Model reference adaptive controller with interpretable fuzzy rules for linear MIMO systems (Q6595346)

The paper starts from the model of a linear control system with fuzzy controller \N\[\N\displaylines{\displaystyle{\dot{x}=A_lx + B_lu(t)\;,\;u_l(t) = u(t) - L\sigma(t)}\cr \displaystyle{u(t) = f(x(t),t)\;,\;\sigma(t)=Hx(t)}\cr \displaystyle{A_l=A+BLH\;,\;L=\operatorname{diag}\{\beta_1,\ldots,\beta_m\}}\cr \displaystyle{(K_j\sigma_j - u_{l,j})u_{l,j}\geq 0\;,\;j=1,\ldots,m}} \N\]\Nwhere the sector condition -- the last of the equations above -- describes the fuzzy controller. An absolute stability analysis is emphasized. Next, the parameter adaptation i.e. their modification according to a standard gradient rule to minimize a certain function of the error between the controlled system dynamics and its model is presented. An illustrating example is given.