Environmental Monitoring Prediction Based on Graph Convolution Neural Network

: Environmental monitoring plays an important role in identifying environmental characteristics. Abnormality is related to negative effects and seriously affects human life. Many sensors can be placed in specific areas and are responsible for monitoring the environmental characteristics of specific phenomena. Sensors report their measurement results to a central system, which can carry out invitational reasoning. Therefore, the system responds to any event related to the observed phenomenon through decision-making. This paper proposes an event recognition mechanism based on sensor measurement, and gives the corresponding decision, so as to realize the real-time event recognition. The system adopts the statistical learning method of data fusion and prediction (time series regression) to effectively collect the measurement data of sensors. Fuzzy logic is used to deal with the uncertainty of derivative alarm decision. This paper introduces the application of GIS, geostationary, meta database and cart in environmental monitoring by querying survey data in meta-data database connected with GIS. The results show that the I | o value of mechanical ventilation is 36% with indoor heat source and 18% without indoor heat source. prediction period is 10 days [1]. Some experts have studied the research and development of base station edge intelligent monitoring and antenna monitoring terminal height measurement technology. By constructing a small edge server based on B / S architecture, combining TCP / IP and web socket protocol, the data communication middleware is completed, and the terminal monitoring data is presented on the web page in real time, which realizes the integration of open measurement and remote monitoring. In order to improve the spatial resolution of climate change information, the change of plant phenology and temperature was combined with GIS overlay technology, and applied to the vector incidence prediction at regional scale. This paper proposes a system model based on three-tier architecture of Internet of things, which is realized by geometric merging of surface data of Internet of things initiator. Considering the temporal and spatial aspects of environmental processes, the institutional basis of this system based approach is discussed, and how to encode environmental processes in the system based on graph convolution neural network technology is explained. This paper introduces some preliminary research results of station scientists in the prediction modeling and early warning of ecological environment. The co-environmental process is affected by many factors, such as hydro-meteorological conditions, biological factors and human activities. Therefore, it is difficult to incorporate all these effects and their interactions into deterministic or analytical models. A prediction model of marine ecological parameters is proposed, which combines time series prediction method with neural network nonlinear modeling. Some experts have studied the design and implementation of the indoor environment monitoring system based on Android, designed and developed a set of indoor environment monitoring system based on Android, which can realize the real-time monitoring of temperature, humidity, PM2.5 and flue gas status, as well as the regulation of the environment. It relies on a framework called adaptive model selection (AMS), which is designed to run time series prediction technology on resource constrained wireless sensors. We use two demos to show our implementation, related to environmental monitoring and video games. TinyOS is the reference operating system of Low-Power Embedded system. Trotsky and convention are wireless sensors. This paper extends the literature of normal and cliometrician simultaneous prediction interval to gamma distribution. Gamma distribution can adapt to all kinds of non normal distribution (with inclined right tail) and the existence of mutability. Gamma prediction limit is the best candidate for