Research On Cow Behavior Discrimination System Based On Wireless Sensor Network

To improve the welfare of dairy cows,improve the production efficiency of dairy industry and reduce the production cost of dairy industry through scientific breeding management had become the goal of dairy industry development.In view of the problems of lagging management methods,low degree of automation of disease monitoring and unable to determine the sudden behavior of cows in time,a discriminant system for dairy cow behavior based on wireless sensor network was developed in this study.The 3 layers mode had been adopted in this system,and through the verification of the hardware design,software part design,algorithm development,software development,and field test,the dairy cow sports information and basic information were collected,stored,managed and displayed.It had strong stability and high accuracy and could realize the effective organization and management of the data.(1)The hardware design follows the design principle of low power consumption,high detection sensitivity and strong running stability.It integrated three axis acceleration sensors ADXL345,processor M430-F149,wireless transceiver CC1101 and so on,which could meet the requirements of accurately collecting the acceleration data of cow motion,long term reliable transmission data and so on.At the same time,the transmission performance test of wireless sensor nodes was completed,and the optimal transmission distance and the optimal node height was determined.The software part was designed by PHP,Javascrip and My SQL,and software function module,database table structure,interface and other parts was designed.(2)A semi supervised fuzzy clustering(FCM,Fuzzy C-Means)algorithm used for real-time behavior discrimination of dairy cows was proposed.The algorithm combined the advantages of supervised learning algorithm and unsupervised learning algorithm,which could effectively improve the accuracy of cow behavior discrimination.Compared with the fuzzy clustering algorithm,K-means algorithm and BP neural network algorithm,the experimental results showed that the semi supervised fuzzy clustering algorithm could effectively improve the accuracy of cow behavior discrimination,and it was characterized by high accuracy,low learning complexity and fast running speed.(3)In the actual experiment,it was found that the accuracy of the algorithm was low for the behavior of eating and standing,so the similarity positioning algorithm based on the D-S evidence theory fusion algorithm were introduced,and the discriminant results of semi supervised fuzzy clustering algorithm and dairy location data information were used as evidence sources,and a single model was established by using the combination rule to make evidence fusion.And semi supervised fuzzy clustering algorithm was used to reclassify the discriminating results of eating and standing behavior.Reclassification results showed that this method could effectively improve the accuracy of eating and standing behavior in leg installation mode.(4)The B/S architecture was adopted and the system software was written in combination with the discriminant algorithm.The development environment with strong stability,high efficiency and mature technology was used to develop and test in detail.The test results showed that the system has characteristics of strong portability,convenient function upgrading,simple and beautiful operation interface,and the clear display results.