肖宏成 , 尹冬梅 , 林庆华 , 等 . 轨道炮复合身管纤维缠绕封装结构优化设计及预应力模拟 [J ] . 高压物理学报 , 2018 , 32 ( 5 ): 152 - 158 .

XIAO H C , YIN D M , LIN Q H , et al. Structural optimum design and pre-stress simulation of fiber housing for railgun [J ] . Chinese Journal of High Pressure Physics , 2018 , 32 ( 5 ): 152 - 158 . (in Chinese)

张峻铭 , 杨伟东 , 李岩 . 人工智能在复合材料研究中的应用 [J ] . 力学进展 , 2021 , 51 ( 4 ): 865 - 900 .

ZHANG J M , YANG W D , LI Y . Application of artificial intelligence in composite materials [J ] . Advances in Mechanics , 2021 , 51 ( 4 ): 865 - 900 . (in Chinese)

BABU U H , SAI N V , SAHU R K . Artificial intelligence system approach for optimization of drilling parameters of glass-carbon fiber/polymer composites [J ] . Silicon , 2021 , 13 : 2943 - 2957 .

EHSANI A , DALIR H . Multi-objective optimization of composite angle grid plates for maximum buckling load and minimum weight using genetic algorithms and neural networks [J ] . Composite Structures , 2019 , 229 : 111450 .

FU X W , RICCI S , BISAGNI C . Minimum-weight design for three dimensional woven composite stiffened panels using neural networks and genetic algorithms [J ] . Composite Structures , 2015 , 134 : 708 - 715 .

徐亚栋 , 钱林方 . 基于改进非支配排序遗传算法的复合材料身管多目标优化 [J ] . 兵工学报 , 2006 , 27 ( 4 ): 617 - 621 . 针对复合材料身管设计时多个性能指标设计要求，基于有限元模型和优化设计方法，建立了复合材料身管多目标优化模型，优化的目标为身管一阶固有频率和身管质量，复合材料各层缠绕角和缠绕厚度为优化设计变量，优化方法采用改进的非支配排序遗传算法（NSGA-Ⅱ）；通过优化求解，获得了复合材料身管的Pareto最优解集；优化算例表明，采用NSGA-Ⅱ获得的Pareto前沿面曲线分布均匀，其所对应的复合材料身管设计方案在刚度和重量方面均有改善。

XU Y D , QIAN L F . Multi-objective optimization of composite barrel based on the improved nondominated sorting genetic algorithm [J ] . Acta Armamentarii , 2006 , 27 ( 4 ): 617 - 621 . (in Chinese)

田会方 , 仇振兴 , 吴迎峰 . 基于改进NSGA-Ⅱ的纤维缠绕落纱点轨迹采样特征权重优化 [J ] . 复合材料科学与工程 , 2024 ( 1 ): 54 - 59 . DOI: 10.19936/j.cnki.2096-8000.20240128.007 http://doi.org/10.19936/j.cnki.2096-8000.20240128.007 针对基于空间特征曲线特征函数的纤维缠绕落纱点轨迹采样算法无法自动选择特征权重的问题,建立以特征权重为变量,以得到采样点线性插值生成曲线与原曲线的MAE,RMSE为目标函数的双目标优化模型。提出基于改进NSGA-Ⅱ算法的双目标优化求解方法以优化特征权重。实例验证表明,与传统NSGA-Ⅱ算法相比,改进NSGA-Ⅱ算法求得Pareto解集的MAE,RMSE平均下降了0.002和0.105,算法选取特征权重的MAE,RMSE比特征权重为(0.1,0.3)的MAE,RMSE分别降低了约12.9%和8.5%,比特征权重为(0.9,0.1)的MAE,RMSE分别降低了约20.6%和11.4%,有效地提高了落纱点轨迹采样的精度。

TIAN H F , QIU Z X , WU Y F . Optimization of feature weights of filament winding dropping point trajectory sampling based on improved NSGA-II [J ] . Composites Science and Engineering , 2024 ( 1 ): 54 - 59 . (in Chinese)

