How does learning control use a mathematical model to improve its operation?

Learning control systems utilize their computational ability to develop and refine representations of the mathematical model of the process they are managing. By continuously building these internal models, the system gains new knowledge about the dynamics of the plant it controls.1

This capability allows the learning control system to modify its own operation to take advantage of this newly acquired information, effectively improving its performance over time. As an advanced evolution of the adaptive control concept, a learning controller does not merely react to current states but actively updates its understanding of the system's behavior to achieve optimal results.1