Method of Mathematical Modeling Based on PSO Algorithms

Abstract

In order to set up universal and non-linear map of variables, a full binary tree is constructed as mathematical model. Leaf nodes of the full binary tree are linear combination of input variables, and used as inputs of next nodes. On the basis of weighting two inputs by selector for inner node, the inputs are again linearly combined and used as output for next node. The inputs and outputs of all the inner nodes are constructed in turn as the same, and the output of root node is the output of mathematical model, implementing segment-linear approximation. With the means of machine learning of particle swarm optimization for data from some areas, all the coefficients of mathematical model are achieved for the special. The mathematical model is applied to seismic inversion to interpret stratum by seismic data, approving it very practical.