李彬 , 谢新 , 唐文勇 , 等 . 基于近似模型的复合材料导管支臂结构性能分析 [J ] . 兵工学报 , 2022 , 43 ( 6 ): 1435 - 1446 . DOI: 10.12382/bgxb.2021.0300 http://doi.org/10.12382/bgxb.2021.0300 为分析气垫船复合材料导管支臂结构性能并进行高效率结构设计，采用一种带有权重系数变量的组合神经网络近似模型替代结构有限元计算。根据各铺层设计参数对结构性能影响的灵敏度进行分析并分组，建立多个神经网络模型，结合权重系数与各模型输出得到响应预测值。基于复合材料可设计性，应用优化拉丁超立方试验设计方法从细观和宏观层面分析研究导管支臂性能中材料参数的作用。结果表明：所提复合材料导管支臂结构组合近似模型方法具有较高精度，纤维弹性模量和基体剪切模量对导管支臂结构材料力学性能具有主导作用；在组分材料属性一定情况下，纤维体积分数的增加能够提高结构刚度与稳定性，在不同受力状态下纤维体积分数与各铺层角对结构失效影响有差异，设计时需进行特殊考虑；所提方法可为复合材料导管支臂结构性能分析及支臂结构材料一体化设计提供理论依据。

LI B , XIE X , TANG W Y , et al. Analysis of structural performance of composite duct supporting structure based on approximate model [J ] . Acta Armamentarii , 2022 , 43 ( 6 ): 1435 - 1446 . (in Chinese) DOI: 10.12382/bgxb.2021.0300 http://doi.org/10.12382/bgxb.2021.0300 The ensemble neural network approximation model method with variable weight coefficient is used to replace the structural finite element calculation in order to analyze the structural performance of the composite duct supporting structure of an air cushion vehicle for efficient structural design. Several neural network models are establishedaccording to the sensitivity of the ply design parameters to the structural performance，and the predicted response values are obtained from the weight coefficients and the output of each model. Based on the designability of composite materials，the influence of material parameters on the performance of the supporting structure is analyzed microcosmically and macroscosmically level by using the optimized Latin hypercube design method. The results show that: the approximate model for the composite material has a high accuracy；and the fiber elastic modulus and matrix shear modulus have significant impact on the mechanical properties of the composite material of the supporting structure. The increase in fiber volume fraction can improve the stiffness and stability of the structure when the properties of component materials are fixed. The influences of fiber volume fraction and ply angles on the structural failure are different under different stress states，so special consideration should be taken in the design.The proposed method provides a theoretical basis for the structural performance analysis and the integrated design of composite material duct supporting structure.

胡春幸 , 侯玉亮 , 铁瑛 , 等 . 基于遗传算法的碳纤维增强树脂复合材料层合板单搭胶接结构的多目标优化 [J ] . 复合材料学报 , 2021 , 38 ( 6 ): 1847 - 1858 .

HU C X , HOU Y L , TIE Y , et al. Multi-objective optimization of adhesively bonded single-lap joints of carbon fiber reinforced polymer laminates based on genetic algorithm [J ] . Acta Materiae Compositae Sinica , 2021 , 38 ( 6 ): 1847 - 1858 . (in Chinese)

COELLO C A C , PULIDO G T , LECHUGA M S . Handling multiple objectives with particle swarm optimizatiom [J ] . IEEE Transactions on Evolutionary Computation , 2004 , 8 ( 3 ): 256 - 279 .

GOPAL V E , PRASAD M V N K , RAVI V . A fast and elitist multi-objective genetic algorithm: NSGA-II [J ] . IEEE Transactions on Evolutionary Computation , 2002 , 6 ( 2 ): 182 - 197 .

黄争鸣 . 复合材料破坏与强度 [M ] . 北京 : 科学出版社 , 2018 .

HUANG Z M . Composite material damage and strength [M ] . Beijing : Science Press , 2018 . (in Chinese)

徐亚栋 , 钱林方 , 陈龙淼 . 复合材料身管三维有限元分析与结构优化 [J ] . 弹箭与制导学报 , 2005 , 25 ( 4 ): 209 - 212 .

XU Y D , QIAN L F , CHEN L M . Three-dimensional finite element analysis and optimization of composite material barrel [J ] . Journal of Projectiles, Rockets, Missiles and Guidance , 2005 , 25 ( 4 ): 209 - 212 . (in Chinese)

戴忠华 , 周穗华 , 张晓兵 . 多目标优化的舰船磁场建模方法 [J ] . 物理学报 , 2021 , 70 ( 16 ): 164101 .

