Infrared sensors of gas methane is one of the main sensors which monitor and test the coal gas explosion.Nonlinear dynamic affects its accuracy and measuring range

which counts against to the production safety.In order to solve this problem

least square support vector machine（LS-SVM） nonlinear dynamic method is used to compensate the sensor

and improve the algorithm.Comparisons are given through simulation experiment

the experimental results show that based on improved least squares support vector machine（ILS-SVM） nonlinear dynamic compensation

measuring accuracy and measuring range of the sensor are greatly improved

which has positive meaning to promote safe fabrication.