Advanced digital twins for conditions monitoring, examinations, diagnosis and predictive remaining lifecycles based Artificial Intelligence

Abstract

Digital Twins (DTs) continue to evolve; however, there is a lack of standardisations, DTs not compliant with the description in the literature, lack of physical design background, lifecycles still based on estimates or ideal scenarios, no attention to the simulation during systems operation, and lack of integration between AI and DTs. This research provides a comprehensive systematic review methodology for DTs while incorporating the science mapping methods. This research deeply examines DTs’ concepts, maturity, creation, values, applications, techniques, and technology to identify and create the most suitable way to implement DTs in the Predictive Maintenance (PdM) of suspension systems. This research proposes a novel concept to conceptualise DTs: “Four dimensions digital replica that continuously simulates the entire behaviour of anything”. The hypothesis of the four dimensions presented is computationally fast, easy to implement and cost-effective. This research uses physics-based simulation models and the DT concept to assess the PdM and forecast the Remaining Useful Life (RUL) of primary springs used in suspension systems. The Euler technique is employed with DTs to solidify the concept of DTs based on physics models and predict springs’ conditions in Current Real Time (CRT) as a novel method and unique technique to these primary springs. This method simulates physics-based models to monitor and predict spring conditions without intrusive methods. This research decreases the dimensions of the DTs modelling from five to four (Physical, Digital, Connection, and CRT), reducing the systems’ complexity and cost. This research reduces the predetermined design average load compared to the regular simulation and experimental results by 35.7 %. This research increases the average lifecycles of coil springs by 12 and 9 times more than the simulated and experimented results. This research increases the average lifecycles of the coil springs by 19.7% compared to the wireless DT model results. Artificial Intelligence (AI) is then integrated with the DTs to maximise efficiency, improve the performance of DTs models, and classify the types of faults that occur with the coil springs. A validation model is proposed and shows a prediction error of less than 0.0001 % and classifying the system’s defects with regression accuracy throughout training, validation, and testing were 0.99997, 0.99954, and 0.99931, respectively, for a total accuracy of 0.99986