Study On Performance Of Leaf-stripping Device Of Sugarcane Combined Harvester

In order to improve synthetic performance of sugarcane combined harvester leaf-stripping device,the study with rotational speed of each roller as experiment factors (rotational speed of inlet roller, rotational speed of first stage leaf-stripping roller, rotational speed of second stage leaf-stripping roller, rotational speed of deliver roller), with the rate of no completely stripped and passing rate of sugarcane stalks as test index in this paper, an indoor experiment was performed, by using leaf-stripping device which self-designed, based on the quadratic general rotary unitized design. The regression mathematical model of the factors and the index was established, while the influence law of each factor and their interaction on performance index was researched and the optimal parameter was combined, corresponding index value was gained. The results showed that the rotational speed of inlet roller was 596r/min, the rotational speed of first stage leaf-stripping roller was 868r/min, the rotational speed of second stage leaf-stripping roller 1082r/min, the rotational speed of deliver roller was 956r/min, and the rate of no completely stripped was 31.73%, the passing rate of sugarcane stalks was 98.19%. The influence mechanism of leaf stripping was analyzed by dynamic simulation during the process of leaf stripping with virtual simulation technology.The last part of the paper focused on prediction of performance index and parameter optimization for sugarcane combined harvester leaf-stripping device basing on BP neural network and Genetic Algorithm. The neural network predictive model of the factors and the index was established, while the parameter optimization was achieved by means of Genetic Algorithm and the best parameter combination, corresponding index value was gained. The results showed that the rotational speed of inlet roller was 623.3r/min, the rotational speed of first stage leaf-stripping roller was 951.6r/min, the rotational speed of second stage leaf-stripping roller 1129.4r/min, the rotational speed of deliver roller was 846.7r/min, and the rate of no completely stripped was 29.45%, the passing rate of sugarcane stalks was 98.48%. Compared with the traditional predictive results of the regression mathematical model, the predictive results of BP neural network model were more close to measured value than the regression mathematical model. The results of parameter optimization basing on Genetic Algorithm made the performance index better. Finally it indicated that the performance index prediction and parameter optimization by BP neural network and Genetic Algorithm were feasible and effective.