Analysis and correction of the consistency problem in the calculation method for Sperling human vibration comfort evaluation index in railway vehicles

Abstract

The Sperling vibration comfort index is widely used to predict passengers’ riding experiences and evaluate railway vehicle systems’ service performance. This evaluation index can be calculated using either the 3^rd power integration and 10^th root operation (algorithm ①) or the 2^nd power integration and 6.67^th root operation (algorithm ②) of the weighted acceleration. Theoretically, the above two algorithms are equivalent when the signals are continuous in the time and frequency domains. However, the real measured signals are always discrete after the analog-to-digital conversion, which leads to significant inconsistencies in the index calculated value based on the two different algorithms mentioned above. Additionally, the calculation results based on algorithm ① are inconsistent when using time analysis fragments (TAF) with different truncation lengths. To address the above issues: Firstly, the inconsistent characteristics of the index value calculated using different algorithms and TAF truncation lengths are analyzed using real measured vibration acceleration data and virtual signals. Secondly, based on the fundamental theories of signal analysis and numerical analysis, the characteristics of the relationship between the index calculated value, the TAF truncation length, and the different algorithms are derived, revealing the inconsistent mechanism of the index calculated value. Finally, a correction method for the inconsistent index calculated value is proposed, and its effectiveness is verified based on real measured vibration acceleration data, to improve the stability and reliability of the human vibration comfort evaluation system for railway vehicles.