针对云模型相似性度量方法中存在的区分度不高、结果不稳定等问题,提出一种基于组合赋权的正态云模型形状相似性度量方法。首先基于期望曲线的位置关系,用正态模糊数的贴近度来表征云模型的形状相似度;然后考虑云滴的离散程度,在云滴方...针对云模型相似性度量方法中存在的区分度不高、结果不稳定等问题,提出一种基于组合赋权的正态云模型形状相似性度量方法。首先基于期望曲线的位置关系,用正态模糊数的贴近度来表征云模型的形状相似度;然后考虑云滴的离散程度,在云滴方差基础上提出基于熵与云滴方差的形状相似度;最后考虑云模型的三个数字特征,基于偏好系数,采用组合赋权,将两种形状相似度进行组合度量云模型相似度。通过仿真实验及时间序列分类实验表明,该方法是有效的,并具有较好的区分度和稳定性。To tackle the challenges of limited distinction and inconsistent outcomes in similarity measurement among cloud models, this paper proposed a method for measuring shape similarity of normal cloud models based on combinatorial weighting. Firstly, the approximation degree of a normal fuzzy number is employed to characterize the shape similarity of the cloud model based on its positional relation with the expected curve. Then considering the dispersion degree of cloud droplets, the shape similarity based on entropy and cloud droplet variance is proposed on the basis of cloud droplet variance. Finally, considering the three digital features of the cloud model, based on the preference coefficient, the combination weighting is used to combine the two shape similarities to measure the similarity of the cloud model. The simulation results show that the method is effective and has good discrimination and stability.展开更多
文摘针对云模型相似性度量方法中存在的区分度不高、结果不稳定等问题,提出一种基于组合赋权的正态云模型形状相似性度量方法。首先基于期望曲线的位置关系,用正态模糊数的贴近度来表征云模型的形状相似度;然后考虑云滴的离散程度,在云滴方差基础上提出基于熵与云滴方差的形状相似度;最后考虑云模型的三个数字特征,基于偏好系数,采用组合赋权,将两种形状相似度进行组合度量云模型相似度。通过仿真实验及时间序列分类实验表明,该方法是有效的,并具有较好的区分度和稳定性。To tackle the challenges of limited distinction and inconsistent outcomes in similarity measurement among cloud models, this paper proposed a method for measuring shape similarity of normal cloud models based on combinatorial weighting. Firstly, the approximation degree of a normal fuzzy number is employed to characterize the shape similarity of the cloud model based on its positional relation with the expected curve. Then considering the dispersion degree of cloud droplets, the shape similarity based on entropy and cloud droplet variance is proposed on the basis of cloud droplet variance. Finally, considering the three digital features of the cloud model, based on the preference coefficient, the combination weighting is used to combine the two shape similarities to measure the similarity of the cloud model. The simulation results show that the method is effective and has good discrimination and stability.
