Given that power systems are essential to modern life and electricity demand continues to rise, ensuring their reliable and secure operation has become a critical priority. Wireless networked control systems (WNCSs), which rely on wireless channels for communication between controllers, sensors, and actuators, are increasingly deployed in energy systems such as multi-area interconnected power systems to enhance flexibility and scalability. WNCSs are susceptible to deception attacks and time-varying communication delays that can compromise interconnection stability and deteriorate performance. This paper presents an observer-based secure control methodology that models deception via independent Bernoulli processes with unknown attack probabilities, while explicitly considering actuation and measurement delays. Using a Lyapunov stability framework, we established computationally feasible linear matrix inequality conditions enabling the co-design of the controller and observer with proven stability and disturbance rejection. A two-area interconnected power system case study validates the approach. The proposed method was tested with offline gains covering nine scenarios. Results indicate that the method sustains closed-loop performance across all nine combined attack/delay scenarios and recovers quickly even in worst-case conditions, supporting secure control of WNCSs in realistic adversarial environments.