Study On The Monitoring And Analysis Method Of Ingestive-related Behaviors Of Dairy Cows Based On Triaxial Acceleration

By the end of 2019,the large-scale breeding ratio of dairy cows in China reached 64%,an increase of nearly 3 percentage points compared with 61.4% at the end of 2018.And the average product of dairy cows also increased from 7.4 tons in 2018 to 7.8 tons in 2019.It can be said that China has made remarkable achievements on the road of large-scale dairy cow breeding.At present,the dairy industry in western developed countries has been developing towards the direction of intensification,factorization and informationization.Informationization is of great significance to large-scale breeding,which can improve production efficiency,optimize production process and save a lot of manpower and material resources.In comparison,the application of information technology in dairy farming in China is still relatively backward.At present,the information equipments and technology depend on import,which is costly and unfavorable to China’s technological progress in this field.Therefore,it is necessary and urgent to develop our own application of information technology in dairy industry.In view of this,this paper proposed a method for monitoring and analyzing ingestive-related behaviors of dairy cows based on triaxial acceleration.The specific research contents include the following aspects:1.Design of a set of monitoring equipment for dairy cows’ ingestive-related behaviors based on triaxial acceleration.The device included control and sensing module,wireless transmission module and other circuit modules.The control and sensing module was mainly composed of STM32 microprocessor and its peripheral circuit,as well as triaxial accelerometer.The STM32 microprocessor controlled the triaxial acceleration sensor to collect the jaw acceleration data at a certain sampling frequency under the premise of ensuring the normal operation of the whole circuit.The wireless transmission module was mainly composed of BC35-G with NB-Io T technology as the core,which could send the collected data to the server at high speed through multiple network transmission nodes through UDP protocol.Other circuit modules were composed of debugging module,memory module and power module,which were respectively used to ensure the debugging of the circuit,the local storage of data and the normal power supply of the circuit.2.Triaxial acceleration data preprocessing and classification model construction.In this part,a whole set of preprocessing methods and classification models for triaxial acceleration data was proposed.Firstly,the triaxial acceleration data were synthesized,and then the time-domain and frequency-domain features were extracted.The extracted features were normalized and PCA dimensionality reduction operations were carried out.Then three machine learning algorithms,K-nearest neighbor,support vector machine and probabilistic neural network,were used to execute classification.During the process,the effects of different data segment lengths on the results were compared.The results showed that the combination of the data segment length of 256 and the K-nearest neighbor algorithm had the best classification effect.The precision,recall,specificity and the area under the curve of recognition for feeding were 92.8%,95.6%,96.1%,and 0.959 respectively.And those of recognition for ruminating were 93.7%,94.3%,97.5% and 0.959 respectively.3.Quantitative calculation and analysis of ingestive-related behaviors.In this part,a quantitative calculation and analysis method for the classification results of ingestive-related behavior was proposed.The basic thought was to eliminate the isolated points in the classification results and optimize the classification results through the combination of single-pointer scanning and actual cow behavior characteristics,so as to further obtain the actual ingestive-related behavior duration and frequency information of cows.Then,through linear regression fitting,MSE,and RMSE the consistency between the results obtained by the algorithm and the actual value was judged.The results showed that the estimated value of the algorithm was in good agreement with the actual value,that is,the algorithm can calculate the frequency and duration of ingestive-related behavior of cows accurately.To sum up,this study designed a set of monitoring equipment based on triaxial acceleration for ingestive-related behaviors of dairy cows that could meet the needs of large-scale dairy cow breeding,and proposed effective monitoring and analysis methods for ingestive-related behaviors of dairy cows.This study provided a reliable technical support for the informatization transformation of the dairy industry,and provided new ideas and new methods for further monitoring and early warning of the health and welfare of dairy cows.