Abstract：This paper proposes a digital twin technology based on numerical simulation and feature extraction, validated using a cement decomposition furnace as an example. To address the challenges of sensor monitoring in harsh industrial environments, the study combines CFD(Computational Fluid Dynamics)simulation and reduced-order modeling techniques, introducing a digital twin construction method based on POD(Proper Orthogonal Decomposition)feature extraction and Artificial Neural Network(ANN)surrogate modeling. Physical field data under various operating conditions are obtained through parameter design and high-precision simulation. The POD method is used for dimensionality reduction of high-dimensional data, and ANN is employed to map parameter groups to POD modal coefficients, achieving efficient and accurate flow field predictions with errors below 4%. Additionally, Gaussian spatial interpolation is applied to downscale mesh resolution, enabling model lightweighting and facilitating rapid deployment of the digital twin. The results demonstrate that this method is highly applicable and scalable, making it suitable for monitoring, fault diagnosis, and lifespan prediction of complex industrial equipment, providing technical support for refined equipment management and IoT applications.
水 沛. 基于特征提取与降阶模型的分解炉数字孪生构建与应用[J]. 水泥杂志, 2025, 0(8): 43-.
SHUI Pei. Construction and application of calciner digital twins based on feature extraction and reduced order model. Journal of Cement, 2025, 0(8): 43-.
2025, Vol. 0 
