[Objective] With the increasing integration of renewable energy into power systems

coal-fired units are required to operate flexibly over a wide load range in response to frequent and large-scale load fluctuations

placing higher demands on the dynamic performance of process control. However

the prevalent high-order

large-inertia characteristics of thermal processes often lead to significant response lag

making it difficult for conventional control methods to effectively improve regulation performance. To compensate for the deficiencies of conventional control methods in dynamic performance and disturbance rejection capability for such processes

this paper proposes a lag compensation based active disturbance rejection control(LC-ADRC) scheme. [Methods] The scheme adopts the active disturbance rejection control(ADRC) method as basic control framework

leveraging its well-established robustness and superior disturbance rejection capabilities. Through a theoretical analysis of ADRC applied to highorder

large-inertia thermal processes

the inherent lag phenomenon and its mechanistic causes are systematically investigated. Based on this

and drawing inspiration from the Smith predictor structure commonly used for handling time-delay processes in conventional control

a dedicated parameter tunable lag compensation structure is innovatively established. Given that thermal processes are predominantly governed by low-frequency dynamics

a low-frequency approximation technique is adopted to formulate practical design guidelines for the compensation parameters

with validation provided through frequency-domain analysis. By integrating the compensation structure with the ADRC structure

the novel LC-ADRC scheme is proposed. Furthermore

using a two-degree-of-freedom equivalent transformation

the relationship between desired closed-loop performance and tunable controller and compensator parameters is analytically established. Supported by robustness analysis

a standardized parameter tuning method based on desired response performance is derived to guide practical implementation. [Results] Comparative simulation results demonstrate that under nominal conditions

the proposed method improves the quantitative disturbance rejection metric by 49.6% over conventional PI control

outperforms two existing ADRC methods

and is only 8.4% lower than that of model predictive control. The tracking performance metric is improved by more than 30% compared to existing ADRC schemes

by 38.2% over conventional PI control

and by 9.1% over model predictive control

with an overshoot of less than 1%. Under two typical parameter perturbation scenarios

the control system remains stable

and the degradation in core performance metrics is kept within 20%. [Conclusion] In summary

compared with conventional algorithms

the proposed method demonstrates improved performance in both tracking response and disturbance rejection

while exhibiting robustness against model uncertainty and parameter perturbations under complex operating conditions. Moreover

in contrast to computationally intensive model predictive control

the proposed method achieves comparable performance at a significantly lower deployment cost

demonstrating favorable engineering applicability and practical deployment potential.