Prediction On Gluing Effect Of Blender Based On Support Vector Machine

Shaving gluing is an important process in the industrial production of particleboard.The blender is the key equipment for shaving gluing.The structural parameters of the blender directly affect the motion trajectory curve and the covering effect of the adhesive on the surface of shavings,thereby affecting the performance and production cost of the particleboard.The mixing of shavings and adhesives in the blender is a complex nonlinear process.Establishing an effective model to predict the gluing effect of the blender is conducive to the precise control of the gluing process,enhancing the adaptability of the gluing system,optimizing the structure and process parameters of the blender,improving the performance of particleboard,and controlling the cost of the production.Taking the gluing process of the single-axis ring blender as the research object,the study determined the main structural parameters of the blender which affects the gluing effect through the gluing process of the blender.Using kinematics and numerical simulation,the research studied the influence of the main structural parameters on the trajectory curve of shavings in the blender to extract the parameters that could characterize the gluing effect of the blender used to describe the process and results of the gluing.The research results show that the distance between claw and inner wall,the installation angle of claws,the axial distance between the claw are main structural parameters of the blender.The gluing effect characterization system of the blender,which consists of the initial discharge time,shavings retention,adhesive coverage and adhesive coverage variation coefficient,can characterize the gluing effect of the blender from the aspects of the mixing efficiency,the adhesive covering effect and the breaking rate of shavings.The study designed and manufactured the ring-type blender used in the test,in order to build a particle gluing test platform.This thesis studied the influence of structural parameters on the gluing effect by the orthogonal test,and analyzed the degree of correlation between the structural parameters of the blender and the gluing effect.The data of the prediction model for the gluing effect of the blender was obtained by the full factor test.The test and analysis results show that the main structural parameters of the blender have no significant effect on the adhesive coverage variation coefficient.It is suitable to determine the initial discharge time,shavings retention and adhesive coverage as parameters for characterizing the gluing effect of the blender.According to the support vector machine regression principle,the study constructed the basic model for predicting the gluing effect of the blender.Using a grid search method to optimize model parameter,the penalty factor and the kernel function parameter,the study established a support vector machine regression prediction optimization model.The prediction optimization model is used to predict the gluing effect of the blender The research results show that the established shaving gluing effect prediction model based on support vector machine can accurately reflect the relationship between the main structural parameters of the blender and the gluing effect.Model regression performance evaluation indicators,mean square error and determination coefficient,meet model accuracy requirements,and can meet production needs.