遗传算法优化的BP神经网络税收模型

A Kind of Taxation Forecasting Model Based on Genetic Algorithm Optimized BP Neural Network

本文将遗传算法与局部寻优算法相结合代替传统的BP算法寻找BP神经网络模型中最佳的权值与阀值,进而建立了税收预测模型。对某区域的数值仿真表明该方法比单纯用BP方法或单纯用遗传算法有更高的精度,效果令人满意。

The author adopts a new combined method of genetic algorithm and local optimization algorithm, instead of traditional BP algorithm, to find the fittest values of weights and biases of the neural network model and then sets up a taxationforecasting model. By simulating this model using the data of certain district,this method has better precision than using BP algorithm only or using genetic algorithm only, and the result is satisfied.