THz spectral analysis-based rapid nondestructive testing model for K value of fresh meat

Abstract

The invention discloses a THz spectral analysis-based rapid nondestructive testing model for a K value of fresh meat. The model comprises the modeling steps of (1) construction of a multi-layer back-propagation neural network BP-ANN, namely sampling a THz spectrum of a fresh meat sample by adopting THz attenuated total reflection ATR detection mode and treating sample spectrum data and regarding the BP-ANN as a nonlinear function, wherein an input value and a predicted value of the BP-ANN are an independent variable and a dependent variable of the nonlinear function separately and the BP-ANN expresses the function mapping relationship from the independent variable to the dependent variable; and (2) training which comprises network initialization, hidden layer output calculation, output layer output calculation, error calculation, weight updating, threshold updating, and judging whether algorithm iteration is completed or not according to the iteration times and the final iteration error. A relationship model between the THz spectrum data and the freshness K value of the fresh meat is constructed through exploring the THz spectrum characteristic of the freshness of the fresh meat.

Description

technical field [0001] The technical solution belongs to the technical field of fresh meat detection, and is specifically a fast non-destructive detection model of K value of fresh meat based on THz spectral analysis. Background technique [0002] Meat is the main source of intake of protein and fat for people. The freshness of fresh meat is especially important in food safety. Taking pork as an example, it is the main type of meat consumption in my country. The consumption of pork in my country accounts for 77% of the total meat consumption. How to control the quality of meat in the production and sales process and ensure food safety has been highly recommended by manufacturers and consumers. And the national quality inspection department pays close attention to it. The quality of fresh meat such as pigs, cattle, sheep, and fish includes elements such as nutrients, flavor, tenderness, water retention, and freshness, among which freshness is an important aspect. [0003] T...

AI Technical Summary

Problems solved by technology

However, during the experiment of the near-infrared spectrometer, the sample chamber needs to be evacuated, and online rapid detection cannot be realized. Although the fiber-optic spectrometer does not need to be evacuated, it needs to use optical fibers to collect the spectral information of the sample. The coupling optical fiber is small and the detection area is limited. Unable to achieve a one-time comprehensive collection of the sample surface

Although visible-near-infrared hyperspectral image analysis technology can achieve comprehensive acquisition, it is necessary to calibrate the spectral acquisition system with a standard reflectance calibration plate before each test and during the long-term test to prevent the temperature of the camera photosensitive device The impact of drift and brightness changes of the light source on the detection system is not conducive to online continuous detection

The electronic nose technology uses a gas sensor to collect the spoilage smell of meat, and measures the freshness of the meat based on the principle that the deeper the spoilage, the stronger the odor release. Time, so that the odor distribution in the detection chamber is even, the odor concentration reaches the detection lower limit of the sensor, and the waiting time for detection is long

Images