Article: Line loss anomaly identification in power grids using grey wolf algorithm-optimised SVR Journal: International Journal of Computer Applications in Technology (IJCAT) 2026 Vol.78 No.6 pp.78 - 88 Abstract: The operating conditions of the power grid have strong nonlinear time-varying characteristics, and the line loss data presents non-stationary features, resulting in a decrease in the accuracy of line loss anomaly identification and a high false alarm rate. Therefore, a grey wolf algorithm is proposed to optimise the identification method of power grid line loss anomalies under SVR. Firstly, LSTM network is used to complete the missing power line loss data. Secondly, by simulating the encirclement mechanism and position update strategy of grey wolf hunting behaviour, the SVR parameters are adaptively adjusted. Finally, with the maximum number of iterations as the termination condition, output the optimal parameter combination to identify the abnormal state of line loss. The experimental results show that the accuracy of the proposed method for identifying line loss anomalies always remains in the high range of 96-98%, and the false alarm rate always remains below 2%. Inderscience Publishers - linking academia, business and industry through research

Title: Line loss anomaly identification in power grids using grey wolf algorithm-optimised SVR

Authors: Bin Wang; Chen Luo; Shiping Yang; Yongqing Zhu; Zhen Li; Yuanming Liu

Addresses: Power Grid Planning Research Centre, Guizhou Power Grid Co., Ltd., Guiyang, Guizhou Province, China ' Power Grid Planning Research Centre, Guizhou Power Grid Co., Ltd., Guiyang, Guizhou Province, China ' Power Grid Planning Research Centre, Guizhou Power Grid Co., Ltd., Guiyang, Guizhou Province, China ' Power Grid Planning Research Centre, Guizhou Power Grid Co., Ltd., Guiyang, Guizhou Province, China ' Power Grid Planning Research Centre, Guizhou Power Grid Co., Ltd., Guiyang, Guizhou Province, China ' Power Grid Planning Research Centre, Guizhou Power Grid Co., Ltd., Guiyang, Guizhou Province, China

Abstract: The operating conditions of the power grid have strong nonlinear time-varying characteristics, and the line loss data presents non-stationary features, resulting in a decrease in the accuracy of line loss anomaly identification and a high false alarm rate. Therefore, a grey wolf algorithm is proposed to optimise the identification method of power grid line loss anomalies under SVR. Firstly, LSTM network is used to complete the missing power line loss data. Secondly, by simulating the encirclement mechanism and position update strategy of grey wolf hunting behaviour, the SVR parameters are adaptively adjusted. Finally, with the maximum number of iterations as the termination condition, output the optimal parameter combination to identify the abnormal state of line loss. The experimental results show that the accuracy of the proposed method for identifying line loss anomalies always remains in the high range of 96-98%, and the false alarm rate always remains below 2%.

Keywords: grey wolf algorithm; support vector machine regression; power grid line loss; abnormal identification.

DOI: 10.1504/IJCAT.2026.153744

International Journal of Computer Applications in Technology, 2026 Vol.78 No.6, pp.78 - 88

Received: 19 Aug 2025
Accepted: 09 Dec 2025
Published online: 22 May 2026 *

Title: Line loss anomaly identification in power grids using grey wolf algorithm-optimised SVR

Keep up-to-date