论文部分内容阅读
为有效缩短脉冲激光烧蚀制备有机硅聚合物聚二苯基硅亚甲基硅烷(PDPhSM)基纳米复合薄膜工艺中繁琐的实验过程,分别采用多层前馈(BP)神经网络和径向基函数(RBF)神经网络对PDPhSM基纳米复合薄膜的制备工艺与聚合效率之间的关系进行建模,并将其运用到聚合效率的预测中去,讨论了激光能量密度、环境压强、靶衬距离、沉积时间和聚合效率之间的关系。克服了以往单因素实验法不能正确反映制备工艺和聚合效率之间复杂的非线性关系的弱点。预测和验证结果均表明实验值和网络预测值之间相对误差都在10%以内,但径向基函数神经网络较多层前馈神经网络能够更精确、更可靠地逼近它们之间的非线性关系。该方法为有效、快捷、经济地开发研制PDPhSM基纳米复合薄膜提供了新的思路和有效手段。
In order to shorten the experimental process of pulsed laser ablation of PDPhSM-based nanocomposite films, multilayer feedforward (BP) neural networks and radial basis (RBF) neural network was used to model the relationship between the polymerization process and the polymerization efficiency of PDPhSM-based nanocomposite films. The RBF neural network was used to predict the polymerization efficiency. The effects of laser energy density, ambient pressure, target liner distance , The relationship between deposition time and polymerization efficiency. Overcoming the weakness of the single-factor experiment method that can not correctly reflect the complicated nonlinear relationship between the preparation process and the polymerization efficiency. Both prediction and verification results show that the relative errors between experimental values and network prediction values are within 10%, but radial basis function neural networks can approach the nonlinearity between them more accurately and reliably than multi-layer feedforward neural networks relationship. The method provides a new idea and effective method for developing PDPhSM-based nanocomposite films effectively, rapidly and economically.