Boundary and distributed control of the viscous Burgers equation (Q557741)

The author considers the standard form of Burgers' equation \(u_{t}=v u_{xx}-uu_{x}+f(x)\). The control \(w(t)\) is at first applied on the boundary, i.e. we have \(w(0,t)\) and \(w(1,t)\) acting as controls. The author uses two cases for conducting his analysis. The system is analyzed when the viscosity \(v\) is ``known'', and an entirely different analysis must be conducted when \(v\) is ``unknown'' because of the use of adaptive control. The fluid has no way of changing its behavior when we either know or do no not know what is its viscosity. But an adaptive response can use the information. In both cases \(v\) is assumed to be constant. Comment: In the case when \(v\) is variable we could rewrite Burgers' equation in the form \[ \Phi(u, u_x, u_t,v)= u_t+ u_x- vu_{xx}= f(x,t)_{x= 0,1}, \] and regard \(u\) as a function of \(v\). We can use the symbol \(s= s(x,t,v)\) for \(\partial u/\partial v\) to obtain a second-order nonlinear partial differential equation involving \(u\) and \(s\). Then much of the author's analysis can be continued for that case. For the case when \(v\) is known, the author uses a classical Lyapunov function: \[ V ={1\over 2} \int^1_0 u^2(x,t)\,dx. \] Then the author suggests a boundary control which makes \(\partial/\partial t(V(x,t))\) negative along the trajectory of the system. In the case when \(v\) is known the adaptive system derivation is slightly more complex. Then the total energy of the system is used as the Lyapunov function. A Galerkin method is used in numerical computations.