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

ListofFigures 5.22. Illustrationof the target temperaturecurve. . . . . . 161 5.23. Control performance ofPIDcontrolmethod (new cavity 3,12sources). . . . . . . . . . . . . . . . 161 5.24. Control performance ofPIDcontrolmethod (oldcavity3,36sources). . . . . . . . . . . . . . . . . 163 5.25. Control results of the linearMPC (a)and nonlinearMPC (b). . . . . . . . . . . . . . . . . . . . 164 5.26. Control ’accident’ in theheating process. . . . . . . 165 5.27. Control results ofLinearMPC with RLS (a) and RKF(b). . . . . . . . . . . . . . . . . . . . . . . . 167 5.28. Control results ofnonlinearMPC using8 heatingsources (newCA3). . . . . . . . . . . . . . . 169 5.29. Thesetupwith independentsmallworkpieces. . . . 169 5.30. Control performance ofPID, linear MPCand nonlinearMPC inthesecond setup. Dueto slightlydifferent duration in the three experiments, the comparisonofstandard deviation lasts until 2760s. . . . . . . . . . . . . . . . 172 5.31. Structureused in theNNC simulations (the NN estimatorwas trainedbased on experimental data fromthenew CA3 with8 heatingsources). . . . . . . . . . . . . . . . . . . . . . 173 5.32. Simulationresults ofNNC. . . . . . . . . . . . . . . 174 5.33. Control performance ofNNC (newCA3,8sources). 175 5.34. Control results ofNNCinitializedwith the sameweights of the controller trained in 5.33 (new CA3,8 sources). . . . . . . . . . . . . . . . . . . 176 5.35. LabVIEW program(steps) toobtain the coordinatesof the maximumtemperature. . . . . . . 177 5.36. Simulationresults of theQ(λ)basedRLCmethod. . 179 5.37. Control resultsof theQ(λ)basedRLC method in threeconsecutive trials. . . . . . . . . . . . . . . . 180 B.1. Errorflow inthe output layer. . . . . . . . . . . . . . 228 B.2. Errorflowsfromhiddenlayers to theoutput layer. . 230 xii . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .