The compression process LS-SVM model of biomass compound coal mixed with sawdust, coal and rice hull was established according to the published experimental data. The test result showed that the maximum relative error of predicted and experimental values of two models was 3.46% and 5.83%, which proved the efficiency of model. On this base, the multi-objective optimization objective function of compression process was designed and fitted, by which the Pareto optimal solutions were gained in a single run. According to the standard target of solid biomass fuels and CV value which represented the comprehensive performance of biomass compound coal, the efficient solutions that are suitable for compression process were selected from the Pareto optimal solutions under different indicators for compound coal mixed with sawdust, coal and rice hull.