A Posteriori Suboptimal Error Estimates in Control Problems

摘要

For the linear quadratic control problem, a lower bound for the performance index is established by simple methods. Using this bound, two important a posteriori error estimates are obtained, the first one measuring the deviation of the performance index, while the other is for the deviation of the state and control variables from the optimal solution. Continuous-time as well as discrete-time systems described by ordinary differential and difference equations are discussed. A numerical example is given for illustration.