Seite - 99 - in Adaptive and Intelligent Temperature Control of Microwave Heating Systems with Multiple Sources

4.1. AdaptiveControl Thefitness functionfi isdefinedas fi= exp(−α ·Ji), 1≤ i≤Ng, whereα is a tuning constant. The value fi reflects how well the control sequence is fit for the optimization problem. After the fit- ness value is calculated for all individuals, the new generation of population will be selected. The principle of selection is defined as that for each old individual i, it has a probability of Pi to be selected as a new individual j in the new generation, where Pi is givenby[Whi94] Pi= fi∑n l=1fl , 1≤ i≤Ng. It is clear that the larger fi is, the more suitable this control se- quence isandthehigherselectedopportunity ithas. • Step3: Crossoverandmutation After the new generation is selected, the population randomly choose every two individuals into one pair and evolves these two individuals into new individuals by doing crossovers. In our case, the crossover occurs only for the first input vector and the rest of inputvectorswill replicate thefirstone, suchas Vci Vi(k)︷ ︸︸ ︷ 0100101011 Vi(k+1)︷ ︸︸ ︷ 0100101011. .. Vcj 10101110011010111001 . .. =⇒ 11001010111100101011. .. 10101010011010101001 . .. The length of crossover is predefined, but the starting point of crossovercouldbedeterminedrandomly. Aftercrossover, thenextstepismutation. ForeachbitofVci, ithas a tiny probability Pm to be switched to the other state (from 0 to 1 or from 1 to 0). The probability is defined as Pm= 1/(mp), where mp is the number of total bits in each individual. Similarly with the crossover part, mutation is also taken only for bits of the first inputvector. 99