Dynamic identification of longitudinal impulse of heavy-haul locomotive: an effective intelligent method for comprehensive index

Abstract Longitudinal impulse could seriously threaten the heavy-haul train operational safety and stability. Traditional longitudinal impulse level assessment could be achieved through coupler force, coupler yaw angle (CYA) and longitudinal acceleration of carbody (LAC). However, the traditional measurements are hampered by the harsh service conditions and multiple evaluation indicators have troubled the accurate understanding of on-site personnel. Therefore, it is difficult to maintain the long-term stability of train running monitoring. To address this issue, a comprehensive evaluation index of longitudinal impulse (CEILI) and a new quantitative identification method of the CEILI are proposed in this paper. First, a novel index called CEILI for heavy-haul trains is proposed to accurately identify the longitudinal impulse level, based on the analysis of extensive measured data and entropy theory. It not only incorporates the quantitative analysis of traditional coupler force and CYA but facilitates the measurement in engineering applications. Subsequently, the convolution fusion module along the channel is employed to fuse the multi-channel data of relative displacement of car body with frame and LAC respectively. Leveraging the fused data, an adaptive multi-scale convolutional network is constructed to extract high-level abstract features. Finally, the extracted features are input into the extreme learning machine module to identify the CEILI. Both dynamics simulation and field test results show that the proposed method can effectively identify the CEILI under different level of longitudinal impulse.