针对多组分混合气体的检测问题,提出了基于红外吸收技术进行探测并利用BP神经网络进行信号分析处理的检测方法。该方法采用宽带中红外光源和前端带有窄带滤光片的探测器进行气体探测,探测器输出的每路微弱电信号对应1种气体吸收波长的光波,在对此电信号放大滤波后经A/D采样送达计算机。对实验采集到的数据利用BP神经网络进行分析,能够很好的消除各组分气体之间的干扰,测量相对误差在5%以内,测量拟合曲线呈现良好的线性关系,完全能够满足多组分气体鉴别和测量的要求。 Aiming at the problem of multi-component mixed gas detection, a detection method based on infrared absorption technology and signal analysis and processing using BP neural network is proposed. The method uses a broadband mid-infrared light source and a detector with a narrow-band filter on the front end for gas detection. Each weak electrical signal output by the detector corresponds to one kind of light wave with the gas absorption wavelength. After amplifying and filtering the electrical signal, A / D Sampling to the computer. The experimental data collected by BP neural network analysis can be very good to eliminate the interference between the gas components, the measurement relative error of less than 5%, the measured curve fitted showed a good linear relationship, fully able to meet the multi-group Sub-gas identification and measurement requirements.