Real time detection of coal mine dust is importance to safety in production.The dust pattern recognition algorithm is presented based on diffraction angular distribution.To improve the dust detecting precision

the error analysis is made

and dust amendment function has be brought up.The eigenvectors of thirty six amendment patterns are worked out and stored in singlechip computer.During measurment

choosing suitable amendment pattern ameliorate the optimum pattern in local scope.Pattern eigenvectors mended is better close to the real time dust signals.The simulation experiment shows the detection precision and speed have be improved.