Research On The Diagnosis Algorithm For Livestock Health Based On Intelligent Collar

The health of livestock is closely related to the economics of the pasture.Traditional monitoring of livestock health still relies on human judgment.With the expansion of farming scale and the scarcity of talents with knowledge of husbandry and veterinary,this method is no longer applicable.Therefore developing the automatic monitoring of animal vital signs is the direction of future development of husbandry.In order to accurately judge the health status of livestock,this thesis studies and designs a diagnosis algorithm for livestock health based on intelligent collars,which can provide a certain reference for farmers' decision-making.The main work of the thesis is as follows: Firstly,by studying the changes of vital signs with the physiological changes of livestock,it is decided to take the body temperature,ruminate time and exercise quantity of the animals as characteristic parameters,and use the linear regression algorithm based on least squares method,peak detection algorithm and K-means clustering respectively.These algorithms extract and corrects the original data,and the results are used as the input to the health diagnosis algorithm.Secondly,the diagnosis model for livestock health based on SVM support vector machine is constructed.Compared the accuracy of the prediction of the test under different normalization modes and different kernel functions obtained by experiments,finally determines that the data should be normalized on [0,1] and the SVM diagnostic model should be based on linear kernel function.In order to improve the performance of the classifier,the CV method is used to optimize the penalty parameter C and the kernel function parameter g,so that the accuracy of the test set prediction classification has been significantly improved.Finally,the STM32-based hardware test platform was designed and built to test its performance.The data collected by the hardware platform was used to test the livestock health diagnosis algorithm,and the test results were analyzed.