Pit mud total acid prediction model building method based on hyperspectral image technology

Abstract

The invention discloses a pit mud total acid prediction model building method based on a hyperspectral image technology. Firstly, hyperspectral image information of infrared and visible light wave bands is acquired, average spectral information of an ROI (region of interest) is extracted through ENVI software, errors caused by non-uniform illumination are eliminated through black and white correction, the average spectral reflectivity of each pixel in an image is estimated, quantitative predication models based on full wave bands and characteristic spectra are built respectively through no processing, SNV (standard normal variate) and different modelling methods, and the best model for quantitative predication of pit mud total acid is screened by comparing and analyzing determination coefficients and root-mean-square errors of a training set and a testing set of each fitting model. The method provides powerful technique support for transformation and upgrading of baijiu brewing industrialization and digital and intelligent online real-time monitoring.

Description

technical field [0001] The invention relates to a method for establishing a prediction model of pit mud total acid based on hyperspectral image technology, and belongs to the technical field of solid-state fermentation index detection. Background technique [0002] Solid-state fermentation technology has a long history in my country and is also a unique brewing process in my country. In the liquor brewing industry, the pit mud is the basis of fermentation, in which the beneficial microbial flora dominated by caproic acid bacteria uses the nutrients in the grains to produce acid and ester, and ethyl caproate is the main body of the Luzhou-flavor liquor. Therefore, the pit mud plays a decisive role in the quality of Luzhou-flavor liquor in the production process, and it is more important to evaluate the quality of the pit mud. Without good cellar mud, it is impossible to produce good quality wine. [0003] At present, many wineries are facing the situation that the acidity i...

AI Technical Summary

Problems solved by technology

[0004] The deficiency of the patent: the patent detects the pH value, that is, the concentration of free hydrogen ions, so the pH is closely related to the type and content of acidic substances, and the characteristic absorption performance of the fermentation substrate to the spectrum is also reflected by the coloration carried by various acidic substances. Group performance, because the optical properties of a substance are generally expressed through functional groups, and different functional groups have different absorption strengths for different wavelength bands, which are related to the molecular structure

According to the pre-experiment and data query, the total acid shows its optical properties mainly through the carboxyl group, and has strong absorption in the near-infrared band, while the patent only collects images and spectral information in the visible light region; no full spectrum Compared with the characteristic spectrum, the spectral information corresponding to the characteristic wavelength is directly selected as the input variable of the model, which is easy to miss important information; without preprocessing and denoising, it is easy to include non-sample signals into the calculation; the modeling method is also relatively simple

Images