Scholarlink

<p>As a novel material with tightly interwoven fiber layers, needle-punched composites have significant applications in the engineering field. However, the process parameters of needled composite preforms are numerous, how to design the needling processes, while ensuring the mechanical properties within the surface, and enhancing the properties outside the surface, has always been the focus of research. In this study, a process optimization method based on artificial neural network and intelligent algorithms has been proposed. Firstly, the representative volume element (RVE) of 3D needle-punched composites specimens have established to conduct finite element simulation and the mechanical properties are obtained. The needling process parameters, the stiffness and strength properties are extracted as the training set in the neural network. Subsequently, a surrogate model based on the back propagation neural network (BPNN) is determined to establish the relationship between needling process parameters and mechanical properties of needled composites. The errors of both the predicted value and true value are less than 11.00%, which balance the improvement of solution efficiency and the demand for prediction accuracy. Then, the established BPNN model is combined with an intelligent optimization algorithm for the process optimization of needle-punched preforms. Taking the out-of-plane tensile strength as the objective function, the optimal combination scheme of processing parameters can be obtained based on the genetic algorithm (GA) and adaptive genetic algorithm (AGA) with an improved genetic strategy. When the design requirements are met, the out-of-plane tensile strength obtained by the AGA method is 1.22% higher than that obtained by original GA. Meanwhile, to avoid the optimization process getting stuck in a local optimal solution, the particle swarm optimization (PSO) algorithm is embedded within the GA, which yields a solution with higher accuracy. The out-of-plane tensile strength is 7.71% higher than the original one. This study realizes the efficient and high-precision optimization of the needled preform processes, and greatly promotes the performance improvement, which can boost the applications of needle-punched composites in the aerospace and high-temperature protection fields.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Process Optimization Design of Needle-Punched Composite Preforms Based on Machine Learning

  • 期刊论文

  • 发表日期：

    2026年5月11日

查看全文
  • Yunchao Qi,
  • Yuqi Liu,
  • Binwang Luo,
  • Xinfang Li,
  • Guizhe Zhao

摘要

As a novel material with tightly interwoven fiber layers, needle-punched composites have significant applications in the engineering field. However, the process parameters of needled composite preforms are numerous, how to design the needling processes, while ensuring the mechanical properties within the surface, and enhancing the properties outside the surface, has always been the focus of research. In this study, a process optimization method based on artificial neural network and intelligent algorithms has been proposed. Firstly, the representative volume element (RVE) of 3D needle-punched composites specimens have established to conduct finite element simulation and the mechanical properties are obtained. The needling process parameters, the stiffness and strength properties are extracted as the training set in the neural network. Subsequently, a surrogate model based on the back propagation neural network (BPNN) is determined to establish the relationship between needling process parameters and mechanical properties of needled composites. The errors of both the predicted value and true value are less than 11.00%, which balance the improvement of solution efficiency and the demand for prediction accuracy. Then, the established BPNN model is combined with an intelligent optimization algorithm for the process optimization of needle-punched preforms. Taking the out-of-plane tensile strength as the objective function, the optimal combination scheme of processing parameters can be obtained based on the genetic algorithm (GA) and adaptive genetic algorithm (AGA) with an improved genetic strategy. When the design requirements are met, the out-of-plane tensile strength obtained by the AGA method is 1.22% higher than that obtained by original GA. Meanwhile, to avoid the optimization process getting stuck in a local optimal solution, the particle swarm optimization (PSO) algorithm is embedded within the GA, which yields a solution with higher accuracy. The out-of-plane tensile strength is 7.71% higher than the original one. This study realizes the efficient and high-precision optimization of the needled preform processes, and greatly promotes the performance improvement, which can boost the applications of needle-punched composites in the aerospace and high-temperature protection fields.