Translation Registration Algorithm for Multi - source Time Series Data Based on the Sliding Window

Most of the existing integration and mining methods of multi-source time series assume that the data has been calibrated, which avoids the problem of time series registration. However, the inconsistent sampling frequency, different time references, and transmission delay make time and frequency mismatch in multi-source time series data, resulting in an unsatisfactory integration and mining and an even poorer effect than that of the single-source data. Time series registration has becoming an important research point in the pre-processing stage of time series integration and mining. To improve the overall integration performance and data mining accuracy, a new translation registration algorithm for multi-source time series data based on the sliding window was proposed. Firstly, the demands for time series big data mining was clarified, and the disadvantages of existing time series registration approaches was analyzed, then a structural model for the translation registration based on the sliding window was presented. Secondly, by using the slide window and the nearest neighbourhood principle, the real-time intervals offset was calculated, the low-frequency sampling of time data was translated to high-frequency, and then the proposed translation registration algorithm based on the sliding window was designed and implemented. Finally, the experimental analysis and verification of the typical algorithms for time series was completed. The experimental results demonstrated that the nearest neighbourhood method can provide a targeted time registration strategy, and the sliding window approach which does not dwell on the starting moment of time series can satisfy the registration requirements of dynamic time series big data. The experiments show that the proposed algorithm’s goodness-of-fit is 96.7%, and it is slightly influenced by missing data and is easy to operate. It cannot only improve the accuracy and timeliness of multi-source time series registration, but also has important theoretical and practical values to the dynamic time series data integration and mining.

[1]  Henry Leung,et al.  An exact maximum likelihood registration algorithm for data fusion , 1997, IEEE Trans. Signal Process..

[2]  Hava T. Siegelmann,et al.  Sensor registration using neural networks , 2000, IEEE Trans. Aerosp. Electron. Syst..

[3]  Ying Wah Teh,et al.  Time-series clustering - A decade review , 2015, Inf. Syst..

[4]  Prem Prakash Jayaraman,et al.  City Data Fusion: Sensor Data Fusion in the Internet of Things , 2015, Int. J. Distributed Syst. Technol..

[5]  Rob J Hyndman,et al.  Forecasting Time Series With Complex Seasonal Patterns Using Exponential Smoothing , 2011 .

[6]  S. Haykin,et al.  Cubature Kalman Filters , 2009, IEEE Transactions on Automatic Control.

[7]  Richard A. Davis,et al.  Introduction to time series and forecasting , 1998 .

[8]  T. R. Rice,et al.  Removal of alignment errors in an integrated system of two 3-D sensors , 1993 .

[9]  Kunikazu Kobayashi,et al.  Time series forecasting using a deep belief network with restricted Boltzmann machines , 2014, Neurocomputing.

[10]  Ali T. Alouani,et al.  On optimal synchronous and asynchronous track fusion , 1998 .

[11]  Amy Loutfi,et al.  A review of unsupervised feature learning and deep learning for time-series modeling , 2014, Pattern Recognit. Lett..

[12]  Tak-Chung Fu,et al.  A review on time series data mining , 2011, Eng. Appl. Artif. Intell..

[13]  YI Xiang-ju.  A Systematic Model of Organization Conflict Factors on Equipment Management , 2007 .

[14]  Liang Kai.  THE STUDY OF MULTI-SENSOR TIME REGISTRATION METHOD , 2006 .

[15]  W. Blair,et al.  Use of a kinematic constraint in tracking constant speed, maneuvering targets , 1993, IEEE Trans. Autom. Control..

[16]  Carlos Agón,et al.  Time-series data mining , 2012, CSUR.

[17]  M. Blanchette,et al.  Sensor alignment using the Earth-centered Earth-fixed (ECEF) coordinate system , 1996, Conference Record of The Thirtieth Asilomar Conference on Signals, Systems and Computers.

[18]  Wang Li-n.  Time Registration Algorithm Based on Adaptive Kalman Filtering , 2013 .

[19]  Nien Fan Zhang,et al.  Forecasting and time series analysis , 1976 .