Detection method for producing areas of Xihulongjing tea based on genetic algorithm optimization

Abstract

The invention provides a detection method for producing areas of Xihulongjing tea based on genetic algorithm optimization. The method is characterized in that an electronic nose is utilized to acquire fingerprints of the smell of Xihulongjing tea from different producing areas, sensors having special contribution to modeling of the producing areas are screened out based on analysis of average fingerprints and a principal component score chart through three key techniques of genetic algorithm, i.e., a roulette selection method, a single-point crossover method and a simple mutation method, with an integral recognition rate as a fitness function, optimization is carried out so as to reduce the amount of sensors needed in modeling, and fingerprint response values of the sensors are subjected to matrix analysis through an F value obtained through variance analysis and principal component load.

Description

technical field [0001] This application relates to a method for optimizing the origin detection of West Lake Longjing tea based on a genetic algorithm. Background technique [0002] For a long time, sensory evaluation has been an important method to evaluate the quality of tea, but this method requires rich knowledge of tea science and evaluation experience. Unless they are professional tea reviewers, distributors or manufacturers, it is difficult for ordinary tea buyers to distinguish the quality of tea from good or bad. Without considerable experience, it is difficult to obtain reliable results. And training a tea judge not only needs to be carefully selected and invested a lot of money, but also the training period is relatively long. Moreover, even for professional tea tasters, the sensitivity of their sensory organs is easily changed by the interference of external factors, thus affecting the accuracy, objectivity and consistency of the evaluation results. For example...

AI Technical Summary

Problems solved by technology

[0013] (4) Signal drift and denoising: Due to changes in instrument measurement parameters, measurement methods, measurement environment, sample sources and other factors, it is easy to cause the sensor response curve to drift, causing errors in intelligent sensory detection, making it unable to adapt to industrialization for a long time Continuous operation, so it is necessary to strengthen research on reducing response signal drift and signal noise analysis and processing technology

[0014] (5) Robustness of the model: Some studies did not discuss the model in detail when establishing the discriminant model, nor did they use independent prediction samples to test the robustness of the model

Although the current optimal combination method uses the concept of combination to a certain extent, this combination is based on the preliminary elimination and combines the grouped sensor arrays, which does not achieve the effect of global optimal combination.

Although the Loading value method avoids the addition of redundant sensors, it does not analyze the response performance of the selected sensor, that is, the repeatability of the response of the same sensor to the same sample and the discrimination of the response to different samples.

Images