Junfei Qiao, Jiabin Zhuang, Haixu Ding
High-fidelity computational fluid dynamics (CFD) models can accurately describe the complex thermo–chemical processes in municipal solid waste incineration (MSWI). However, their high computational cost makes them difficult to apply in online monitoring, prediction, and control of industrial incineration systems. To overcome this limitation, this article proposes a fast and dynamic MSWI modeling framework that combines physics-based simulation with reduced-order modeling and data-driven learning to enable efficient temperature field prediction. First, a mechanism-based numerical model of a mechanical grate MSWI furnace is developed by coupling solid-phase waste combustion on the grate with gas-phase combustion simulated in Fluent, generating physically consistent temperature field data. Second, considering mechanism constraints, industrial controllability, and computational cost, key operating variables are selected and an orthogonal experimental design is employed to construct a representative simulation dataset. Third, proper orthogonal decomposition is applied to reduce the dimensionality of high-dimensional temperature fields, extracting dominant thermal modes and corresponding modal coefficients. Finally, an improved interval type-2 fuzzy neural network is used to establish the nonlinear mapping between industrial operating variables and temperature modal coefficients, enabling rapid temperature field reconstruction and short-term prediction. Simulation results show that the proposed method can effectively predict the dynamic evolution of the furnace temperature field within a 3 s horizon after operating condition changes, while reducing the full-field reconstruction time to approximately 10 s. The proposed framework significantly improves computational efficiency while preserving physical consistency, providing a practical solution for online modeling and intelligent operation of MSWI systems.