Quantitative Recognition Of Toxic Gas Mixture Based Multi Sensor

With the rapid development of society,many flammable,explosive,and toxic gases are widely used as chemical raw materials or as energy gases.However,there are often safety hazards during the storage of these gases.When the gas leaks into the air to a certain concentration,it can cause an explosion when it meets an open flame,or people live and work in low concentrations of toxic gases for a long time,it can cause irreversible damage to human organs,so it is very important to have effective and accurate detection of toxic gases.In order to realize that the gas sensor can detect the low-concentration mixed gas existing in the environment,and give the concentration and type of the mixed gas.In this paper,pure ZnO and CuO gas sensitive materials are prepared by hydrothermal method,and high-performance ZnO and CuO gas sensitive materials doped with Y₂O₃are prepared respectively.The prepared four gas-sensitive materials are made into a sensor array,combined with a pattern recognition algorithm,to realize the quantitative prediction of the concentration of the mixed gas(formaldehyde,ammonia,ethanol and acetone)and the qualitative identification of the type.The main findings of this article are as follows:(1)The prepared pure ZnO has a nanosheet structure,presenting an umbrella shape,and the doped Y₂O₃is attached to the ZnO nanosheet as micro-particles.The prepared pure CuO is nanorods,and the doped Y₂O₃wraps part of the CuO nanorods to form a core-shell structure.Because of the presence of Y₂O₃,the surface area of ZnO and CuO gas-sensitive materials is enlarged,and the concentration of carriers is increased at the same time,and the performance of the gas-sensitive materials is improved.Four gas-sensing materials are made into gas-sensing sensors,and the response performance of formaldehyde,ammonia,ethanol and acetone are tested at low concentrations.The response of doped ZnO and CuO to the four gases is generally improved.Among them,the sensitivity of yttrium-doped ZnO to 10ppm formaldehyde gas is 69.2%,and the sensitivity of yttrium-doped CuO is 78.5%,so the prepared The gas sensor can detect low-concentration gas,and the sensor array has good stability.(2)Use the prepared gas sensor to collect the response data of the mixed gas.In order to reduce the response error of the gas sensor itself,the response curve is filtered and normalized.The maximum value in the response curve is extracted as the characteristic value.There is a significant difference in the maximum value between the response curves of different sensors at the same concentration.The difference algorithm is used to process the response curve,and the rising and falling edges of the curve are respectively extracted as eigenvalues,which realizes the expansion of the dimensionality of the eigenvalues and improves the utilization rate of the response curve.(3)Combine the two extracted eigenvalues with the BP neural network to complete the quantitative prediction of the mixed gas concentration.The genetic algorithm is used to search the weights and thresholds of the BP neural network globally to realize the optimization of the BP neural network.The prediction results show that the difference algorithm combined with the optimized BP neural network has better prediction performance.The prediction errors(formaldehyde,ammonia,ethanol and acetone)are4.37%,5.74%,6.74%,and 8.56%.Combine the extracted maximum eigenvalues with the support vector machine to complete the qualitative recognition of the mixed gas type,and the recognition accuracy is 95.25%.In order to obtain the optimal values of c and g,the genetic algorithm is used to optimize the support vector machine.The recognition results show that the optimized support vector machine has better recognition performance with a recognition accuracy of 97.41%.