Calibration And Concentration Prediction Of Gas Sensor For Gases Alarm Wearable System

With the development of society,the domestic construction of various industries continue to improve.The scale of various types of construction sites,such as industrial production,municipal maintenance,mining and other gradually expanded.Along with the development of construction,gas pollution is more and more serious.The CO because of its wide range,colorless,tasteless without stimulation and it has high toxic,it's very difficult to be perceived by the human body,and has a serious effect on people especially for people who works in the places which can produces CO gas.Thanks to the rapid development of smart clothing,sensors,machine learning,electronics and computer industry,the toxic gas detection system for special operations has also developed greatly.Which wear-type work site poison gas warning system CO gas sensor embedded in the field of work protection equipment,without human intervention and independent work to improve the safety and health of workers operating time resolution and spatial resolution.In this paper,the sensitivity of the gas sensor is calibrated and the CO gas concentration is predicted based on the wearable alarm system of harmful gases in work field.The hardware platform of this paper is based on Raspberry PI Zero,CO gas sensor using Solidsens CO1000 Micro3.Raspberry PI Zero runs Linux system,it supports Python language.Due to Python simplicity,readability and scalability,and has rich and strong library such as Numpy,Matplotlib.Python also integrates GUI and many other tools,so it's more suitable for practical projects compared to Matlab.Because the sensitivity of the electrochemical sensor will be affected by the temperature,humidity and air pressure in the environment,the sensitivity of the electrochemical gas sensor will be predicted by the improved least squares method.The concentration of CO gas collected by the sensor is a series of time series that changes over time.There are many ways to predict time series,such as moving average method,exponential autoregressive model and other classical methods.In recent years,with the development of advanced algorithms such as machine learning and artificial neural network,machine learning has been gradually applied to the prediction of time series.In this paper,we use the decision tree regression,support vector regression,moving average model,exponentially weighted moving average model to predict the CO gas concentration,and compare the effect of the four methods,taking into account the operating speed of the processor and the accuracy of the algorithm,the support vector regression performance is more superior.