Page - 66 - in Adaptive and Intelligent Temperature Control of Microwave Heating Systems with Multiple Sources

3. ModelingMicrowaveHeating timator is introduced in this section, and the NN controller will be addressed inchapter 4. 3.3.1. IntroductionofNeuralNetwork Theneuralnetwork(NN)structurewasoriginallyinspiredbythecon- cept of biological neural networks, and then gradually extended and extensivelyutilizedindifferentareas. Incontrolengineering, it iscon- sidered as a powerful alternative to conventional modeling and con- trolmethods [LV09]. Ithasbeenprovenin[HSW89] that,with theap- propriate network architecture, a multilayer feedforward neural net- work is a universal approximator for all kinds of linear and nonlinear dynamics. Due to this universal approximation ability, NN has be- come one of the most popular system identification methods that are applied invariousapplications. N1,1 N1,2 N2,1 N2,2 N2,3 N3,1 Input Layer Hidden Layer Output Layer W11,1 W21,1 Y1 u1 W13,3 Bias = 1 N3,2 Y2 W22,4 Bias = 1 u2 Figure3.4. Neuralnetworkstructure A basic NN structure is illustrated in figure 3.4. Normally it con- sists of several elements such as nodes, synaptic links and weights. A 66