Resesarch On Prediction Method Of Material Removal Rate In Belt Grinding Based On Spark Image

Abrasive belt grinding is cold grinding.With the advantages of high grinding effieiency and flexibility,it is widely used in the finishing of free-from surface ports.The suteamatic belt grinding techniques usually use the forces tracking method to control the material removal rate.Thte grinding force is re-planned acednding to the material remonal rate predication model.Doe to the influence of belt wear,vibration and theorctical model assurptions,the material removal rated model is not accurate enough to reflect the actual situation.In this poper,a method for prefileting the material remonal rate of abrasive belt grinding based on spark image features is proposed.The main research contencs include:(1)The overall strocture and control scheme of the abrasive belt grinding experiments are designed and developed.The collection scherae of relerans experimental data,such as spark images and material senwoval rates are introdneed.The experimental data acquisition scheme,the overall experimental scheme,and the grinding experimental poramenters are planned.The experiments provide data for the modeling of the prodication material renoval rat.(2)Aiming at the problem of the charge of the posithan of the COMS camera thring the motion of the grinding mechanism and the charge of the posithan of the COMS cantera thuring the grinding process,a calibration method of visual distances of the spark images based on thegeometrie parameters of the objects is proposted.Firstly,the Sobel opertor and OSTU adaptive threshold method are used to extract the edge features of the Image.Then the edge thinning algorithm is used to single-ploel the edges.The geonetric parameters of the selexted objects are identified based on the Hough paramter transformation method.The zoom ratio of the image is calculated.Fimily,the Image is proportionally scaled using the bi-linear imerpolation method.(3)An automatic segmentation method of the spark image based on the color and geconetric features of the spark image is proposied.Firsity,based on the color characteristics of the spark image,combined with the OSTU maximum inter-class variance method and the maximum entropy method to calculate the adaptive threshold based on the R component,the foreground and are segmented from the spark image.Then,the center and radius of the contact wheel and the tension wheel identified by the Hough transform are used to realize the segmentation of the belt grinding device.Finally,based on the ratio of the R component and the G component of the image combined with the connected domain labeling algorithm,the spark field is completely segmented.(4)The definition and quantified method of the spark image features related to the material removal rate are given.By analyzing the influence of the different grinding parameters on the spark image,such as grain size,workpiece material,belt velocity,workpiece feed rate,theoretical grinding depth,etc.,the material removal rate-related features of spark image are preliminarily determined as area,intensity,brightness,and energy spectrum.These features are defined and their quantified methods are given.Features of the spark image that have a strong correlation with the material removal rate are determined as area,intensity,and energy spectrum by analyzing the dispersion and the correlation between each feature and the material removal rate.(5)Material removal rate prediction models based on spark image features are established.The least-squares method is used to establish a single feature-based regression model and multiple features-based linear regression model.Multi-feature-based material removal rate prediction model is established,using support vector regression and BP neural network.The five indexes of mean absolute error,maximum absolute error,mean square error,determination coefficient,and mean relative error are used to evaluate the established prediction models.The results show that the prediction error of the method proposed in this paper is small enough for application requirements.(6)A grinding depth and material removal contour model based on the material removal rate of abrasive belt grinding are established.Through the Hertz contact theory,the pressure distributions of the contact area between the abrasive belt and convex surface,concave surface are analyzed.The calculation formulas of the material removal profile and the maximum grinding depth calculation formula are derived.The feasibility and correctness of the model are verified using the plane experimental data.