PM2.5 Observation And Prediction Syetem

At this stage,air pollution problems,especially smog problems,are still a very serious problem.PM2.5 have taken up an important position in haze problems,and a real-time measurement of PM2.5 around itself can be performed.It is very meaningful to understand its follow-up trend.At this stage,the government is paying more and more attention to environmental issues,and PM2.5 measurement sites have been set up in most cities,but only by means of fixed-po:int measurement to estimate PM2.5 values,it is not possible to accurately give specific microscopic The PM2.5 value of the position,so observing PM2.5 and predicting subsequent values is still meaningful.In this paper,a portable PM2.5 monitoring equipment based on embedded is designed,and a PM2.5 prediction algorithm based on BP neural network is proposed.At the same time,the BP neural network is easy to fall into the local minimum,and it is optimized by genetic algorithm.To more accurately predict PM2.5.The main work is as follows:1.Design circuit based on the modular principle,the modules of the PM2.5 monitor are divided and the corresponding circuit design is carried out.The driver software and the overall driver software of each module are designed,and the final circuit schematic diagram and PCB board diagram are given.2.The historical data of AQI related gas and temperature in Changsha for the past three years were collected.The relative error between the expected value and the actual predicted value was used as the loss function of the neural network.The PM2.5 prediction model based on BP neural network was designed and trained by using the collected data.BP neural network,and use the test data to actually test the predictive performance of the model.After testing,the model has certain predictive functions and has certain stability.3.Genetic neural network is proposed for the purpose of overcoming the and existence of local minimum in conventional BP neural network.Through the data test,the average prediction accuracy is 0.125,which is 13%higher than the average accuracy of the BP neural network 0.253.The experimental results show that the genetic neural network model proposed in this paper improves the accuracy of the prediction results to some extent and verifies the validity of the model.