Intelligent Processing And Analysis Of Abnormal Monitoring Data Of Sensor Networks On Farmland

Agricultural informatization is an important part of the national economy informationization,and its development plays a significant role in promoting the mechanized production and intelligent management of agricultural production in China.Internet of things technology provides a new solution to farmland information acquisition,and farmland field sensor networks is a typical application of Internet of things.It is convenient to realize real-time crop growth monitoring in a long period with the deploying sensor networks on farmland.Moreover,with the application of farmland sensor network,the field management will be more precise and smart.Since the farmland sensor nodes are always deployed in an open field environment,it is inevitable that the collection of sensed data will be affected by the severe environment.And it will bring in noises and outliers in acquired data.According to the data receiving process and characteristics of the sensed data from Smart Farming System developed by National Engineering and Technology Center for Information Agriculture(NETCIA)in Nanjing Agricultural University,this paper designs and implements an "aggregation-decomposition-reconstruction" process of abnormal monitoring data detection method for farmland sensor network.Outlier detection of spectral monitoring data is conducted with LOF algorithm,and reconstruction of miss values are achieved by EMD.The proposed methods were tested and applied on the rice and wheat growth monitoring data from farmland sensor networks deployed in Rugao and Xinghua of Jiangsu province.The results shows that the processed data is more accurate and reliable to reveal the characteristics,trends and the patterns of crop growth.The technique of open RESTful web service is helpful in sharing monitoring data of farmland sensor networks,and is an available technique for integrating and updating of existing agro-software.This paper builds a RESTful web service interface with Node.js framework and Python programming language to realize the functions of abnormal monitoring data detection.The monitoring data is stored in a Microsoft SQL Server database,and a WebGIS visualization strategy is also utilized to visualize the aforementioned process.