中文摘要：

随着矿井环境信息感知、危险源辨识等技术的发展,对气体传感器检测精度和可靠性的要求显著提高。为改善矿用气体传感器的性能,针对气体传感器补偿方法存在的技术难题,提出一种微粒群优化支持向量机（PSO-SVM）的非线性补偿方法。以CO传感器为例,采用Matlab软件进行数值仿真,BP神经网络方法将误差从18.48%降到8.51%,而采用微粒群优化支持向量机方法将误差降到5.28%。实验结果表明：PSO-SVM补偿方法能有效消除非目标参量对传感器输出结果的影响从而完成非线性补偿,提高了矿用CO传感器的可靠性与检测精度。

英文摘要：

With the development of mine environment information perception,hazard identification and other technologies,the requirement for the accuracy and reliability of the gas sensor is significantly increased.In order to improve the performance of the coal mine gas sensors,according to the technical problems of compensation method of gas sensors,the nonlinear calibration method for particle swarm optimization support vector machine was proposed.Taking the carbon monoxide sensor as an example,using Matlab software for numerical simulation,the BP neural network method reduced the error from 18.48% to 8.51%,and the particle swarm optimization support vector machine method reduced the error from 18. 48% to 5. 28%. The experimental results show that the PSO-SVM compensation method can effectively eliminate the effect of non-target parameters on the output of the sensor to complete the nonlinear compensation,and improve the reliability and detection accuracy of mine carbon monoxide sensor.