The use of Smart Grid (SG) technologies has modernized and improved the current electrical infrastructure. Modern computer, control, and advanced networking technologies form the backbone for SGs. The Information and communication technologies (ICTs) have been incorporated into the SG, which utilizes the conventional electrical power system. By empowering both the suppliers and users of electrical utilities, whereas an integration enhances the effectiveness and the power system availability as well as continuously monitors, governs, and manages client needs. Cyber-physical SG systems need to be secure from evolving security risks and intrusions. In SGs, flooding attacks and Distributed Denial-of-Service (DDoS) attack have received the greatest attention. These hacks have the potential to compromise SGs' efficient operation and cause significant financial losses, damage to equipment, and malicious regulation. Several researchers have reveal that Machine Learning (ML) techniques outperform traditional attack detection algorithms in detecting attacks. The ML approaches were used to examine malicious activity and intrusion detection challenges at the network layer of SG communication system. This paper focuses on ML algorithm that have been utilized for classifying the measurements as either being attacked (Unstable) or secured (Stable) and it can be obtained through bat sequence algorithm. The dataset used in this research is Electrical Grid Stability Simulated Dataset has been used to classify the stability assess in SGs and try to minimize the power loss by avoiding the unstable (attacked) distribution. The accuracy of proposed bat sequence algorithm using Artificial Neural Network (ANN) is 99.43% which is higher while compared to other classification method.