TY - GEN
T1 - Intelligent analysis and evaluation of orbital quality state based on machine learning and decision-level fusion predictive analysis
AU - Jiaqi, Chen
AU - Jianjun, Qu
AU - Kunzhen, Liu
AU - Zimo, Song
AU - Pan, Liu
AU - Liang, Wang
N1 - Full text of this publication does not contain sufficient affiliation information. With consent from the author(s) concerned, the Research Unit(s) information for this record is based on the existing academic department affiliation of the author(s).
PY - 2025
Y1 - 2025
N2 - The Rail Quality Index (TQI) is an important indicator for assessing the state of railway tracks and directly affects the safety and efficiency of railway operations. Traditional TQI assessment methods rely on human experience, and existing models are not capable enough to handle the demands of complex railway data, high-precision predictions, and the analysis of various influencing factors. Based on the various experimental data obtained from the railway construction site, this paper first builds a machine learning model, uses the experimental data for model training and prediction, evaluates and analyzes the prediction results, compares the model performance, and combines the SHAP theory to analyze and calculate the shaply weight values of each influencing factor as a preliminary analysis. Further, based on Dempster-Shafer Theory of Evidence, the shaply values obtained from each model are fused at the decision level, and the weights of each influencing factor after fusion are analyzed as the final result. This method enables efficient analysis and prediction of railway TQI data, conducts weight analysis on various TQI influencing factors, and provides new methods and ideas for artificial intelligence to predict TQI analysis influencing factors. © COPYRIGHT SPIE.
AB - The Rail Quality Index (TQI) is an important indicator for assessing the state of railway tracks and directly affects the safety and efficiency of railway operations. Traditional TQI assessment methods rely on human experience, and existing models are not capable enough to handle the demands of complex railway data, high-precision predictions, and the analysis of various influencing factors. Based on the various experimental data obtained from the railway construction site, this paper first builds a machine learning model, uses the experimental data for model training and prediction, evaluates and analyzes the prediction results, compares the model performance, and combines the SHAP theory to analyze and calculate the shaply weight values of each influencing factor as a preliminary analysis. Further, based on Dempster-Shafer Theory of Evidence, the shaply values obtained from each model are fused at the decision level, and the weights of each influencing factor after fusion are analyzed as the final result. This method enables efficient analysis and prediction of railway TQI data, conducts weight analysis on various TQI influencing factors, and provides new methods and ideas for artificial intelligence to predict TQI analysis influencing factors. © COPYRIGHT SPIE.
KW - decision-level fusion
KW - Machine learning
KW - SHAP
KW - Track Quality Index (TQI)
UR - http://www.scopus.com/inward/record.url?scp=105013060129&partnerID=8YFLogxK
UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-105013060129&origin=recordpage
U2 - 10.1117/12.3070966
DO - 10.1117/12.3070966
M3 - RGC 32 - Refereed conference paper (with host publication)
SN - 9781510693166
VL - 13685
T3 - Proceedings of SPIE - The International Society for Optical Engineering
BT - Proceedings of SPIE
A2 - Wang, Xingjun
A2 - Liang, Wang
PB - SPIE
T2 - 8th Global Intelligent Industry Conference (GIIC 2025)
Y2 - 29 March 2025 through 31 March 2025
ER -