Seite - 107 - in Adaptive and Intelligent Temperature Control of Microwave Heating Systems with Multiple Sources

4.2. IntelligentControl Plant (HEPHAISTOS) NN Controller NN Estimator Target Temperature Measured Temperature Control Input Estimated Dynamics Figure4.6. Semi-direct control structure. The blue dashed line represents two possible learningapproaches. The main advantage of the semi-direct control structure is that the overall weight update of the NN controller is significantly acceler- ated. The average weight update rate can be increased to once per control period or even higher. In addition, there is no extra waiting time spent on the states or outputs recovery. The NN controller us- ing one-measurement SPSA learning can quickly evolve from the ini- tialized state to a nearly optimal state, which guarantees acceptable control performance in the starting period. Defining the modulo op- erationas r= Mod(a,n) =a−nba n c, for a,n,r∈R, wherer is thereminderofadividedbyn. The complete procedures of the semi-direct control scheme is shown infigure 4.7. 4.2. IntelligentControl Intelligent control is a class of control methods that utilize ideas and approaches from the artificial intelligence or biological systems to solve the control problem [WSF94]. Technically, both aforementioned nonlinear GA control and NNC methods can be classified as intelli- gent control approaches. Nevertheless, in this dissertation intelligent 107