论文部分内容阅读
为提高偏微分方程的计算求解精度,设计了以多元二次径向基神经网络为求解单元的偏微分计算方法,给出了多元二次径向基神经网络的具体求解结构,并以此神经网络为求解基础,给出了具体的偏微分计算步骤.通过具体的偏微分求解实例验证方法的有效性,并以3种不同设计样本数构建的多元二次径向基神经网络为计算单元,从实例求解所需的计算时间以及解的精度作对比,结果表明,采用基于多元二次径向基神经网络的偏微分方程求解方法具有求解精度高以及计算效率低等特点.
In order to improve the computational precision of PDEs, a partial differential method based on multivariate quadratic radial basis function neural networks was designed and the solution structure of multivariate quadratic radial basis function neural networks was given. Network is used to solve the problem and the specific step of partial differential calculation is given.The validity of the method is verified by a concrete example of partial differential method.The multivariate secondary RBF neural network with three different design samples is used as the calculation unit, Comparing with the calculation time and the precision of solution, the results show that the method of solving PDEs based on multivariate quadratic radial basis function neural network has the characteristics of high solution accuracy and low computational efficiency.