3D-printed scale model for detection of railway wheel flats using augmented vibration data from axle box

Abstract

As data-driven methods for defect detection become more prevalent in the railway industry, the demand for high-quality data continues to grow. However, field experiments are often time-consuming and constrained by practical limitations. This study introduces a methodology that uses Fused Deposition Modeling (FDM) 3D printing to develop a scale model for simulating wheel flat-induced vibrations, combined with a Long Short-Term Memory (LSTM)-based generative model to produce synthetic vibration data. This approach improves data quality by enhancing quantity, variety, and velocity, while increasing data volume and reducing the need for extensive experimental testing. The LSTM-based model generates realistic synthetic data, minimizing reliance on labor-intensive field experiments and offering a broader spectrum of defect scenarios. By accelerating the data generation process, this method provides an effective alternative in a laboratory setting and contributes to foundational research aimed at improving defect detection and maintenance processes in the railway industry.