Article: Transformer fault diagnosis model for power system based on improved CNN-BiLSTM and multimodal feature fusion Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.88 pp.73 - 92 Abstract: To solve the problem of low precision of power system fault diagnosis, a Transformer fault diagnosis model for power system based on improved CNN-BiLSTM and multimodal feature fusion is proposed. Firstly, this paper adopts improved CNN to extract infrared image features of power system transformer, and uses BiLSTM network to extract oil chromatograms and local discharge sequence features of power system transformer. Moreover, multimodal feature fusion algorithm is utilised for feature fusion and fault diagnosis. Finally, simulation experiments are conducted for verification. Experimental results show that the average false detection rate and average loss detection rate of proposed model are 2.99% and 3.53%, respectively, which are about 6% lower than those of other commonly used fault diagnosis models. Therefore, the proposed model improves the precision of early warning and fault diagnosis for power system transformers, providing technical support for ensuring the normal operation of the power system. Inderscience Publishers - linking academia, business and industry through research

Title: Transformer fault diagnosis model for power system based on improved CNN-BiLSTM and multimodal feature fusion

Authors: Yun Dong; Kaijie Liu; Zhipeng Meng; Qi Chen; Shibo Zhang; Chen Wang; Wenyu Zhang; Yaxiong Liu

Addresses: Digital and Intelligent Operation Center, Guangxi Power Grid Co., Ltd., Nanning Guangxi, 530015, China ' Digital and Intelligent Operation Center, Guangxi Power Grid Co., Ltd., Nanning Guangxi, 530015, China ' Digital and Intelligent Operation Center, Guangxi Power Grid Co., Ltd., Nanning Guangxi, 530015, China ' Digital and Intelligent Operation Center, Guangxi Power Grid Co., Ltd., Nanning Guangxi, 530015, China ' Power Grid Planning Research Center, Guangxi Power Grid Co., Ltd., Nanning Guangxi, 530015, China ' Liuzhou Power Supply Bureau of Guangxi Power Grid Co., Ltd., Liuzhou Guangxi, 545000, China ' Digital Department, Guangxi Power Grid Co., Ltd., Nanning Guangxi, 530015, China ' Digital and Intelligent Operation Center, Guangxi Power Grid Co., Ltd., Nanning Guangxi, 530015, China

Abstract: To solve the problem of low precision of power system fault diagnosis, a Transformer fault diagnosis model for power system based on improved CNN-BiLSTM and multimodal feature fusion is proposed. Firstly, this paper adopts improved CNN to extract infrared image features of power system transformer, and uses BiLSTM network to extract oil chromatograms and local discharge sequence features of power system transformer. Moreover, multimodal feature fusion algorithm is utilised for feature fusion and fault diagnosis. Finally, simulation experiments are conducted for verification. Experimental results show that the average false detection rate and average loss detection rate of proposed model are 2.99% and 3.53%, respectively, which are about 6% lower than those of other commonly used fault diagnosis models. Therefore, the proposed model improves the precision of early warning and fault diagnosis for power system transformers, providing technical support for ensuring the normal operation of the power system.

Keywords: power system; transformer; fault diagnosis; CNN network; BiLSTM network; multi-modal feature fusion algorithm.

DOI: 10.1504/IJICT.2026.155569

International Journal of Information and Communication Technology, 2026 Vol.27 No.88, pp.73 - 92

Received: 15 Apr 2026
Accepted: 09 Jun 2026
Published online: 05 Aug 2026 *

Title: Transformer fault diagnosis model for power system based on improved CNN-BiLSTM and multimodal feature fusion

Keep up-to-date