Research On Data Stream Processing Method For Breeding Environmental Monitoring

With the rapid development of the wisdom agriculture, the Internet of things, networking, cloud computing, and other technologies, the massive data emerge greatly in agricultural field. It means that the information society has entered the era of big data. Big data has two forms. One is static data, the other is data stream. Data stream that generated from various sensors in the wisdom agriculture is the main source of agricultural big data. Big data in agriculture has many features,such as numerous, diverse forms, complex.It’s difficult to use single model to solve all of demands.Now how to establish a high-level model according to the different data characteristics and calculation characteristic is the primary problems..In this paper, a series of stream computing work focused on clustering, traceability, prediction and modeling are studied.The main research contents and conclusions were as follows:(1) When the input datas are continuously random variables, the existing clustering method based on discrete random variables can not meet the requirements of efficiency and accuracy. In order to solve the problem mentioned above,a new method named cumicro algorithm is proposed. First, the Gaussian mixture model as the basic representation of uncertain data streams was used.Second, a clustering method which can find clustering in time dimension is proposed. This method can make up for the deficiency of traditional clustering which can’t find the non-spherical clustering. Finally, the compared result shows that the proposed algorithm promotes the accuracy of clustering.(2) For the existing traceability model could not express the mixing process quantitatively in the context of big data stream, the uncertain data were brought into traceability system and the model based on these uncertain data was built up.A traceability inquiry method using uncertain data was proposed.The multi-source tracing problem was solved by using basic representation and inquiry method of uncertain data. The functions of simple inquiry, node evaluation and single node abnormal deduction were achieved based on the proposed model and the solving method of multi-node abnormal deduction was presented.(3) A prediction methods based on the GM(2,1) was proposed to predict the data stream value. While the updating cycle of sliding window increasing,the prediction success ratio decreased. While the sampling frequency changing, prediction success rate decreased. While the width of future data window increasing, the mean absolulte deviation increased.The experimental results showed that the proposed algorithm is suitable for the recent prediction.(4) A data stream collection model of pig breeding was designed. In the big data stage, business research should be carry out in the center of data.The research not only analyse the time domain, but also consider the influenc while the calculation process and physical process through the network real-time interaction.