Using Least Squares Support Vector Machine and Polynomial Partial Least Squares Method Quantitative Analysis of Gases in Mines

摘要

At present, the range of index gases of coal quantitative analysis is small, when the gas concentration is high, the analysis result error is large. A method based on least squares support vector machine (LS-SVM) and polynomial partial least squares(PPLS) is proposed to establish a quantitative analysis model of mixed gas of coal. The least squares support vector machine was used to classify the concentration interval, the concentration was divided into several sub-intervals, and establishes the polynomial partial least squares model for each sub-interval. Finally, the proposed method is compared with the partial least squares method(PLS) and LSSVM-PLS, the results shows that the method has the smallest RMSE of the root mean square error and the the largest R2 of predictive determinant coefficient, especially when the gas concentration is high, the analysis results are more accurate. Which significantly improves the prediction accuracy of the gas. The results show that the proposed method can accurately carry out quantitative analysis of the index gas under the mine.