Attention-enhanced partial least squares prediction method for sensory evaluation of strong-flavor base liquor
By constructing an attention-enhanced partial least squares prediction model, the problems of subjectivity and insufficient precision in the sensory evaluation of strong-aroma raw liquor were solved, realizing the objective quantification and intelligent management of sensory quality, and improving the accuracy and stability of prediction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ANHUI GUJING DISTILLERY CO LTD
- Filing Date
- 2026-04-29
- Publication Date
- 2026-07-21
AI Technical Summary
Existing technologies for sensory quality evaluation of strong-aroma raw liquor suffer from problems such as strong subjectivity, difficulty in quantification, computational redundancy, and insufficient model prediction accuracy, making it difficult to meet the standardized control and intelligent management of the production process.
An attention-enhanced partial least squares prediction method for sensory evaluation of strong-aroma raw liquor is adopted. Through residual update and weight recursive optimization mechanism, combined with attention weight allocation module, the intrinsic relationship between flavor characteristics and sensory attributes is explored, and a flavor perception partial least squares regression network is constructed to quantitatively predict sensory attributes.
It improves the accuracy and stability of sensory attribute prediction, realizes the objective quantification of sensory evaluation, reduces the impact of individual differences and state fluctuations, and improves the repeatability and comparability of evaluation.
Smart Images

Figure CN122432636A_ABST