Robust LPV control design based on HOSVD

Considering the trade-off between the accuracy and complexity of the linear parameter varying (LPV) modeling of hypersonic vehicles, a novel polytopic LPV modelling technique based on grid tensors and higher-order singular value decomposition (HOSVD) was developed. Firstly, through Jacobian linearization, the grid tensor description of a hypersonic vehicle with large flight envelopes was introduced, and a polytopic LPV model was constructed using HOSVD. The resulting model was described by a linear combination of vertices of linear time invariant (LTI) models and corresponding weighting functions. The discarded singular values were treated as a modeling error index. A robust variable gain controller which combines a robust controller and gain scheduling technique was then designed for the 6 degree-of-freedom nonlinear model of the hypersonic vehicle. Simulation results demonstrate that the proposed method not only reduces the computational complexity, but also guarantees the modeling accuracy. Furthermore, under the proposed controller, the closed loop system can track the attitude commands rapidly, and good stability and robustness are achieved.