Multi-scale direct learning control of linear time-varying high-order systems (Q1961193)

The authors propose a new so-called direct learning control scheme, which aims to handle learning control problems characterized by the occurrence of non-repeatable trajectories. They discuss a multi-scale approach in the sense that patterns with different shapes can be handled and learned, which was impossible for known single-scale approaches. They provide a detailed theoretical analysis as well as an illustrative example.