Digital twin-driven water-wave information transmission and recurrent acceleration network for remaining useful life prediction of gear box

Abstract

Abstract Traditional methods for predicting the remaining useful life (RUL) of the gear box fail to consider the relationship between physical world data and virtual world data, leading to low prediction accuracy that can disrupt the normal functioning of robot. To tackle this problem, this paper introduces a RUL prediction method that combines digital twin DT with water-wave information transmission and recurrent acceleration network (DT-WITRAN). This approach leverages the high simulation capabilities of DT along with the robust data processing features of WITRAN. Initially, this paper creates a DT system for analysing the lifespan characteristics of the gear box. The DT system simulates the operating conditions under various scenarios, enabling effective monitoring and awareness of the gear box system. Once the DT system is operational, this paper can derive the theoretical RUL value. Subsequently, this paper uses the experimental data to train the WITRAN model, which outputs the predicted RUL value. Finally, the central particle swarm optimization (CPSO) algorithm merges the theoretical RUL value from the DT with the predicted RUL value from WITRAN. The experimental result indicates that the prediction accuracy of the DT-WITRAN model surpasses 99.3%, thereby enhancing the capability and reliability of the RUL prediction.