基于遗传神经网络的科研项目经费预算方法

针对科研经费预算准确性与实用性无法兼顾的难题,提出基于遗传神经网络的科研项目经费预算方法.分析了科研项目经费构成及影响预算的关键因素,建立了影响因素的量化表达式,将量化结果作为输入,项目经费预算结果作为输出,利用遗传算法优化BP(Back Propagation)神经网络的权值和阈值,通过训练遗传神经网络预测模型建立输入与输出之间的非线性关系.运用预算实例对预测方法进行了有效性验证.仿真结果表明该方法有较高的预测准确性.

Scientific Research Project Budget Estimate Method Based on GA BP Neural Network

For the problem of accuracy and practicality of research budget can not be considered at the same time. A research budget estimating method based on GA-BP neural network was presented. Firstly, the component and influencing factors of research budget were analysed. Secondly, a research budget estimating model based on GA-BP neural network was built, GA was used to optimize the weights and thresholds of BP neural network. A nonlinear expression between influencing factors (input) and budget (output) was created. Finally, the model was implemented by Matlab. The effectiveness of the method was tested. Testing experiment showed the model based on GA-BP neural network is reliable and accurate.