基于混合改进的遗传算法-误差反向传播多层前馈神经网络(GA-BP神经网络)算法预测了在不同中主应力比(0、0.25、0.5、0.75、1)与主应力轴偏转角度(0°、15°、30°、45°、60°、75°、90°)的应力路径下的应力分量-应变发展规律。将预测数据与实测数据相对比,多轮印证了数据回归预测模型在剪切特性预测的实用性,并且利用该网络反演不同工况下的扭剪应力-应变发展规律。探讨了训练集数据等对网络识别精度的影响,证明了改进的GA-BP神经网络的预测精度较原始BP神经网络与常规GA-BP神经网络的普遍更高。最后通过两个算例对模型的可行性和精度进行了验证。
Based on the hybrid improved Genetic Algorithm-Error Backpropagation Multilayer Feedforward Neural Network (GA-BP Neural Network) algorithm, stress component-strain development law under the stress path of different medium to principal stress ratios (0, 0.25, 0.5, 0.75, and 1) and angles of deflection of the principal stress axis (0°, 15°, 30°, 45°, 60°, 75°, and 90°). By comparing the predicted data with the measured data, the practicality of the data regression prediction model in shear property prediction has been verified through multiple rounds. Moreover, this network is utilized to inversely analyze the development regularities of torsional shear stress-strain under different working conditions. The influence of training set data on network recognition accuracy is discussed, demonstrating that the improved GA-BP neural network generally achieves higher prediction accuracy compared to the original BP neural network and conventional GA-BP neural network. Finally, the feasibility and accuracy of the model are validated through two numerical examples.
[1] 乔金丽,范永利,刘波,等.基于改进BP网络的盾构隧道开挖地表沉降预测[J].地下空间与工程学报,2012,8(2):352-357,374.
[2] 易黄智,高飞.基于GA-BP神经网络的地铁变形预测模型[J].合肥工业大学学报(自然科学版),2021,44(11):1513-1517.
[3] 李洛宾,龚晓南,甘晓露,等.基于循环神经网络的盾构隧道引发地面最大沉降预测[J].土木工程学报,2020,53(增1):13-19.
[4] 马祥森,史治宇.基于GA-BP神经网络的结构损伤位置识别[J].振动工程学报,2004(4):81-84.
[5] Leila N,Shokrollah Z,Hossein N M.Intelligent identification of soil and operation parameters in mechanised tunnelling by a hybrid model of artificial neural network-genetic algorithm(case study:Tabriz Metro Line 2)[J].Civil Engineering and Environmental Systems,2022,39(4):287-308.
[6] 胡少伟,李原昊,单常喜,等.基于改进的PSO-BP神经网络的边坡稳定性研究[J].防灾减灾工程学报,2023,43(4):854-861.
[7] Tashayo B,Behzadafshar K,Tehrani S M,et al.Feasibility of imperialist competitive algorithm to predict the surface settlement induced by tunneling[J].Engineering with Computers,2019,35(3):917-923.
[8] 石开荣,吉古雪梅,姜正荣.基于PGSA的B样条蒙皮自由曲面薄壳结构形态优化方法研究[J].建筑结构学报,2024,45(10):158-169.
[9] 朱一凯,陈安妮,余哲帆,等.采用BP神经网络优化的振动信号压缩感知方法[J].振动工程学报,2023,36(5):1234-1243.
[10] Li Y,Zheng S,Dong L,et al.Mechanical properties of corroded steel bars on the tensile and buckling considering stirrup constraints[J].Construction and Building Materials,2024,417:135181.
[11] Koçer M,Öztürk M,Arslan H M.Determination of moment,shear and ductility capacities of spiral columns using an artificial neural network[J].Journal of Building Engineering,2019,26:100878.
[12] 范厚明,徐振林,李阳,等.混合遗传算法求解多中心联合配送路径问题[J].上海交通大学学报,2019,53(8):1000-1009.
[13] Negash B A,Wonhee Y,Le J,et al .Parameter Identification of Bouc-Wen Model for Magnetorheological(MR)Fluid Damper By A Novel Genetic Algorithm[J].Advances in Mechanical Engineering,2020.12(8).
[14] 陈长征,王楠.遗传算法中交叉和变异概率选择的自适应方法及作用机理[J].控制理论与应用,2002(1):41-43.
[15] Chen Z Y,Liu L Z,He J Y,et al.A Genetic Algorithm Based Framework for Local Search Algorithms for Distributed Constraint Optimization Problems[J].Autonomous Agents and Multi-Agent Systems,2020,34(2):1-31.
[16] 储琴,吕继东.机器人运动学参数标定的改进麻雀搜索算法[J].机械设计与研究,2022,38(3):52-55,59.
[17] Yang X S,Deb S .Cuckoo Search for Inverse Problems and Topology Optimization[J].Advances in Intelligent Systems and Computing,2013,174:291-295.
[18] De Jong K A.An Analysis of the Behavior of a Class of Genetic Adaptive Systems[D].Michigan:University of Michigan,1975.
[19] Rudolph G.Convergence properties of canonical genetic algorithms[J].IEEE Transaction on Neural Networks,1994,5(1):96-101.
[20] 王睢.考虑主应力轴旋转的Q2黄土静动力变形行为及本构关系研究[D].重庆:重庆大学,2018.
[21] 王睢,钟祖良,赵鹏,等.考虑主应力方向的Q2原状黄土定向剪切试验研究[J].地下空间与工程学报,2023,19(4):1160-1168.
[22] 张季如,陈敬鑫,王磊,等.三轴剪切过程中排水条件对钙质砂颗粒破碎、变形和强度特性的影响[J].岩土力学,2024,45(2):375-384.
[23] 孙越,肖杨,周伟,等.钙质砂和石英砂压缩下的颗粒破碎与形状演化[J].岩土工程学报,2022,44(6):1061-1068.
