Intelligent fault tolerance control using long short-term memory for efficient system performance under fault conditions

Fault-Tolerant Control (FTC) is a crucial field within control systems engineering that focuses on designing systems capable of maintaining desired performance and stability even in the presence of faults. This study introduces a data-driven fault-tolerant control system that enhances the operation of control systems in the presence of faults. The system is designed on a single Long Short-Term Memory (LSTM), which replaces the units responsible for diagnosis and control reconfiguration. The LSTM-FTC system does not require diagnostic and process models, which is a significant advantage over traditional model-based methods. The factory I/O is interfaced with MATLAB through the implementation of the digital twin idea, which allows for the simulation and validation of the suggested approaches. These approaches are then applied to an assembler case study that included both faultless and multiple faulty sensors. The training process reaches 6553 iterations with Root Mean Square Error (RMSE) equal to [Formula: see text] at six minutes and 17 s. The results of the simulation demonstrate the effectiveness of the proposed approaches. The accuracy of the system outputs in the faultless and worst-case scenarios are 92.81% and 67.16% respectively.