Data compression approach for the home energy management system. (1st August 2019)

• Record Type: Journal Article
• Authors: Jia, Kunqi; Guo, Ge; Xiao, Jucheng; Zhou, Huan; Wang, Zhihua; He, Guangyu
• Is Part Of: Applied energy. Volume 247(2019)
• Page Start: 643
• Page End: 656
• Publication Date: 2019-08-01
• Subjects: Energy-cyber-physical system -- Online data compression -- Home energy management system -- Data mining -- Pattern discovery -- Similarity metric; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2019.04.078
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: Use extracted patterns to compress data with linear time complexity. Propose a pattern discovery method for pattern-based data compression. Outperform state-of-art data compression algorithms on real datasets tests. Abstract: As a typical energy-cyber-physical system (e-CPS), home energy management system (HEMS) plays a critical role in power systems by accommodating higher levels of renewable generation, reducing power costs, and decreasing consumer energy bills. HEMS can help understand the home appliances energy use and learn the users' preference so as to optimize home appliances operation and achieve higher energy efficiency. HEMS needs massive historical and real-time data for the above applications. Since HEMS is always based on a wireless sensor network, a more effective online data compression approach is necessary. The efficient data compression methods can not only relieve data transmission pressure and reduce data storage overhead, but also enhance data analysis efficiency. This paper proposes an online pattern-based data compression approach for the data generated by home appliances. The proposed approach first discovers the patterns of the time series data and then utilizes these patterns for the online data compression. The pattern discovery method in the proposed approach includes an online adaptive segmenting algorithm with incremental processing technique and a similarity metric based on piecewise statistic distance. The key issues of parameter selection and data reconstruction are also presented. Real-world common home appliance datasets are employed for comparing the performance of the proposed approach with those of six state-of-the-art algorithms. The experimental results demonstrate the outperformance of the proposed approach. Further complexity analysis shows that the proposed approach has linear time complexity. To the best of our knowledge, this is the first paper that performs online data compression based on the extracted patterns of the time series.