A Unified Control and Fault Diagnosis Framework for Single-Area Load Frequency Control: Addressing Sensor Bias Fault Insensitivity in Integral Control Approaches

This paper presents a unified control and fault diagnosis (UCFD) framework for single-area load frequency control (LFC) systems subject to load disturbances and sensor bias faults. Although integral-based LFC can eliminate steady-state frequency errors under nominal conditions, persistent sensor biases may be interpreted as genuine frequency deviations, resulting in erroneous control actions. To address this limitation, the proposed framework replaces integral action with a state-feedback and feedforward control structure and employs two dedicated observers: an augmented-state observer for reconstructing the system states and sensor bias as an additional state, and a residual-generation observer for generating a residual associated with load disturbances. By explicitly incorporating the measured load disturbance into the augmented observer, the proposed approach separates physical load variations from sensor-induced measurement errors and enables unbiased state reconstruction. The residual-generation observer, together with an auxiliary first-order compensator, is then used for load-disturbance rejection, while sensor bias accommodation is performed at the state-estimation level. An analytical expression for the compensator gain is derived from the steady-state disturbance-rejection condition, while its time constant is selected to provide appropriate separation between observer and compensation dynamics. The proposed framework is evaluated through four simulation cases involving load disturbances and sensor bias faults with different temporal profiles, including a combined scenario with model-parameter uncertainty and stochastic load demand. The results are compared with an integral state-feedback approach using frequency-regulation and control-effort metrics. Although integral control provides slightly better nominal performance, the proposed framework achieves substantially improved frequency regulation under faulty conditions and maintains zero steady-state frequency deviation in the presence of persistent sensor bias. Moreover, under faulty and uncertain conditions, the proposed framework reduces both the cumulative squared control effort and the peak control demand, demonstrating enhanced resilience to sensor-bias faults without imposing excessive actuator requirements.