Radial-basis-neural-network point allocation method of solving statics response of composite material structure containing interval parameters

Abstract

The invention discloses a radial-basis-neural-network point allocation method of solving statics response of a composite material structure containing interval parameters. According to the method, firstly, a series of sample points are selected in a parameter interval domain of the composite material structure, and a finite-element method is used to solve structure statics response values of the sample points; the data are used as training samples of a radial-basis neural-network; the number of radial basis functions in a hidden layer of the neural network is selected according to the number of the sample points, then a K-means algorithm is used to obtain a center value of each radial basis function, and then a recursive least square method is used to obtain weight values between the hidden layer and an output layer of the neural network; the trained radial-basis neural-network is used as an approximate response function of an original structure response function; and then a genetic algorithm is utilized to solve maximum and minimum values of the radial-basis neural-network to use the same as an upper bound and a lower bound of the statics response of the composite material structure containing the interval parameters.

Description

technical field [0001] The invention relates to the technical field of static response analysis of composite material structures with interval uncertain parameters, in particular to a radial basis neural network collocation method for solving the static response of composite material structures with interval parameters. Background technique [0002] Structural analysis plays an important role in mechanical engineering, civil engineering, vehicle engineering and aerospace engineering. In order to ensure that the structure can work safely and reliably within its economic life, its strain, stress, displacement and other indicators need to be carefully checked before it is put into service. Structural analysis plays an important role in the whole link of structural design, and static response analysis is one of the most basic and important links in structural analysis. Traditional structural static analysis often regards various parameters of the structure as definite values, a...

AI Technical Summary

Problems solved by technology

When using the Monte Carlo method to solve the problem, a large number of samples are required in the uncertainty domain, which requires a large number of finite element calculations, which cannot be implemented in actual engineering; the perturbation method is a first-order approximation method, which can only Deal with problems with small nonlinear and uncertain intervals; the vertex method can only deal with monotone problems; and the collocation method can deal with problems with strong nonlinearity and can obtain more accurate solutions, but the processing process is more complicated

Images