Design Of Indoor Air Quality Evaluation And Detection System

With the significant changes of living style and working conditions in modern society,urban population is gathering gradually,and the use of materials in the process of interior decoration and construction is increasing day by day.Due to the lack of strict requirements for indoor air quality in some construction processes,indoor air pollution is becoming increasingly serious.Therefore,the assessment of indoor air quality has become a research hotspot of scholars at home and abroad.Based on the harm degree of indoor air quality to human body,this paper selects five environmental pollutants:CO₂,CO,TVOC,formaldehyde and particulate matter as the evaluation factors of indoor air quality,combined with the latest national indoor air quality standards,establishes the evaluation standards of classroom air quality in school environment.Firstly,an embedded system is designed and manufactured by using STM32F103ZET6microprocessor and industrial grade gas sensor to detect the concentration of various indoor air pollutants.The data is processed by microprocessor and displayed on the LCD screen.At the same time,wireless WIFI module or bluetooth technology is used to transmit the pollutant concentration data to the APP client of mobile phone,and bee is used the buzzer will give local real-time alarm.In order to evaluate indoor air quality scientifically and accurately,this paper uses BP neural network to establish indoor air quality evaluation model,and uses additional momentum method,variable learning rate method,elastic BP algorithm and LM algorithm respectively to optimize BP neural network.Through the simulation of algorithms one by one,it is found that these algorithms have different degrees of defects in network training speed and evaluation accuracy.For the above problems,this paper uses the momentum algorithm with variable learning rate to optimize the BP neural network,and uses the algorithm to select the appropriate network topology and parameters to establish the indoor air quality evaluation model.Through MATLAB software to simulate 30 groups of test data for multiple simulation tests,it is concluded that when the number of nodes in the hidden layer of the network is 6,the simulation results are the best,and the minimum evaluation accuracy is 93.33%at this time,and the convergence rate is 57.1%higher than that of the traditional BP neural network,which greatly improves the training speed and evaluation accuracy of the network model.In order to verify the practicability of the momentum algorithm with variable learning rate,a set of indoor air pollutant concentration data is collected from 8:00 to 17:00 every other hour in ten classrooms of a university in Dalian from 1.2m above the ground by using the detection system designed in this paper.A three-layer BP neural network of 5-6-1 is established to evaluate the air quality of the collected data.From the evaluation results,it can be seen that the evaluation accuracy of the model is 96%,which is 60%faster than the convergence speed of the traditional BP neural network.It shows that the model can evaluate the indoor air quality reasonably and effectively,and has a good application prospect.