Framework Design of Environment Monitoring System Based on Machine Learning
Lingxiao Meng,Shudong Li,Xiaobo Wu,Weihong Han
DOI: https://doi.org/10.1007/978-3-030-57881-7_34
2020-01-01
Abstract:With the gradual deepening of industrialization process, a series of problems brought about by industrial pollution. Urban development has exerted tremendous pressure on the ecological environment and has resulted in a serious impact on the lives and residents. Therefore, real-time monitoring of air quality and accurate forecasting are very important for the full utilization of the environmental data index. With an original platform, this paper deeply considers the actual needs and the performance defects of users, and build the platform of system: (1) The data map function has been added. Real-time data is identified on the map, and different colors are displayed in different intervals for PM2. 5; (2) Alarm function is added. When data exceeds a certain threshold value, logos will be displayed on the map; (3) Trend analysis functions are added, and BP artificial neural network is used to predict the data of environmental data in the future. The data scrolls playback on the map over time; (4) Reconstructing the change curve of data makes the system more intuitive, simple and easy to use. Based on the theory of BP artificial neural network, the influence of temperature, humidity, PM2.5, atmospheric pressure and illumination intensity on one of five factors above are carefully analyzed respectively. And then a reliable and suitable environmental index for predicting the five factors above is obtained. Finally, the forecast results are shown on a heat map for better visualization.