Abstract: In order to solve the problems of high head breakage rate and impurity rate in mechanized sugarcane harvesting, a sugarcane harvesting quality prediction model based on least squares support vector machine（LSSVM） was proposed. The LSSVM model was optimized by particle swarm optimization（PSO） and genetic algorithm（GA）, and MAE and MSE were used as evaluation indexes. The results showed that when PSO-LSSVM model was used to predict the breakage rate of sugarcane, MAE value was 0.168 75 and MSE value was 0.027 55. Also, when predicting the impurity rate of sugarcane, MAE value was 0.107 5 and MSE value was 0.024 43, which was better than other models. PSO-LSSVM prediction model was selected in LabVIEW software to provide algorithm support for the system. Combined with MySQL data management software, a set of decision support systems for operation parameters of sugarcane harvester was developed, which analyzed factors affecting sugarcane harvesting quality, prediction of sugarcane harvesting quality, and decision support for operating parameters of key sugarcane components. From the field test, the results revealed that after adopting the system decision-making suggestions, the breakage rate of sugarcane roots decreased from 9.02% to 5.76%, the impurity rate decreased from 8.74% to 4.94%, and the harvest quality of sugarcane improved. As such the results provide a feasible solution for improving the quality of mechanized sugarcane harvesting.