A wheel slip detection method of electric locomotive based on time-frequency analysis

Rapid and effective identification of wheel slip plays an important significance to improve traction force of electric locomotive. A novel detection method of wheel slip was studied in this paper. Firstly, the Hilbert energy spectrum of wheel speed was calculated with the Hilbert-Huang Transform method. The tendency of wheel slip was obtained with the energy spectrum change. Then the acceleration was calculated after filtering the wheel speed data. The wheel slip was detected using the tendency of wheel slip and acceleration combined with the analysis of time domain and frequency domain. Finally, a water spring wheel-slip test was conducted with a HXD2C-type electric locomotive to verify the effectiveness of the detection method.

I. Introduction

The transport policies of Chinese railway are high-speed and heavy-load. Therefore the power of the electric locomotives is increased greatly. The driving and braking force of an electric locomotive depends on the adhesion force between rail and wheel. If the traction force exceeds the maximum adhesion force, the slip phenomena will occur and the traction torque decrease sharply. Wheel-to-rail slipping degrades running performance and leads to a critical abrasion of wheel and rail. Therefore effective identification of wheel slip is essential to anti-slip control systems. Hybrid slip control methods are the most commonly used wheel slip protection in rail vehicles [1]–[4]. The slip velocity and acceleration are used to detect wheel slipping. If the slip velocity or acceleration exceeds the specific thresholds, the slip is detected. But it is difficult to determine the threshold if the rail conditions are different. The anti-slip/skid control system base on disturbance observer is used to suppress slip/skid phenomena^[5]–[6]. The adhesion force coefficient can be estimated by the disturbance observer and the slip can be known. The performance of the anti-slip schemes based on a disturbance observer is to a large extent affected by noises. Wheel slip detection with axle vibration is a novel detection technique^[7]. The methods detect the wheel slip indirectly from the wheel set dynamic changes which caused by the wheel sets operating in the unstable region of the non-linear contact characteristics as a consequence of wheel slip. But the variation of axle vibration should be studied further when the rail conditions are different. How to handle unknown conditions should be studied further. Song Wang^[8] proposed a multi-rate EKF algorithm including input and output algorithms for load torque estimation in the induction motor based on multi-rate control theory and the extended Kalman filter theory. The load torque estimation detection can be used to detect the wheel slip. Zhao and Liangan^[9] proposed a re-adhesion controller based on extended Kalman filter. The creep force and friction coefficient can be estimated by measuring the voltage, current and speed of the AC traction motor and using an extended Kalman filter.