Summary: | Refrigeration control is usually realized by means of model-based feedback controllers, which requires high-computational load and time-consuming model identification efforts. The implementation of feedback control requires a compromise between performance and robust stability. Considering these difficulties, an online learning operation controller for one-stage refrigeration cycle is presented, which consists of two components: a model-based feedback component and a learning feedforward component. The feedback controller is utilized to guarantee robustness. Meanwhile, the optimized performance is reached by the learning feedforward controller including a one-hidden-layer structure with B-spline basis functions. The comparison results of benchmark problems validate the effectiveness of this strategy and show that a perfect tracking performance can still be achieved without extensive modeling.
|