As tactical and information technologies become more integrated, cyber risks to organisational power systems become more complicated. Ransomware, DDoS attacks, insider threats, and advanced ongoing attacks on vital systems can’t be stopped with traditional defences. This research talks about preventative security methods that use cryptography, AI to find intrusions, and zero-trust designs. In addition, mathematical models like statistical and game-theoretic frameworks, as well as attack-defense modelling methods, are used to measure and lower risks. Results from experiments show that there is better robustness, less delay, and more accurate identification. The suggested structure improves privacy, availability, and integrity, making sure that the power system can work reliably even as hacking problems change.