Malicious traffic attacks can reduce the timeliness of normal data flows by seizing the limited bandwidth. For networked load frequency control (LFC), traffic attacks will cause stability margin decline, frequency deviation rise or even over-limitation accidents. In most existing control schemes, a single and fixed controller is used to ensure asymptotic stability under maximum attack intensity. This has disadvantages such as many design constraints and great conservatism. A novel LFC strategy is proposed to adaptively adjust the controller gain according to attack intensity. First, based on deterministic network calculus, the upper data transmission delay bound in a no-attack scenario is obtained. Then a series of transmission delay ranges representing different attack intensities are preset. Second, by constructing a Lyapunov function, controller design criteria for each attack intensity are derived. Finally, in a switched control framework, the maximum attack switching frequency that the proposed adaptive scheme can tolerate is derived. Simulation results show that compared with the existing schemes, the frequency deviation amplitude with the proposed method can be reduced by 12.60%, and the integral of the absolute error of the area control error can be decreased by 10.85%.