基于L-M算法的复数神经网络方法拟合粘弹性材料复模量

In this paper, a complex neural network method based on L-M algorithm was proposed to fit the complex modulus of viscoelastic material. It solved the problem that viscoelastic material constitutive relationship of solid rocket motor was complex and the complex modulus is difficult to fit effectively. The complex modulus expression of the material was obtained by generalized Maxwell model. It can construct the neural network with undecided complex network parameters Then the neural network solution was provided, which the input and the output samples were all complex. Referring to the real L-M algorithm, this algorithm was derived from the complex field. Through training of neural network, we achieved high precision fitting complex modulus of viscoelastic material. Numerical examples showed that compared with the traditional neural network fitting method, the proposed methoComplex Neural Network Method Based on L-M Algorithm for Fitting Complex Modulus of Viscoelastic Materiald had its superiority in training speed and generalization ability.