Fault Detection and Fault Tolerant Control of Pressure System by Reinforcement Learning Approach

In this paper, a reinforcement learning-based methodology is introduced to detect sensor fault and FTC in a pressure system. The methodology includes reinforcement learning skills to attain optimal performance and the computational time is also as compatible with the systems tracking. The method is deliberately designed to offer thoughtful yet workable solutions that support sustainability, safety, and dependability in presented attempt to have a balance between system performance and system complexity. By providing a thorough framework that maximizes performance of framework while lowering complexity and costs, the work provides intelligent ways of industrial solutions. A novel proportional integral(PI) controller augmented by reinforcement learning, especially for accurate fault detection (FD) as well as (FTC) in pressure system. The results are demonstrate the superiority of the presented controller over conventional methods of control system design in terms of accuracy, effectiveness, and usability. In this, a model-based environment approach is used to ensure robust fault detection, resulting in a quick and precise diagnostic technique for sensor faults. A model-based FTC method is used in the case that a sensor malfunctions, enabling the system to continue operating effectively in spite of disturbances. The efficiency of presented method is vividly demonstrated by simulation results using a pressure system subjected to sensor failures in the MATLAB environment. The results clearly demonstrated the benefits of proposed method, which qualify it for industrial use due to its enhanced accuracy, quicker operation, affordability, and simplified simplicity.