Abstract: Large amount of measured data will be produced during the measurement process of light scattering particle size distribution and there is not a method could be utilized to get accurate analysis of all measured data, thereby, the accuracy and speed of the distribution reconstruction will be reduced, which could cause errors occurring in the measurement result of the coal dust and some hidden dangers that might endanger the safety of production and the staff. The inversion algorithm for the reconstruction of particle size distribution has become one of the important topics in the field of particle size measurement. This paper combined the artificial bee colony algorithm with generalized pattern search algorithm to reconstruct unimodal R-R distribution. Comparing this method with genetic algorithm and Phillips-Twomey method to get a result, it shows that the hybrid algorithm can be effectively applied to the inversion of unimodal R-R distribution reconstruction and can provide effective and reliable method for industrial applications.