D. Augustin et H. Maurer, Second order sufficient conditions and sensitivity analysis for optimal multiprocess control problems, CONTROL CYB, 29(1), 2000, pp. 11-31
Second order sufficient optimality conditions (SSC) are derived for optimal
multiprocess control problems. For that purpose the multiprocess control p
roblem is transformed into a single stage control problem with augmented st
ate variables which comprise the state variables of all individual stages a
s well as the switching times as choice variables. This transformation allo
ws to apply the known SSC for single stage control problems. A numerical te
st of SSC involves the solution of an associated Riccati equation together
with boundary conditions adapted to the multiprocess. Sensitivity analysis
of parametric multiprocess problems can be based on SSC. A numerical exampl
e of the optimal two-stage control of a robot illustrates both SSC and sens
itivity analysis.