seldom evaluate user comfort in a more quantifiable way to determine a more favorable energy saving policy.
To sum up, there are three main contributions in this work: (1) We associate power consumption level with a context of interest so that we can provide users more thorough feedbacks. More specifically, we will identify the power usage of those implicit appliances when a context is recognized. As a result, such a correlation between a context and its power consumption can be utilized to facilitate more spontaneous power saving. (2) In order to make multi-user activity recognition far less intractable, we reformulate this problem and take a group of users in the same area/zone as a whole to greatly reduce the complexity of data association inherent in a multi-user activity problem. (3) Using a composite and standard-based index to comprehensively evaluate real user comfort and to make appropriate energy saving policies without compromising both user comforts.

並列關鍵字

Pervasive Home Environment ； Activity Recognition ； Energy Saving ； Group Activity ； Dynamic Bayesian Network (DBN) ； User Comfort ； Context-Aware

參考文獻

[1]D. H. Wilson and C. G. Atkeson, "Simultaneous Tracking and Activity Recognition (STAR) Using Many Anonymous, Binary Sensors," presented at the the 3rd International Conference on Pervasive Computing, Munich, Germany, 2005.

[2]D. J. Cook and S. K. Das, "How Smart Are our Environments? An Updated Look at the State of the Art," Pervasive Mobile Computing, vol. 3, pp. 53-73, 2007.

[4]E. Kim, S. Helal, and D. Cook, "Human Activity Recognition and Pattern Discovery," IEEE Pervasive Computing, vol. 9, pp. 48-53, 2010.

[7]M. Brand, N. Oliver, and A. Pentland, "Coupled Hidden Markov Models for Complex Action Recognition," in IEEE Computer Society Conference on Computer Vision and Pattern Recognition, San Juan, Puerto Rico, pp. 994-999, 1997.

[8]L. McCowan, D. Gatica-Perez, S. Bengio, G. Lathoud, M. Barnard, and D. Zhang, "Automatic Analysis of Multimodal Group Actions in Meetings," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 27, pp. 305-317, 2005.