When using bp neural network deals with the data of the big numerical value like cost , using common methods of changing concealed node number etc . do n ' t improve the things that the simulated curve is low precision or transpires . so data processing course of network is investigated . a new data processing method is advanced to apply graphics zoom theory 在用bp神經(jīng)網(wǎng)絡(luò)進(jìn)行建模的過程中,發(fā)現(xiàn)對處理費用這種大數(shù)值數(shù)據(jù)時,采用通過改變隱層節(jié)點數(shù)等常用方法,都無法改善擬合的曲線精度低或出現(xiàn)發(fā)散這樣的情況,為此通過研究網(wǎng)絡(luò)的數(shù)據(jù)處理過程,提出了一種在網(wǎng)絡(luò)運(yùn)行前后應(yīng)用圖形圖像縮放原理來對數(shù)據(jù)進(jìn)行處理的新方法,實例證明了該方法的可行性和有效性。