Multi factor prediction model of highway electromechanical equipment faults based on panel data model
QIN Yu, YU Quan, REN Guang li

（Beijing University of Technology, Beijing Collaborative Innovation Center for Metropolitan 
Transportation, Beijing 100124, China）

Abstract: Aiming at the problems caused by many factors in the mechanical and electrical equipment failure, a panel data model was developed. The electromechanical faults represented the response variable, and traffic flow, temperature difference, relative humidity, and wind speed were selected as explanatory variables. The data was based on an investigation of electromechanical equipment faults, and the abovementioned environmental variables were collected for six highways in Beijing during 20122013. Panel data models for mixed regression models with individual fixed and random effects were established. A unit root test and cointegration test were performed on the data sequence. Finally, theFtest and the Hausmantest were used to compare the three prediction models and determine the optimal model. The results indicate that the individual fixed effects model is superior, positive effects of temperature, relative humidity, and wind speed on the electromechanical equipment faults are significant. Traffic flow has a significant negative effect on the electromechanical equipment faults. This results can provide theoretical support for fault prediction and inform needs for preventive maintenance of the highway electromechanical equipment.

Key words: highway; mechanical and electrical equipment equipment; failure; panel data; multiple factors; model