提出一种优化的核模糊C均值聚类算法(WBAKFCM).该算法首先通过改进蝙蝠算法(Weight bat Algorithm,WBA)确定最优聚类中心集合,然后用核模糊C均值聚类算法指导聚类划分.一方面,改进的蝙蝠算法在传统的蝙蝠算法中引入佳点集理论和速度权重...提出一种优化的核模糊C均值聚类算法(WBAKFCM).该算法首先通过改进蝙蝠算法(Weight bat Algorithm,WBA)确定最优聚类中心集合,然后用核模糊C均值聚类算法指导聚类划分.一方面,改进的蝙蝠算法在传统的蝙蝠算法中引入佳点集理论和速度权重,分别用于调节种群的初始化和个体位置的自适应更新.另一方面,在核模糊C均值聚类算法(Kernel Fuzzy C-Means,KFCM)中,选用了高斯核函数,从而将数据映射到高维特征空间进行聚类划分.实验结果表明,优化的核模糊C均值聚类算法在聚类准确率与时间效率上明显优于传统算法.展开更多
本文提出一种新的数据驱动建模思路及方法,即面向建模误差概率密度函数(Probability density function,PDF)形状与趋势拟合优度(相似度)的动态过程多目标优化建模方法.首先,针对均方根误差(Root mean square error,RMSE)等常规一维性能...本文提出一种新的数据驱动建模思路及方法,即面向建模误差概率密度函数(Probability density function,PDF)形状与趋势拟合优度(相似度)的动态过程多目标优化建模方法.首先,针对均方根误差(Root mean square error,RMSE)等常规一维性能指标不能完全刻画建模误差在时间和空间二维随机动态特性的问题,引入PDF指标来对动态过程的建模误差在时间和空间进行二维尺度的全面刻画和评价,并采用核密度估计技术对关于时间的建模误差序列的PDF进行估计;其次,为了刻画动态过程数据建模的拟合趋势,进一步引入趋势拟合优度指标,从而构造综合描述数据建模误差PDF形状与趋势拟合相似性的多目标性能指标;在此基础上,采用NSGA-II算法优化数据模型的参数集,获取一大类满足上述多目标性能优化的智能模型参数解.数值仿真及工业数据验证表明,所提方法的建模误差PDF逼近设定的期望PDF,并且模型输出与样本数据拟合趋势接近,好于常规最小化一维RMSE指标的数据建模方法.展开更多
We define a class of confidence bands for distribution functions,named simple confidence bands.The class of bands includes the common step bands and continuous bands,some of which may perform better than the smoothed ...We define a class of confidence bands for distribution functions,named simple confidence bands.The class of bands includes the common step bands and continuous bands,some of which may perform better than the smoothed bands not belonging to the class,e.g.,the kernel smoothed bands.It is shown that under some mild assumptions,the simple bands with exact coverage for continuous distribution functions are all step bands.The unbiasedness problem of the step bands is also investigated.It is proved that most of two-sided step bands are biased and one-sided step bands are unbiased.展开更多
文摘提出一种优化的核模糊C均值聚类算法(WBAKFCM).该算法首先通过改进蝙蝠算法(Weight bat Algorithm,WBA)确定最优聚类中心集合,然后用核模糊C均值聚类算法指导聚类划分.一方面,改进的蝙蝠算法在传统的蝙蝠算法中引入佳点集理论和速度权重,分别用于调节种群的初始化和个体位置的自适应更新.另一方面,在核模糊C均值聚类算法(Kernel Fuzzy C-Means,KFCM)中,选用了高斯核函数,从而将数据映射到高维特征空间进行聚类划分.实验结果表明,优化的核模糊C均值聚类算法在聚类准确率与时间效率上明显优于传统算法.
文摘本文提出一种新的数据驱动建模思路及方法,即面向建模误差概率密度函数(Probability density function,PDF)形状与趋势拟合优度(相似度)的动态过程多目标优化建模方法.首先,针对均方根误差(Root mean square error,RMSE)等常规一维性能指标不能完全刻画建模误差在时间和空间二维随机动态特性的问题,引入PDF指标来对动态过程的建模误差在时间和空间进行二维尺度的全面刻画和评价,并采用核密度估计技术对关于时间的建模误差序列的PDF进行估计;其次,为了刻画动态过程数据建模的拟合趋势,进一步引入趋势拟合优度指标,从而构造综合描述数据建模误差PDF形状与趋势拟合相似性的多目标性能指标;在此基础上,采用NSGA-II算法优化数据模型的参数集,获取一大类满足上述多目标性能优化的智能模型参数解.数值仿真及工业数据验证表明,所提方法的建模误差PDF逼近设定的期望PDF,并且模型输出与样本数据拟合趋势接近,好于常规最小化一维RMSE指标的数据建模方法.
基金supported by National Science Foundation for Post-doctoral Scientists of China (Grant No.20090450603)National Natural Science Foundation of China (Grant No.10771015)
文摘We define a class of confidence bands for distribution functions,named simple confidence bands.The class of bands includes the common step bands and continuous bands,some of which may perform better than the smoothed bands not belonging to the class,e.g.,the kernel smoothed bands.It is shown that under some mild assumptions,the simple bands with exact coverage for continuous distribution functions are all step bands.The unbiasedness problem of the step bands is also investigated.It is proved that most of two-sided step bands are biased and one-sided step bands are unbiased.
