Support Vector Machine And Signal Matching Algorithm In The Gas Detection Of Sound Spectrum

The use of characteristics of sound propagation in gas components detection has advantages of fast response, low cost and small losses that other traditional methods are quite difficult to complete all, and the use of Support Vector Machine and signal matching algorithm in the detection of gas acoustic relaxation absorption spectrum also has advantages such as high efficiency, excellent flexibility and stable detection accuracy that other detection methods don't have, and it is quite an advanced research that has much potential!This thesis firstly constructs the database of sound spectrum curves based on gas acoustic absorption theory, secondly demonstrates the conclusion that we can use single point such as the peak point or the inflection on the spectrum to fully identify gas acoustic relaxation absorption spectrum based on Support Vector Machine and signal matching algorithm. Then we focus on the theoretical and experimental theory of the research in the qualitative and quantitative detection of gas acoustic relaxation absorption spectrum. In the qualitative analysis of the spectrum, we use the method of support vector machine and the analysis of the peak point. In the quantitative analysis of the spectrum, we propose to the ideal of constructing the maximum fit factor by using characteristic parameters such as the angle chain and area of the spectrum to identity the spectrum with high accuracy. Based on the above research, we designed and implemented the software of the automatic detection of sound spectrum, the software can complete all the function of each module such as the structure of theoretical data, the acquisition and import of experimental data, the analysis of the peak point and the threshold, and the software can also directly obtain the recognition result only by the import of data. Through accurate test of the software, we can guarantee the completeness of the function and performance of the software, we can also enhance the accuracy of the spectrum detection to 1% by using the software.We have done a lot of data processing efficiently by using the software of the automatic detection of the sound spectrum, and we draw the following conclusion through the analysis and summary of the results of my experiment: In the quantitative analysis of gas acoustic relaxation absorption spectrum, we can accurately identify the concentration of the spectrum in the error range of 6% with the accuracy of 1% if the composition of the spectrum is already known. Then we compared the above result with other gas detection method, we can find that our method has advantages of higher efficiency, better flexibility and more stable detection accuracy. This conclusion proves the correctness of the theoretical research and the analysis of the algorithms that we can detect the gas acoustic relaxation absorption spectrum by Support Vector Machine in machine learning and signal matching algorithm, it also proves the availability and accuracy of the software of the automatic detection of the sound spectrum!...