Method and system for predicting outlet moisture in tobacco shred drying process based on KAN (Karan Area Network)

The method for predicting the moisture content at the outlet of tobacco shreds based on KAN network solves the problems of traditional models' difficulty in modeling the internal mechanism of the drying process and the lack of interpretability of deep learning. It achieves high-precision prediction of the moisture content at the outlet of tobacco shreds and is suitable for the process control needs of industrial sites.

CN122286147APending Publication Date: 2026-06-26CHINA TOBACCO YUNNAN IND
0 Cites 0 Cited by

Patent Information

Application Number
CN202610365573.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-03-24
Publication Date
2026-06-26

AI Technical Summary

Technical Problem

Traditional soft measurement models are difficult to effectively model the intrinsic mechanism of the wire drying process, and deep learning methods lack interpretability, making it difficult to provide a reliable basis for process analysis and operation adjustment.

Method used

A method for predicting the moisture content at the outlet of tobacco shreds during the drying process based on KAN network is adopted. Key process variables are screened through the maximum correlation minimum redundancy algorithm. Combined with the dynamic normalization strategy of sliding window and sparsity regularization constraint, a KAN prediction network integrating SiLU basis function and parameterized B spline function is constructed. The network parameters are iteratively optimized and the generalization performance is verified to achieve real-time prediction of the moisture content at the outlet of tobacco shreds.

Benefits of technology

It achieves high-precision and high-reliability prediction of moisture content at the outlet during the tobacco drying process, provides a preliminary reference for process control, compensates for the lag of direct measurement, and adapts to the continuous production needs of industrial sites.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a cut tobacco drying process outlet moisture prediction method and system based on a KAN network, relates to the technical field of tobaccos, and aims to anchor key process parameters associated with cut tobacco drying outlet moisture, avoid redundant variable interference and adapt to cut tobacco drying multivariable strong coupling working condition characteristics. Wherein in the step S1, key process variable screening is completed, and the strong relevance between input features and a prediction target is guaranteed; s2, strengthening model working condition robustness through standardization processing adaptive to industrial data characteristics; the network structure constructed in the step S3 breaks through the nonlinear expression limitation of a traditional fixed activation function; s4, optimizing model parameters, and improving generalization performance and interpretability; s5, verifying the generalization ability of the model, and guaranteeing the prediction stability under unseen working conditions; and S6, real-time prediction of outlet moisture is realized, and detection hysteresis is made up. The high-precision and high-reliability prediction of the moisture at the cut tobacco drying outlet is integrally realized, and the prediction performance and the industrial application value are both considered.
Need to check novelty before this filing date? Find Prior Art