A Bibliometric Review of Hybrid CNN-BiLSTM Deep Learning Frameworks for Fault Detection, Classification, and Location in Power Transmission Systems

This paper presents a bibliometric review of hybrid Convolutional Neural Network–Bidirectional Long Short-Term Memory (CNN-BiLSTM) models used for fault detection, fault classification, and fault localization in electrical power systems. A systematic literature search was conducted using the Scopus database following the PRISMA 2020 guidelines. After screening and eligibility assessment, 43 peer-reviewed journal articles and conference papers published between 2023 and 2026 were selected for analysis. The selected studies were examined based on publication trends, document and source distribution, keyword co-occurrence, citation analysis, and reported model performance. The review shows a significant increase in research on CNN-BiLSTM-based fault diagnosis since 2024. Keywords such as fault diagnosis, deep learning, and BiLSTM appeared most frequently, indicating the growing interest in hybrid deep learning techniques. Most studies reported diagnostic accuracies above 95%, demonstrating the effectiveness of CNN–BiLSTM models for power system fault diagnosis. However, the review also identifies several research gaps, including the limited availability of unified models for simultaneous fault detection, classification, and localization, the lack of standardized benchmark validation, and the need for robust real-time implementation under noisy operating conditions. These findings provide useful insights and highlight future research directions for developing intelligent fault diagnosis frameworks for power transmission systems.