Neural network algorithm for solving triple integral

摘要

In this paper, according to the triple integral definition and the characteristic of common network topology, a new model and learning algorithm is proposed for the triple integral. This method randomly selects certain nodes on each direction of integral domain in the initial, then the weight of the neural network is optimized, finally the quite precise integral result is obtained. Example is provided to show the high accuracy and fast convergence rate of the proposed neural network.