A novel similarity algorithm for triangular cloud models based on exponential closeness and cloud drop variance
Jianjun Yang,Jiahao Han,Qilin Wan,Shanshan Xing,Hongbo Shi
DOI: https://doi.org/10.1007/s40747-024-01416-0
IF: 6.7
2024-04-20
Complex & Intelligent Systems
Abstract:Abstract Cloud model similarity algorithm is an important part of cloud modelling theory. Most of the existing cloud model similarity algorithms suffer from poor discriminability, poor classification, unstable results, and low time efficiency. In this paper, a new similarity algorithm is proposed that considers the triangular cloud model distance and shape. First, according to the $${{D}}_{\text{T}}$$ D T distance formula, a new exponential closeness measure is defined, with which the distance similarity of cloud models is characterized. Then, the shape similarity is calculated according to the variance of the cloud model cloud drops. Finally, the two similarities are synthesized to define a similarity algorithm for determining the distance from the $${{D}}_{\text{T}}$$ D T distance formula and shape based on the triangular cloud model (DD T STCM). In this paper, discriminability, stability, efficiency and theoretical interpretability are taken as the evaluation indices. Equipment security system capability evaluation experiment, cloud model differentiation simulation experiment and time series classification accuracy experiment are set up to verify the effectiveness of the algorithm in terms of the four above aspects. The experimental results show that DD T STCM has good differentiation and excellent classification effects. In the classification experiment for the time series, the average classification accuracy of DD T STCM reaches 91.78%, which is at least 2.78% higher than those of the other seven commonly used algorithms. The CPU running efficiency of DD T STCM is also extremely high, and the average CPU running time of group training is always on the order of milliseconds, which effectively reduces the time cost. Finally, a case study is conducted to analyse a risk assessment problem for China’s island microgrid industry, and the evaluation results based on DD T STCM are in line with human cognition and have good value for engineering applications. Graphical abstract
computer science, artificial intelligence