择优RBF神经网络下切削率的优化  

Fitting and Predicting of Excision Rate with Optimization-Making RBF Artificial Neural Network

作　　者：龚敏庆[1,2] 孙明伟[1] 黄敏[1] 金明仲[2] 

机构地区：[1]贵州大学理学院,贵州贵阳550025 [2]贵州民族大学理学院,贵州贵阳550025

出　　处：《数理统计与管理》2013年第6期993-1001,共9页Journal of Applied Statistics and Management

基　　金：贵州2010省长基金;黔省专合字(2010)28号;全国统计科学重点项目(编号:2012LZ054);贵州省民委基金;贵州科技厅:黔科合J字[2011]2106;黔科合GY字[2011]3055;黔科合计Z字[2009]4002;贵阳科技局:筑科合同[2011101]1-32;筑科合同[2012101]2-17

摘　　要：本文通过择优RBF(径向基函数,Radial Basis Function)神经网络对影响切削加工过程的切削参数进行建模,对切除率进行拟合预测;提出松弛误差作为衡量网络精度的指标,使RBF选择最优的分布密度,从而有效提高RBF神经网络的拟合预测能力;并将择优RBF的拟合和预测结果与BP的相应结果进行了比较,结果显示择优RBF神经网络的拟合和预测精度大大优于BP神经网络.With Optimization-Making RBF Neural Network the excision rate was fitted and predicted via building on cutting parameters, which affecting cutting machining process. By proposing slack error as the indicator of the network＇s accuracy, radial basis function was made to select the best optimized distribution density in order to advance the fitting and forecasting capability of RBF Neural Network. The result of Optimization-Making RBF was compared with BP Neural Network＇s, what showed that the fitting and forecasting accuracy with Optimization-Making RBF was much higher than the accuracy with BF Neural Network.

关 键 词：择优RBF神经网络 切削参数 松弛误差 学习过程统计性质 

分 类 号：O212[理学—概率论与数理统计]