Addressing the challenge of real-time multi-parameter coupled optimization in cogeneration unit combustion systems

this study proposes a digital twin-based hybrid modeling approach integrated with offline global optimization and online case-based reasoning for real-time optimization. First

a mechanism-data fused digital twin model is constructed for the combustion system of a circulating fluidized bed cogeneration unit

establishing a high-fidelity real-time simulation platform. Deviations between model predictions and measured values remaine within ±1% of instrument full scale. Crucially

computational errors for core combustion stability parameters

including steam drum pressure and oxygen content

are below 0.7%

with dynamic process mean relative error under 0.6%. Then

this paper develops an energy efficiency assessment index incorporating feedwater characteristics to provide a unified benchmark for performance analysis across diverse operating conditions. Finally

a hybrid offline global optimization and online case-based reasoning framework is established. Genetic algorithms are employed offline so as to build an optimized case library for typical operating scenarios. The case-based reasoning algorithm then performs real-time case retrieval and matching

delivering operational recommendations via the UWinTech platform. Results demonstrate a 1.03% improvement in system energy efficiency and a 3% increase in average coal-to-steam ratio (standard coal equivalent)

while critical operating parameters including furnace pressure and bed temperature remained within safe thresholds. This methodology provides an engineerable digital twin solution for enhancing cogeneration unit safety and economic operation.