科技通报
科技通報
과기통보
Bulletin of Science and Technology
2015年
11期
211-214
,共4页
高层建筑结构设计%改进BP神经网络算法%自适应调整策略%误差修正%精度优化
高層建築結構設計%改進BP神經網絡算法%自適應調整策略%誤差脩正%精度優化
고층건축결구설계%개진BP신경망락산법%자괄응조정책략%오차수정%정도우화
high- rise building structure design%improved BP neural network algorithm%adaptive adjustment strategy%error correction%precision optimization
针对传统的BP神经网络算法在对高层建筑进行结构设计时还存在精度不高、误差较大等问题，本文提出了一种基于自适应和误差修正BP神经网络算法的高层建筑结构设计模型，该模型在BP神经网络算法的基础上，首先采用自适应调整策略对其网络模型进行优化，然后采用增加动量项、误差累积处理和陡度因子优化等误差修正策略提高原算法的训练精度。仿真试验结果表明，本文提出的基于自适应和误差修正BP神经网络算法的高层建筑结构设计模型相比较传统的BP神经网络算法精度要高，具有较好的鲁棒性。
針對傳統的BP神經網絡算法在對高層建築進行結構設計時還存在精度不高、誤差較大等問題，本文提齣瞭一種基于自適應和誤差脩正BP神經網絡算法的高層建築結構設計模型，該模型在BP神經網絡算法的基礎上，首先採用自適應調整策略對其網絡模型進行優化，然後採用增加動量項、誤差纍積處理和陡度因子優化等誤差脩正策略提高原算法的訓練精度。倣真試驗結果錶明，本文提齣的基于自適應和誤差脩正BP神經網絡算法的高層建築結構設計模型相比較傳統的BP神經網絡算法精度要高，具有較好的魯棒性。
침대전통적BP신경망락산법재대고층건축진행결구설계시환존재정도불고、오차교대등문제，본문제출료일충기우자괄응화오차수정BP신경망락산법적고층건축결구설계모형，해모형재BP신경망락산법적기출상，수선채용자괄응조정책략대기망락모형진행우화，연후채용증가동량항、오차루적처리화두도인자우화등오차수정책략제고원산법적훈련정도。방진시험결과표명，본문제출적기우자괄응화오차수정BP신경망락산법적고층건축결구설계모형상비교전통적BP신경망락산법정도요고，구유교호적로봉성。
According to the defects such as low accuracy and large error of the traditional BP neural network algorithm in the design of high-rise building structure, this paper presents a high-rise building structure design model based on BP neural network algorithm with adaptive and error correction. The model based on BP neural network algorithm, first network model is optimized by the adaptive adjustment strategy, and then the error correction strategies such as increasing momentum item and cumulative error handling and steepness factor optimization are used to improve the training precision of the original algorithm. Simulation test results show that, compared with the traditional BP neural network algorithm, the proposed high-rise building structure design model based on BP neural network algorithm with adaptive and error correction has higher precision and good robustness.
