Seite - 175 - in Adaptive and Intelligent Temperature Control of Microwave Heating Systems with Multiple Sources

5.3. ResultsofDifferentControlMethods ods. Compared with the PID controller, NNC is also much better be- causeof thebetter temperaturehomogeneity. 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 ° C ~ 5 . 5 ° C T i m e ( s ) T a r g e tT 1T 2T 3T 4T 5 DT1 = 5 ° C ~ 5 . 5 ° C 0 2 0 4 0 6 0 8 0 1 0 0 Figure5.33. ControlperformanceofNNC(newCA3,8sources). When the weights of the NN controller are properly initialized, this temperature lagphenomenoncanbesignificantlyreduced. Forexam- ple, the control results of another NNC experiment is shown in figure 5.34. In this experiment, the controller is initialized with the same weights as the trained controller in 5.33. In other words, after the ex- perimentshowninfigure 5.33, thesamecontroller isusedagaininthe sameenvironment todealwith thesamecontrol task. From the results in figure 5.34, a big improvement can be observed is that thecontrolledtemperatures followthetarget temperatureclosely almost from the beginning of the control process. There is no tem- perature overshoot and the temperature window is smaller than the resultsshowninfigure 5.33. Thisperformanceimprovementisdueto the appropriate initialization of the NN controller. A good initializa- tion would speed up the entire controller update and lead to a more reliablecontrolperformance in thebeginningperiod. From the control results shown by these pictures, all three adaptive control methods, including the linear MPC, the nonlinear MPC and 175