关键词: 神经网络, 切比雪夫多项式, 多元函数, 权值快速计算

Abstract: Compared with the traditional multi-layer perceptron model, Chebyshev neural network has fast convergence, low complexity and generalization ability advantages. However, the one variable Chebyshev neural network to solve multivariate problems in the practical application is great limitations. For this problem, the paper extends and proposes multivariate Chebyshev neural network model and gives fast weight determination algorithm which base on Chebyshev polynomials orthogonal. The simulation results show that compared to traditional multi-layer perceptron neural network, the method has obvious advantages in the calculation accuracy and computing speed.

Key words: neural networks, Chebyshev polynomial, multivariate function, quick calculation of the weights