Enhancing the future adaptability of boiler systems using multi-input extremum seeking control algorithms

Traditional boiler control systems demonstrate inherent limitations in sustaining operational efficiency under frequent load fluctuations and alternative fuel co-firing conditions, primarily attributable to open-loop optimization architectures and delayed response to automatic generation control (AGC) signals. This investigation develops a Multi-input Extremum Seeking Control (MESC) algorithm that dynamically optimizes secondary air distribution parameters (specifically auxiliary air/close-coupled over-fire air/separated over-fire air) through a closed-loop Simulink-APROS integration framework, circumventing conventional reliance on prior system modeling. Experimental validation under 1000-900 MW flexible operation scenarios demonstrate a 1.53 % reduction in the composite objective function jointly evaluating coal consumption rate of power supply (CCR) and nitrogen oxide (NOx) emissions. The proposed method achieves rapid convergence near optimal values within 1800 s even under incomplete combustion and fuel quality disturbances. The algorithm demonstrates effective trade-offs between CCR and NOx concentration. Under load variation scenarios and coal quality disturbances, it achieves 27 % and 25 % NOx reductions respectively, while prioritizing optimization of CCR under suboptimal initial operating conditions. Parametric analysis of weighting factors reveals increased coal pricing amplifies prioritization of CCR optimization, necessitating strategic equilibrium between economic objectives and emission constraints. By enabling model-free real-time self-optimization, this novel approach enhances operational resilience of coal-fired units in renewable-penetrated power networks, offering a practical solution to evolving decarbonization mandates.