摘要
在分析织物热传递性能与相关影响因素之间关系的基础上,建立了织物热传递性能预测的广义神经网络模型(GRNN).并与传统的BP网络模型仿真结果进行了比较,结果表明:GRNN网络设计简单,学习收敛快,在解决小样本问题的学习中,具有更好的的预测和泛化能力,验证了GRNN网络预测的优越性和有效性.
The paper analyzes the relationship between the heat permeability of the fabric and relative factors, and then establishes the general regression neural network (GRNN) model for the heat permeability of the fabric forecast. Compared with the BP neural network, GRNN is simpler, the calculation time needed for convergence is shorter, and it can give better prediction and generalization performances in small sample space. The superiority and effectiveness of using GRNN to forecast the heat permeability of the fabric is demonstrated.
基金
安徽省自然科学基金资助项目(004719)
关键词
GRNN
神经网络
热传递性
预测
GRNN
neural network
heat permeability
prediction