Design And Realization Of Automatic Weather Monitoring And Forecasting System

With the development of science and technology,the field of meteorological observation is moving from traditional manual monitoring to automatic processing.Due to the large size of traditional weather stations and the fact that most of them use short-distance transmission,the application scenarios and functions of automatic weather stations are limited.Moreover,with the development of technology,the accumulated meteorological data volume also increases gradually.At this time,the traditional algorithm is still used for meteorological prediction,which will cause serious waste of data resources.The field of meteorological monitoring urgently needs some kinds of automatic equipment to improve the efficiency of meteorological monitoring and forecasting.After conducting the system requirements analysis,this paper designs and implements a portable automatic weather monitoring and forecasting system.The system can collect temperature,pressure,humidity,wind direction and wind speed,and forecast the temperature for the next 24 hours on the hour by using the temperature prediction model based on depth learning.The system is mainly composed of two parts:the Meteorological Data Monitoring Terminal and the air temperature prediction model,in which the main control chip of the monitoring terminal processes the data collected by the sensors and transmits it to the database of the cloud server through 4G,in the network environment,users can monitor the current weather information and the results of temperature forecast in real time by PC,at the same time,they can query and export the historical weather data.The temperature prediction model uses deep learning,which is developing rapidly to carry on the time series modeling of air temperature,through the massive historical data training and applies the actual measurement data to carry on the experiment contrast analysis,then the temperature prediction model applicable to the system is obtained,and it is deployed in the cloud server for system call.In this paper,the LSTM and GRU based on RNN are proposed to construct the temperature prediction model.While solving the long-term dependence problem of traditional RNN,the parameter setting of the model was improved by combining the problems in the experiment,by adding regularization,Adam Algorithm and other optimization strategies,the model can achieve a better convergence effect.After the overall test and evaluation of the system,the results show that the system's meteorological monitoring function meets the Automatic weather station's requirements for reliability and low power consumption;The study of temperature prediction model based on deep learning also expands the regional temperature prediction method and shows good prediction ability.The system has a good application and development prospects.