DAI Z H , ZHOU S H , ZHANG X B . Multi-objective optimization of ship magnetic field modeling method [J ] . Acta Physica Sinica , 2021 , 70 ( 16 ): 164101 . (in Chinese)

周泉 , 章兢 . 基于克隆选择原理的免疫算法 [J ] . 计算机工程与应用 , 2005 , 41 ( 21 ): 61 - 63 .

ZHOU Q , ZHANG J . Immune algorithm based on clonal selection principle [J ] . Computer Engineering and Applications , 2005 , 41 ( 21 ): 61 - 63 . (in Chinese)

VALIAN E , TAVAKOLI S , MOHANNA S , et al. Improved cuckoo search for reliability optimization problems [J ] . Computers & Industrial Engineering , 2013 , 64 ( 1 ): 459 - 468 .

刘琼 , 吴小俊 . 一种改进的免疫克隆选择算法 [J ] . 山东大学学报(工学版) , 2009 , 39 ( 6 ): 8 - 12 .

LIU Q , WU X J . An improved immune clonal elesction algorithm [J ] . Journal of Shandong University (Engineering Science) , 2009 , 39 ( 6 ): 8 - 12 . (in Chinese)

张玲 , 白中英 , 谢康 . 动态疫苗接种的入侵检测方法 [J ] . 北京邮电大学学报 , 2014 , 37 ( 增刊1 ): 77 - 82 .

ZHANG L , BAI Z Y , XIE K . Dynamic introduction detection with vaccination [J ] . Journal of Beijing University of Posts and Telecommunication , 2014 , 37 ( S1 ): 77 - 82 . (in Chinese)

ZHU Q L , LIN Q Z , CHEN W N , et al. An external archive-guided multi-objective particle swarm optimization algorithm [J ] . IEEE Transactions on Cybernetics , 2017 , 47 ( 9 ): 2794 - 2808 .

VALIAN E , TAVAKOLI S , MOHANNA S , et al. Improved cuckoo search for reliability optimization problems [J ] . Computers & Industrial Engineering , 2013 ,64: 459 - 468 .

ZHOU A M , ZHANG Q F , JIN Y C . Approximating the set of pareto-optimal solutions in both the decision and objective spaces by an estimation of distribution algorithm [J ] . IEEE Transactions on Evolutionary Computation , 2009 , 13 ( 5 ): 1167 - 1189 .

刘朝华 , 章兢 , 李小花 , 等 . 免疫协同微粒群进化算法的永磁同步电机多参数辨识模型方法 [J ] . 自动化学报 , 2012 , 38 ( 10 ): 1698 - 1708 .

LIU Z H , ZHANG J , LI X H , et al. Immune co-evolution particle swarm optimization for permanent margent synchronous motors parameter identification [J ] . Acta Automatica Sinica , 2012 , 38 ( 10 ): 1698 - 1708 . (in Chinese)

SMITH J S , WU B , WILAMOWSKI B M . Neural network training with Levenberg-Marquardt and adaptable weight compression [J ] . IEEE Transactions on Neural Networks and Learning Systems , 2019 , 30 ( 2 ): 580 - 587 . DOI: 10.1109/TNNLS.2018.2846775 http://doi.org/10.1109/TNNLS.2018.2846775 Difficult experiments in training neural networks often fail to converge due to what is known as the flat-spot problem, where the gradient of hidden neurons in the network diminishes in value, rending the weight update process ineffective. Whereas a first-order algorithm can address this issue by learning parameters to normalize neuron activations, the second-order algorithms cannot afford additional parameters given that they include a large Jacobian matrix calculation. This paper proposes Levenberg-Marquardt with weight compression (LM-WC), which combats the flat-spot problem by compressing neuron weights to push neuron activation out of the saturated region and close to the linear region. The presented algorithm requires no additional learned parameters and contains an adaptable compression parameter, which is adjusted to avoid training failure and increase the probability of neural network convergence. Several experiments are presented and discussed to demonstrate the success of LM-WC against standard LM and LM with random restarts on benchmark data sets for varying network architectures. Our results suggest that the LM-WC algorithm can improve training success by 10 times or more compared with other methods.