Research And Design Of Air Quality Monitoring System Based On STM32

With the development of science and technology and the continuous improvement of living standards,people are paying more and more attention to the protection of the environment.The requirements for the outdoor environment have also been continuously improved.The current air quality monitoring system has been difficult to meet people’s needs.Establishing a small air quality monitoring system that can be released in various environments is of great significance.The paper takes the air quality monitoring terminal as the research object,combines the embedded technology and the long-distance communication technology,designs an air quality monitoring system based on STM32,uses GPRS(General Packet Radio Service)as the long-distance communication means,collects the Daliang passenger station of Shunde District,Beijiao Guangdong Industrial Design area,Chencun Shunlian Plaza and Lecong RT-Mart.The sensor data of the four monitoring sites in the city,uploaded to the server.Based on the nonlinearity of the data at the four monitoring sites,a prediction model based on BP neural network was proposed to achieve human comfort and AQI(Air quality index).Monitoring and information release have achieved good forecastingresults.The main work of the dissertation is as follows:(1)STM32F103RCT6 is used as a processor and a software program is written using C language to construct a data acquisition terminal with temperature,humidity,PM2.5,and PM10 acquisition functions.It is proposed to transplant Free RTOS real-time operating system to solve the problem of difficult allocation of system resources and difficulty in multi-task scheduling.For the problem that the format of the cross-platform transfer data between the processor and the server does not match,the Cjson is used as the data transmission format,and a long connection with the server is achieved and the collected sensor data is uploaded in a stable manner.(2)Analyze the human body comfort index and the main influencing factors of AQI.For a single data collection point,it is difficult to meet the human comfort index and AQI in the predicted area.Collect sensor monitoring data of multiple monitoring points in the area(four)and propose a BP nerve.As a predictive model,the network trains the model with the data collected from the four monitoring points and imports the model into the display STM32.GPRS is proposed as a means of long-distance communication for the problem that many places cannot publish monitoring information through the Internet.Using python language for program writing,a multi-layer BP(back propagation)neural network model is built.Because the number of layers,the numberof neurons,and the learning rate are not provided with reliable principles,this paper selects relatively good network layers through multiple experiments.,the number of neurons and the learning rate and other hyperparameters.The experimental results show that the accuracy of the training set of the human comfort index prediction model is 84.3%,the accuracy of the test set is 66.5%,the accuracy of the training set of the AQI prediction model is 82.7%,and the accuracy of the test set is 69.4%.Indicates that the model has deficiencies.(3)In order to further improve the accuracy of the prediction and solve the problem that the BP neural network is slow in convergence and easily falls into a minimum value,the Adam-BP neural network algorithm is proposed for optimization.The experimental results show that the accuracy of the training set is 95.7% in the human comfort index prediction model and 81.5% in the test set.The accuracy of the training set in the AQI model is 97.3%,which is accurate for the test set.The rate is 83.6%,achieving the desired effect.