Experimental study of road identification by LSTM with application to adaptive suspension damping control. (1st September 2022)

• Record Type: Journal Article
• Authors: Liang, Guanqun; Zhao, Tong; Shangguan, Zhengwei; Li, Ningfei; Wu, Mingyu; Lyu, Jingcheng; Du, Yongchang; Wei, Yintao
• Is Part Of: Mechanical systems and signal processing. Volume 177(2022)
• Publication Date: 2022-09-01
• Subjects: Road identification -- Signal processing -- Suspension control -- CDC damper; Adam Adaptive optimization algorithm -- CAN Controller Area Network -- CDC Continuous Damping Control -- ECU Electronic Control Unit -- FC Fully Connected -- GPU Graphical Processing Unit -- IMU Inertial Measurement Unit -- LSTM Long Short-Term Memory -- LiDAR Light Detection and Ranging -- RMS Root Mean Square -- RNN Recurrent Neural Network; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2022.109197
• Languages: English
• ISSNs: 0888-3270

Abstract

Graphical abstract: Highlights: All study is based on and validated by proving ground experiment. Real time identification with acceleration by LSTM network is proposed. A section of serial acceleration signal without preprocessing is needed. Identification is robust to different velocity and damping with high accuracy. Damping is adaptive to current road conditions. Abstract: Road unevenness provides essential evidence for semi-active suspension control while it is difficult to estimate accurately in real time. This paper proposes a real-time method of identifying road unevenness with serial acceleration signals and unevenness-correlated adaptive suspension damping control. This new method uses Long Short-Term Memory (LSTM) network to identify the characteristics of the time domain signal and classify it into different unevenness classes. A proving ground experiment is designed and conducted to construct and validate the algorithm. The good estimation accuracy and robustness is realized at different vehicle velocities and suspension damping levels. Based on this real-time estimated road unevenness, the developed suspension controller can be enhanced by adaptive correlation of the comfort weight and adaptive damping coefficient with corresponding road unevenness. A calibration experiment is conducted to provide evidence for comfort weight selection. The validation experiment result proves that the adaptive damping control could substantially reduce RMS of body acceleration on different kinds of roads, demonstrating the effectiveness of the proposed new method.