Novel Dynamic Security Assessment Technique for Data Driven Stability Analysis with False Data Injection Attack Prediction in Smart Grid

Mohammad Kamrul Hasan^1,*, A. K. M. Ahasan Habib¹, Shayla Islam^2,*, A. K. M. Zakir Hossain^3,*, Masrullizam Mat Ibrahim³, Rosilah Hassan¹, Rahul Thakkar⁴, Nguyen Vo⁴
1 Center for Cyber Security, Faculty of Information Science and Technology, Universiti Kebangsaan Malaysia (UKM), Bangi, Hulu Langat, Malaysia
2 Institute of Computer Science and Digital Innovation, UCSI University, Kuala Lumpur, Malaysia
3 Centre for Telecommunication Research and Innovation, Fakulti Teknologi dan Kejuruteraan Elektronik dan Komputer, Universiti Teknikal Malaysia Melaka, Durian Tunggal, Melaka, Malaysia
4 Department of Information Technology and Systems, Victorian Institute of Technology, Melbourne, Australia
* Corresponding Author: Mohammad Kamrul Hasan. Email: email ; Shayla Islam. Email: email ; A. K. M. Zakir Hossain. Email: email

Computer Modeling in Engineering & Sciences https://doi.org/10.32604/cmes.2026.083521

Received 05 April 2026; Accepted 24 June 2026; Published online 01 September 2026

Download PDF

Abstract

Dynamic security assessment (DSA) of power system devices in smart grid (SG) power systems is currently essential for minimizing widespread blackouts and preventing cyberattacks. For stability processes that are difficult to perform in real time, security evaluation approaches for current SG devices may therefore require extensive historical domain training. Given that predictions are instantaneous, machine learning (ML) can be used to predict DSA. To classify and predict the time margin and transient energy margin (TEM) for a specific fault position and operating condition, input features, reactive power outputs, SG device transient energy function (TEF) terms, and fault location are used. False data injection (FDI) attacks are analyzed to predict the vulnerability using the developed Stacked Autoencoder (SAE) with ML models. In particular, the hybrid voting machine classifier (VMC) values are modified, and the data features are used to identify attack targets. The proposed technique is more effective because it uses ML models to develop an instance-based DSA method that identifies high-importance features to inform effective predictive countermeasures against FDI attacks in the SG system. The proposed algorithm’s viability is confirmed using test power systems on the real-time 128-feature IEEE 3-bus dataset and our developed Industrial Internet of Things (IIoT) dataset, namely the UKMNCT_IIoT_FDIA dataset (Published in Data in Brief; dataset available at: https://zenodo.org/records/14864902).