Research On Water-injected Meat Detection Method Based On System Related Identification

Water-injected meat(WIM)refers to meat,such as pigs,cattle,sheep,which are forcibly injected with water before and after slaughter to increase the weight of the meat.Traditional methods such as sensory method and test paper method have low accuracy,while quantitative analysis methods such as drying method and near infrared spectroscopy can not be applied in slaughter,storage and transportation,trade field.Rapid identification of Water-injected meat has always been a difficult problem to be solved urgently.This thesis proposes a WIM detection method based on system correlation identification.The method uses the m-sequence as the excitation signal,finds the time-domain impulse response of the meat according to the WienerHe equation,and uses the support vector machine algorithm to identify the water-injected meat.The main research contents include:1.Starting from the principle of m-sequence generation,this thesis introduces its basic properties:periodicity,binaryness,run-length,autocorrelation approximation to unit impulse function,strong anti-interference ability and so on;research system correlation identification principle,m sequence as an excitation signal,the system’s time-domain impulse response can be obtained by finding the cross-correlation function of the input and output signals to reflect the system behavior.In this thesis,the system behavior refers to the quality of the meat.2.A set of FPGA-based WIM hardware detection platform is designed.The platform uses Xilinx Artix7 series FPGA chip XC7A100T-2FGG484I as the control core,and uses FPGA to generate excitation signal m sequence.After bipolar conversion and voltage control current source module is converted into current.The excitation signal is injected into the meat to be tested;the AD sampling chip AD9248 is controlled to synchronously sample the excitation signal and the response signal;and the cross-correlation operation of the excitation and response signals can obtain the time domain impulse response of the system.3.Take 15 fresh pigs,cows,sheep and chickens respectively,take 15 pigs,cows and mutton each,and find 5 test points for each piece of meat.Insert the measuring electrodes into the test points and collect 2 times for each test point.Real-time excitation signal and response signal,so that each piece of meat can measure 10 sets of excitation signal and response signal,and 150 sets of excitation signal and response signal are measured for each meat.A total of 105 pieces of meat were measured and 1050 sets of data were obtained.The respective time domain impulse responses are obtained separately.4.Support vector machine(SVM)was used to establish the WIM identification model of pig,beef and mutton respectively,and the identification model of meat type was established for fresh pigs,sheep,cattle and chickens.Firstly,the meat was labeled,then the training samples and test samples were selected.In order to ensure the stability of the model,each model was established five times.After the model is built,the original SVM classification model is judged by using the confusion matrix to compare the original mark and prediction mark of the test sample,and the correct rate,sensitivity and specificity of the pork and beef WIM recognition model are obtained.The average of five times is 100%,and the five average values of the accuracy,sensitivity and specificity of the mutton meat-in-the-water identification model are 92.44%,91.28%,and 93.53%,respectively,and the four meat products in the meat cultivar recognition model.The average value of the correctness rate is 100%.In this thesis,based on the characteristics of m-sequence pseudo-random signals,a method of water-injection detection based on system-related identification is designed and developed,and the classification model is established by SVM to realize the identification.The feasibility of system-related identification principle for water-injected meat identification is preliminarily verified...