Recognition Of Behavioral Characteristics Of Beef Cattle Based On Posture Sensor

With the development of large-scale and intensive beef cattle breeding, the implementation of animal behavior recognition is especially important. The actual experience now observed on beef cattle behavior mainly depends on the degree of understanding the breeder, with their cattle, to judge the health and behavior characteristics of beef cattle, which not only increased the burden of the breeder, and the work efficiency is low, the behavior of the individual state change of beef cattle are not fast enough, is not suitable for the large-scale farm demand. In recent years,the continuous development of information technology, precision breeding of beef cattle to provide technical support, observation and monitoring the behavior of the image processing technology of beef cattle with the help of related workers all over the world, however, these techniques require relatively high on the environment.Based on the attitude sensor as a tool for fattening beef cattle as the main body, the behavior data of beef cattle for dimension reduction through principal component analysis, using the Calman filter algorithm, BP neural network algorithm to establish the classification model of beef cattle behavior. By constructing a typical data model of daily behavior, we can identify and monitor the behavior characteristics of beef cattle, and build a prediction model for beef cattle diseases. The main contents include:(1) according to the data of the attitude sensor on beef cattle were collected and analyzed by principal component analysis algorithm, the data characteristics of beef cattle 9 behavior set for dimensionality reduction, data reduction can be obtained on behalf of all behavioral characteristics of beef cattle, in the modeling calculation and can reduce the amount of calculation and increase measurement accuracy.(2) the behavior data of Calman filter algorithm on beef sets iteration, to explore the relationship between the degree of influence and the classification model between the first attempt, this is the Calman filter in terms of classification behavior of beef cattle.With beef cattle activity occurred at a time, researched and analyzed the attitude data model correspond to the characteristics of daily behavior of beef cattle, beef cattle through the classification behavior of Calman filter rate was 62.5%,the recognition rate is low, that the linear system in the beef cattle for recognition can not be used as theoretical support, Research on beef cattle behavior classification the need for further research in nonlinear systems.(3) the identification using Calman filter to beef cattle behavior classification rate is too low, not enough to act as the theoretical basis of beef classification, this paper also uses BP neural network algorithm research on beef cattle behavior classification results show that the BP neural network algorithm in beef cattle for classification than the Calman filter realized many better classification to improve the accuracy of nearly 30% based on Calman filter algorithm, the results show that the classification model of beef cattle behavior accuracy in using nonlinear system has greatly improved, the method can be used as the theoretical support in the field of beef cattle behavior recognition.(4) the deployment of sensor in beef cattle body position and record the time interval of the corresponding study research shows that: the sensor placed in the temporal position of beef cattle, the actual behavior of activity measured sensor data can satisfy most cattle; without affecting the classification accuracy of beef cattle behavior under the condition of the sensor records the time interval is 2S can reduce the amount of calculation, accelerate the speed of data processing.