Seite - 176 - in Adaptive and Intelligent Temperature Control of Microwave Heating Systems with Multiple Sources

5. ExperimentalResults 0 5 0 0 1 0 0 0 1 5 0 0 2 0 0 0 2 5 0 02 0 4 0 6 0 8 0 1 0 0 1 2 0 DT2 = 5 . 5 ° C ~ 6 . 0 ° C T i m e ( s ) T a r g e tT 1T 2T 3T 4T 5 DT1 = 2 . 8 ° C ~ 3 . 5 ° C 0 2 0 4 0 6 0 8 0 1 0 0 Figure5.34. Control results of NNC initialized with the same weights of the controller trainedin 5.33 (newCA3,8sources). the NNC methods, achieve the same level of control performance, regarding the final temperature window and the temperature homo- geneity aspects. Although there is no essential difference among the controlresults,differentpropertiescanbeobservedforindividualcon- trolmethods. ThelinearMPCmethodisthemostrobustandeasilyap- plicable control method. Benefiting from the well established system identification method (which estimatesAandB simultaneously) and (partially analytical) control solution, the computation requirements in the linear MPC is the lowest among the three methods. Therefore it is able to control high input/output dimensional systems with a guaranteed control performance. Compared with the linear MPC, the performance of the nonlinear MPC is partly influenced by the input dimension. In order to get a more reliable control performance, the number of heating sources has to be limited. However, on the other hand,aslongasthecontroldiversityandtheperformancecanbeguar- anteed, there is no need to use more heating sources. In addition, the nonlinear MPC could achieve a slightly better control result than the linear MPC method in the second setup, which means it is preferred tobeapplied insituationswithmorestablecoolingeffects. 176