Page - 86 - in Adaptive and Intelligent Temperature Control of Microwave Heating Systems with Multiple Sources

4. ControlSystemDesign At time k FuturePast Prediction Horizon kk-1 k+1 . . . k+pk+2 Target output Real output Predicted output Real control Predicted control Time u(k) Figure4.1. PrincipleofMPCcontrolalgorithms. • Apply the first control action of the sequence as the real control inputanddroptherest. • Repeataboveproceduresateachtimeinstants. The idea of MPC is illustrated by figure 4.1. Further elaborations of MPCcanbefoundin[CBCB04]and[MHL99]. MPC has natural advantages over other control methods (mentioned in thebeginningof thischapter) indealingwithcomplexsystems like HEPHAISTOS: • Theprinciple is intuitiveandeasy implement. • The state-space model formed control law is suitable when the in- putandoutputdimensionsare large. • A great variety of dynamics including delay times and external disturbancescanbeeasilyhandled. • It involvespredictionsforthefuturebehaviorswhichareusefulfor trackingtime-varyingtrajectories. • It can easily be extended to treat different kinds of input and out- putconstraints. 86