Optimization strategy for black tea digital blending by fusing image and spectral information

Area of Science:

• Food Science
• Agricultural Engineering

Background:

• Tea blending is crucial for product standardization and sensory quality.
• Traditional methods rely on experienced masters and subjective sensory evaluation.
• Objective, data-driven approaches are needed for consistent tea production.

Purpose of the Study:

• To develop a digital black tea blending model using machine vision and near-infrared spectroscopy.
• To optimize tea blending considering both cost and quality parameters.
• To validate the digital blending model against professional sensory evaluations.

Main Methods:

• Machine vision captured image information (golden pekoe, color, cord).
• Near-infrared spectroscopy analyzed chemical components for taste evaluation.
• Multi-objective optimization algorithms (GDE3, NSGAII, NSGAIII) were employed.
• The GDE3 algorithm demonstrated superior convergence speed and degree.

Main Results:

• A digital black tea blending model was successfully developed and implemented.
• The model integrated visual and spectral data for objective blending decisions.
• The GDE3 algorithm proved effective for optimizing the multi-objective blending problem.
• Results were validated through professional sensory evaluations, confirming model efficacy.

Conclusions:

• Digital blending offers a novel and effective method for standardizing black tea.
• The integration of machine vision and spectroscopy enables objective quality control.
• This approach has the potential to revolutionize large-scale, consistent tea production.