Digital Twin Unmanned Health Monitoring System for Reciprocating Industrial Equipment Using Deep Learning

Reciprocating industrial equipment is a crucial component of the industrial equipment, and unmanned health monitoring is the linchpin for ensuring its efficient operation. This paper proposes a digital twin unmanned health monitoring system for reciprocating industrial equipment utilizing deep learning for unmanned health monitoring. Initially, monitoring data are expanded through the integration of digital twin, simulation technology, and gradient penalized conditional wasserstein generative adversarial network. Subsequently, the accuracy of monitoring is enhanced by improving classification and regression trees, and unmanned monitoring is achieved by developing applications. Finally, the efficacy of the proposed method is corroborated using a reciprocating plunger experimental machine. The experimental results show that the digital twin model exhibits exceptional fidelity. Moreover, gradient penalized conditional wasserstein generative adversarial network learns the distribution of the simulated samples, thereby enhancing the robustness and the generalization ability of monitoring model while addressing the issue of limited or difficult to obtain field monitoring data. The improvement of the classification and regression trees led to a significant increase in the accuracy of the monitoring model. Furthermore, the mean square error and the mean absolute percentage error are reduced to some extent, which validates the effectiveness of the proposed method.