Abstract: The on-line detection technology of dust concentration is an important basis for mine ventilation safety and pneumoconiosis warning,and the core of the technology is the detection error.In order to reduce the error of on-line dust concentration detection,a new dust concentration detection technology based on multi-sensor fusion was proposed.It was found that the light scattering method and charge induction method are the most widely used methods for the on-line detection of dust concentration.Based on the basic principles of the two methods,the light scattering sensor subunit and the charge sensing sensor subunit were developed.In order to solve the pollution problem of light scattering sensor subunit,a light scattering sensor subunit with air curtain dust separation was designed by means of air curtain dust control technology.Then the structure of the two sensor subunits was fused to obtain a series multi-sensor fusion unit.Then,based on the experimental preparation and system,a series of experiments were carried out on the multi-sensor fusion unit.The multi-sensor data fusion method was used to fuse the AD values of the light scattering sub-unit and the charge induction sub-unit,and a dust concentration detection algorithm was proposed to form the multi-sensor fusion dust concentration detection technology.Finally,the experiment shows that the air curtain dust separation structure of the light scattering sensor sub-unit can avoid the dust pollution to the optical devices.The detection error of the multi-sensor fusion unit is ≤8.5%,which is 6.4% smaller than the light scattering sensor subunit and 6.2% smaller than the charge sensing sensor subunit.The average calibration sensitivity of the multi-sensor fusion unit is 2 911.4,which is 32.3 times higher than that of the light scattering sensor sub-unit,and 124.4 times higher than that of the charge sensing sub-unit.It is proved that the multi-sensor fusion technology overcomes the limitations of the light scattering method and the charge induction method in the detection of dust concentration.Combining the advantages of both methods,it further reduces the detection error and improves the calibration sensitivity.