基于RBF神经网络代理模型的复合材料壳体封头双区补强优化

Optimization of dual-zone reinforcement for composite case dome based on RBF neural network surrogate model

  • 摘要: 传统的复合材料壳体封头整体补强导致壳体冗余质量较高,壳体封头分区补强技术已在工程实践中得到应用,然而壳体封头分区补强工艺往往依赖于工程经验,分区补强工艺参数对壳体应力影响机制尚不明确。为此,针对内压载荷下壳体封头破坏行为,基于复合材料渐进损伤失效原理,对壳体失效模式、应变响应进行了分析,并准确预测了壳体的爆破压强。在此基础上,根据壳体封头应力分布和结构形式,基于RBF神经网络代理模型方法建立了壳体双区补强优化模型,采用单向布和编织布两种工程上常用的补强材料,揭示了单向/编织布层数比、补强范围对封头易失效位置应力响应的影响机制,获得最优的补强层数比和范围,并通过试验进行验证。结果表明:封头赤道圆附近轴向刚度较弱,宜采用编织布补强,补强范围需补至壳体筒身;而接头肩部附近宜采用90°单向/编织布组合补强形式,并设置合适的补强层数比,以提高轴向和环向刚度。通过水压爆破试验结果可知,优化后壳体筒身发生爆破,特征系数提升了14.85%,为复合材料壳体封头设计提供了理论参考。

     

    Abstract: Traditional full-area reinforcement of composite case domes results in excessive redundant case mass. Sectional reinforcement technology for these domes has been applied in engineering practice, yet the corresponding process often relies on engineering experience, and the influence mechanism of its process parameters on case stress remains unclear. To address this, aiming at case dome failure under internal pressure, its failure mode, strain response and burst pressure were analyzed and accurately predicted based on composite progressive damage theory; subsequently, a dual-zone reinforcement optimization model was established via RBF neural network surrogate model (using common unidirectional/woven fabrics), revealing how layer ratio and reinforcement range affect stress at failure-prone dome locations and obtaining optimal parameters. Results show weak axial stiffness near the dome equator requires woven fabric reinforcement extending to the case cylinder, while 90° unidirectional/woven composite reinforcement with proper layer ratio suits the joint shoulder; hydrostatic burst tests confirmed optimized cases burst at the cylinder with a 14.85% increase in the performance factor, providing theoretical support for composite case dome design.

     

/

返回文章
返回