Abstract
Piecewise linear representation (PLR) techniques have been widely adopted to recharacterize high-dimensionality time series data in numerous fields for purposes of dimensionality reduction, fluctuation filtering, and overall trend extraction. However, this technique has not yet been applied to pressure waves of high-speed trains (HSTs). This study therefore introduced PLR techniques to recharacterize typical high-dimensionality pressure waves for the first time. A well-performing PLR algorithm based on perceptually important points (PIPs) was specifically designed for pressure waves of HSTs. The results reveal that the measurement methods of data point importance and assessment methods for segmentation errors, particularly the former, have impacts on identification priority and even the final result of PIPs of pressure waves. The PLR_PIP algorithm using vertical distance as the measurement of data point importance (PLR_PIP_VD) achieves a more reasonable PLR of pressure waves compared to those of the Euclidean distance and orthogonal distance. Through comparisons among cumulative error, average error, and maximum error, the PLR_PIP_VD algorithm employing cumulative error as the assessment method of segmentation errors accomplished a preferable PLR of pressure wave. The design of the PLR_PIP algorithm applicable to the PLR of pressure waves was finally proposed. This provides a novel processing method for pressure waves of HSTs.
Original language | English |
---|---|
Pages (from-to) | 2411-2426 |
Number of pages | 16 |
Journal | Journal of Central South University |
Volume | 30 |
Issue number | 7 |
DOIs | |
Publication status | Published - 8 Aug 2023 |
Keywords
- algorithm design
- high-speed train
- piecewise linear representation
- pressure wave
- time series
- tunnel
ASJC Scopus subject areas
- General Engineering
- Metals and Alloys