摘要
研究了高斯函数、多二次函数、逆多二次函数、薄板样条函数、三次函数和线性函数对RBF-ARX模型的影响。选取Mackey-Glass混沌方程、Lorenz吸引子和Box-Jenkins煤气炉3种标准时间序列作为测试模型的数据,采用一种快速收敛的结构化非线性参数优化方法辨识RBF-ARX模型。研究结果表明:最优基函数的选择并不一定是最常用的高斯函数,而是与问题相关,因而,在实际建模中,评价各种基函数有助于选择最优结构的RBF-ARX模型。
The effects of different basis functions including Gaussian,multiquadratic,inverse multiqudratic,thin plate spline,cubic and linear on the radial basis function network-style coefficients auto regressive model with exogenous variable（RBF-ARX） model were examined.Several benchmark time series including Mackey-Glass,Lorenz attractor and Box-Jenkins gas furnace were used as the test data.A fast-converging estimation method was applied to optimizing the RBF-ARX model parameters.The simulation results show that the optimal choice of basis function is not a normal Gaussian function but a problem dependent and evaluating all the recognised basis functions suitable for the RBF-ARX model is advantageous.
出处
《中南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2010年第6期2231-2235,共5页
Journal of Central South University:Science and Technology
基金
湖南省科技计划国际合作重点资助项目(2009WK2009)
国家创新研究群体科学基金资助项目(70921001)
