Tehnički vjesnik, Vol. 33 No. 4, 2026.
Izvorni znanstveni članak
https://doi.org/10.17559/TV-20251023003086
Short-Term Power Meteorological Disaster Risk Assessment Method Based on Sparrow Search Algorithm and Support Vector Machine
Yufeng Tai
; Jilin Electric Power Research Institute Co., Ltd., 130021, Changchun Jilin, China
*
Junbo Liu
; Jilin Electric Power Research Institute Co., Ltd., 130021, Changchun Jilin, China
Saipeng Zhang
; Jilin Electric Power Research Institute Co., Ltd., 130021, Changchun Jilin, China
Changlong Gao
; State Grid Jilin Electric Power Company Limited Electric Power Research Institute, 130021, Changchun Jilin, China
Qunying Yu
; State Grid Jilin Electric Power Company Limited Electric Power Research Institute, 130021, Changchun Jilin, China
Feilong Yi
; Jilin Electric Power Research Institute Co., Ltd., 130021, Changchun Jilin, China
Jiashuai Li
; State Grid Jilin Electric Power Company Limited Electric Power Research Institute, 130021, Changchun Jilin, China
* Dopisni autor.
Sažetak
Short-term meteorological disasters pose significant operational risks to power systems, making timely and accurate risk assessment crucial for system protection and emergency response. This study proposes a hybrid short-term power meteorological disaster risk assessment model that integrates an Improved Sparrow Search Algorithm (ISSA), a self-attention-enhanced Support Vector Machine (SISSA-SVM), and SA-optimized K-means clustering algorithm (SK-means). ISSA is employed to optimize model parameters and enhance global search capability, while SK-means provides high-quality meteorological scenario classification. SISSA-SVM then performs risk identification with improved representation learning and classification accuracy. Experimental results using ten years of meteorological and power fault data demonstrate that the proposed model converges rapidly, achieves robust fitting performance, and attains a risk identification accuracy of 97.32%. Moreover, in thunder and rainstorm scenarios, the model yields superior recall and markedly lower omission rates than competing methods. These findings indicate that the proposed hybrid model provides an efficient and reliable solution for short-term meteorological disaster risk assessment in modern power systems.
Ključne riječi
K-means; meteorological disaster; power risk assessment; SSA; SVM
Hrčak ID:
348704
URI
Datum izdavanja:
30.6.2026.
Posjeta: 0 *