Real Time Prediction Method of Sensor Output Time Series

Abstract:

In order to improve the real time prediction precision of sensor output time series, the predictable inner mechanism of time series is analyzed, and a method using wavelet filtering and neural network is proposed. Sensor output time series are first handled with wavelet filtering, and then predicted by neural network method. The proposed method can eliminate effect of measurement noise on prediction precision. Simulation experiment shows a higher prediction precision by the method. A new idea is given to increase prediction precision of sensor output time series by neural network-based methods.

• Periodical: Advanced Materials Research (Volumes 912-914)
• Pages: 1322-1326
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.912-914.1322
• Online since: April 2014
• Authors: Zhi Cheng Liu*
• Keywords: Neural Network (NN), Prediction Precision, Time Series, Wavelet Filtering

References

[1] Dongxiao Niu, Mian Xing. Systems Engineering-Theory & Practice, 5 (1999): 89-92.

[2] Xiaomu Ding, Shiqun Jin, Yetai Fei. Journal of Hefei University of Technology, 26 (2003): 1127-1130.

[3] Hang Luo, Houjun Wang, Bing Long. Computer Engineering and Applications, 8 (2008): 16-19.

[4] H. Leung, T. Lo, S. Wang. IEEE Transactions on Neural Networks, 5 (2001): 1163-1172.

[5] Ying Zhong, Bingwen Wang. Systems Engineering and Electronics, 4 (2004): 9-11.

[6] G. Yang, J. E. Meng. Fuzzy Sets and Systems, 2 (2005): 331-350.

[7] Zhicheng Liu, Xiangguang Chen, Li Bing. Journal of Data Acquisition & Processing, 2 (2007): 161-16.

[8] Jingxia Chen, Lideng Pan. Journal of Beijing University of Chemical Technology, 6 (2003): 84-87.