# Neural network for identification A three input

Fig. 14. Flowchart for determining the optimal value [27].Figure optionsDownload full-size imageDownload as PowerPoint slide

In order to use the genetic algorithm, an “individual” for Nanaomycin A should be defined as the first step. Each individual represents a candidate for the optimal value. Since the optimal value is the 8-step ON–OFF intervals of watering, an individual can be given by the 8-step ON–OFF intervals of watering T1, T2,…, T16 . They were all coded as 6-bit binary strings, which gives numerical values between 000000 (=20−1=0) and 111111 (=26−1=63 in decimal). The amount of watering as a value from 0 to 63 by using 6-bit binary strings. For optimization, however, the watering duration was restricted to 3600 s in order to save water and fit the real system (Eq. (18)). Individual i=Ti1,Ti2,??,Ti16= 000000,111000,??,010101 .

A set of individuals is called a “population”. They evolve toward better solutions. GA work with a population involving many individuals. The population size varies according to the use of genetic operations with a smaller population size tending to converge to the local optima. Local optima mean local optimization, not a global one. When an objective function is characterized by a complex function, a poor search technique sometimes falls into the local optima. A local optima does not allow obtaining a global (true) optimal solution.