Research On Key Technologies Of Internet Of Things For Monitoring Wheat Growth

Given its long growth cycle and wide planting range of environmental conditions, wheat is sensitive to stresses, disasters and diseases, which leads to the reduction in its yield and quality. The growth and development of crops follow not only internal processes but also external influences. The corresponding environmental conditions result in physiological and physical responses of plants, which are highly relevant to the environmental factors of crop growth besides the factors of wheat genes and varieties. Consequently, an approach that uses timely environmental information in the field to estimate the growth status and identify the stresses, disasters and diseases of crops is feasible. Obtaining wheat information in a timely and accurate manner to raise the value of the monitoring information from original raw data for decision support and improving the capability of the quantification analysis and diagnosis of plant growth, disasters and diseases remain a research challenge and considerable research area.On the basis of the existing Central Monitoring Platform of IoT of Wheat, this study carried out further research in the process of data acquisition and analysis using multi-source information fusion method and technology. Aiming at data acquisition of perception layer of IoT, the key technologies of Wireless Sensor Network (WSN) and in-network data fusion were proposed and developed. And aiming at data analysis of application layer of IoT, three titles of diagnosis methods were further studied which is wheat seedling classification based on data fusion from sites to areas, wheat agrometeorological disasters diagnosis based on models coupling with IoT, and wheat leaf diseases diagnosis based on image processing and support vectors machine (SVM). Finally, the above four prototype systems were designed and developed which were integrated with the IoT of Wheat, and could be regarded as the core applications to provide actual services about seedling growth, disasters and diseases information for users of nationwide wheat production regions.The main innovation contributions of this paper include:1) In the process of wheat seedling information analysis, we studied the relations between the farmland environmental factors and the four wheat seedling growth factors including main stem leaf age (MSLA), number of tillers per plant (TPP), number of secondary roots per plant (SRPP), number of group stems (GS), proposed a weighted algorithm of wheat seedling index, established a wheat seedling classification model based on the index, and established a spatial distribution simulation based on data fusion from sites to areas in an automatic, quantitative and networked manner.2) In the process of disasters information analysis, we proposed a method of coupling monitoring data of IoT with traditional diagnosis model of wheat agrometeorological disasters, established a revised diagnosis model of wheat drought, freeze, and dry hot wind disasters based on multi-regions, multi-points, and real-time monitoring data of IoT as input, which could be provide both dynamic on-going diagnosis and historical statistical analysis of disasters in a manner of both point scale and regional scale.3) In the process of diseases information analysis, we studied image recognition methods with image processing and support vector machine (SVM), determined an optimized SVM classifier for images of wheat leaf diseases with selected combination of color, texture, and shape feature vectors as inputs, and selected Radial Basis Function as kernel function, and established an intelligent diagnosis model aiming at physical conditions of wheat leaves including health, powdery mildew, stripe rust, and leaf rust in a manner of in-field acquisition, self-learning classification, and on-going detection. Moreover, the modeling methodology could be widely used in real-time intelligent diagnosis applications with whole automated process of in-field image acquisition and self-learning feature classification to on-going detection.Finally, we developed the above four prototype systems:field station -wireless sensor network (FS-WSN), diagnosis system for wheat seedling classification (DS-WSC), diagnosis system for wheat agrometeorological disasters (DS-WAD), and diagnosis system for wheat leaf conditions (DS-WLC), which were integrated into the IoT of Wheat, and could be regarded as the core applications to provide actual services about seedling growth, disasters and diseases information for users of nationwide wheat production regions.The study in this paper laid a research groundwork and provided an application case for agricultural information acquisition in farmland, multi-sources information fusion and dynamic analysis. Consequently, we could conclude that the study reached expected goal and realized good practical application value.