Design And Implementation Of PM2.5 Detection System Based On Raspberry Pi

PM2.5 is extremely harmful to human health.Nowadays,the monitoring of PM2.5 concentration mainly relies on officially established air quality monitoring sites,which are costly and unevenly distributed in cities.Consequently,it is hard to monitor PM2.5 concentration in a fine-grained way.Meanwhile,the application of machine learning technology has become a common thing in our daily life.By designing various machine learning algorithms,we are able to learn and analyze the accumulated historical data,which can help us a lot to estimate the current or future events.People have achieved great success,especially in fields of speech recognition,image recognition and recommendation system.Therefore,the study of using machine learning technology to monitor PM2.5 concentration is worthy of in-depth investigation and has broad prospects.This thesis firstly introduces the research background,research content,main work and the key technologies used in the system.After that,a demand analysis of the system is carried out.According to the demand analysis,the thesis gives a overall structure of the system and the division of function modules,then describes the detailed design and implementation of each module.The work of this thesis is based on a participatory perception platform,which consists of Raspberry Pi,Server,and Client.The Raspberry Pi is responsible for taking images and uploading them to the Server.The Server crawl the corresponding meteorological data according to the EXIF information of the image.In order to extract image features which are highly correlated to the PM2.5 concentration,brightness normalization,image segmentation and other techniques are used for image preprocessing.On this basis,the Server combined with meteorological features and image features as model input data,and call the LSTM algorithm model which is formed by offline training and suitable for processing time series data to estimation the PM2.5 concentration.The Client is responsible for initializing the Raspberry Pi,managing the site,and obtaining real-time PM2.5 concentration data.Finally,the thesis conducts functional module tests so as to verify the usability and stability of the system,and the summary of the full text is also performed.The PM2.5 monitoring system based on the Raspberry Pi can achieve fine-grained monitoring of PM2.5 concentration on both time and space dimensions,so that provide perception platform users with real-time PM2.5 concentration monitoring,PM2.5 concentration trend changes and other services.Therefore,users can get the air quality status of different scenes,fully understand their environment and plan travel reasonably.What’s more,the collected image data and meteorological data can provide strong data support for the government’s urbanization management.