Linear combination rule in genetic algorithm for optimizationof finite impulse response neural network to predict naturalchaotic time series

摘要

A finite impulse response neural network, with tap delay lines after each neuron in hiddenlayer, is used. Genetic algorithm with arithmetic decimal crossover and Roulette selectionwith normal probability mutation method with linear combination rule is used for optimi-zation of FIR neural network. The method is applied for prediction of several important andbenchmarks chaotic time series such as: geomagnetic activity index natural time series andfamous Mackey-Glass time series. The results of simulations shows that applying dynamicneural models for modeling of highly nonlinear chaotic systems is more satisfactory withrespect to feed forward neural networks. Likewise, global optimization method such asgenetic algorithm is more efficient in comparison of nonlinear gradient based optimizationmethods like momentum term, conjugate gradient.