ANN and Astrom's Smith Predictor‐Based PI Control Design for Semi‐Markovian Jump Systems With Input Time‐Delay

ABSTRACT

This comprehensive study elucidates the intricacies of tracking control in semi‐Markovian jump systems. It meticulously examines the challenges arising from input delay and disturbances, offering a thorough understanding of their impact on system performance. An artificial neural network based on the Astrom modified Smith predictor scheme and a proportional‐integral control structure ensures accurate tracking capability and high‐precision disturbance reduction. The classic Smith predictor setup is integrated with an artificial neural network‐based disturbance estimator, which provides sufficient rectification of both delay and external disturbances. Also, the principles of Lyapunov stability are complemented by matrix inequalities, which serve as mathematical tools to rigorously analyze and guarantee the stability of semi‐Markovian jump systems. Additionally, the stated matrix inequalities are solved concurrently with the calculation of the controller parameters. Subsequently, the numerical examples provide a detailed insight into how the proposed control method operates in different contexts, highlighting its effectiveness across a range of conditions. Comparative analysis further strengthens these findings by juxtaposing the performance of the proposed method against existing approaches.