Crossover is implemented using arithmetic crossover operator . then unsymmetrical mutation is conducted using the new mutation operator which can expand the scope of chromosome gene value , at the same time , the chromosome with the highest fitness values are retained for each iteration . a lot of experiments are implemented to obtain the optimized initial weighs and bias 生成了三維矩陣表示的染色體,進(jìn)行了聯(lián)賽選擇,利用算術(shù)交叉算子進(jìn)行了交叉運(yùn)算,利用構(gòu)造的新的變異算子,進(jìn)行了非均勻變異,同時(shí)保留了每次進(jìn)化運(yùn)算后最優(yōu)的適應(yīng)值,通過(guò)大量實(shí)驗(yàn),實(shí)現(xiàn)了遺傳算法優(yōu)化bp網(wǎng)絡(luò)的初始權(quán)值和閾值的目的。
On the base of analysing the shortcoming of genetic algotithms , three improved techniques for genetic algorithms are bring forward in this paper : fuzzy penalty fitness function , random dislocation arithmetic crossover , fuzzy parameter adjust policy , which improve genetic algorithms capability of global convergence and convergent speed . at the same time , the improved genetic algorithms are applied to nonlinear mixed integer problems and complex nonlinear function optimization 在分析實(shí)數(shù)型遺傳算法不足的基礎(chǔ)上,本文研究了遺傳算法的關(guān)鍵技術(shù),分別提出了模糊懲罰評(píng)價(jià)函數(shù)、隨機(jī)錯(cuò)位算術(shù)雜交算子、模糊自適應(yīng)參數(shù)控制等改進(jìn)技術(shù),以提高遺傳算法的全局收斂性和收斂速度,并應(yīng)用于求解非線(xiàn)性混合整數(shù)規(guī)劃問(wèn)題和復(fù)雜高維的函數(shù)優(yōu)化問(wèn)題。