Kafa W., Gavrilov D.A., Buzdin V.E.,
Tatarinova E.A., Fateev A.S., Merkelov A.A.,
Sichkar D.S., Potkin O.A., Murkhizh Yazan
This review delves into the pivotal role of Model Predictive Control (MPC) based neural networks across diverse control
domains, highlighting its effectiveness in optimizing complex systems while acknowledging inherent challenges. While not
delving into specifi c mathematical details, the review explores the innovative integration of neural networks with MPC, presenting advancements in system modelling, optimization, and robustness. Despite challenges such as hardware constraints
and data requirements, the combined application of neural networks offers promising solutions. Emerging trends, including
advanced network architectures and real-time optimization, signify a promising future. Practical considerations for implementation underscore the need for a holistic approach, ensuring adaptability, safety, and effi ciency. As researchers address
challenges and explore trends while committing to the practical considerations we highlighted, we believe that the integration
of neural networks with MPC will continue to shape a promising landscape for control systems.
Keywords: Model Predictive Control, Neural Networks, Control theory, Optimization, Modelling, Model Learning, Controller Imitation.
DOI: 10.25791/asu.8.2024.1525
Pp. 20-35. |
