Research On Model Of Pressure Straightening Stroke And Error Compensation Algorithm

Linear parts such as half shaft,drive shaft,crankshaft,Laura,linear guide,etc are widely used in mechanical industry.Heat treatment is an important process during the manufacturing process,but bending deformation often occurred during the process of heat treatment due to the uneven inter-stress.In addition,the external forces in transportation and use also bend the parts.The straightening becomes a necessary process in order to satisfy the requirement of the part design tolerance.Pressure straightening method is widely used in mechanical industry,the accuracy of the pressure straightening method mainly depends on the accuracy of the model of straightening stroke.This thesis mainly studies the pressure straightening stroke modeling method and the compensation algorithm of straightening stroke error.Firstly,the straightening stroke model is established by a variety of methods;Secondly,the error source of straightening stroke is analyzed,and BP neural network is used to establish the compensation model of straightening stroke error;Thirdly the straightening stroke prediction system based on the model of straightening stroke and error compensation algorithm are established and tested.The main content of this thesis is as follows:(1)Two straightening stroke models are established based on the classic elastic-plastic theory and finite element method,a straightening stroke modeling method based on data mining is proposed.The straightening data is acquired by doing straightening experiment,and the polynomial straightening stroke model and Gaussian kernel straightening stroke model are established;Secondly,the K-fold cross-validation algorithm is used to calculate the generalization error of the straightening stroke model,and the optimal straightening stroke model based on polynomial and optimal straightening stroke model based on Gaussian kernel are established;Finally,an experiment is done to compare and analyze the four straightening stroke models.(2)A compensation model of pressure straightening stroke error is established.Firstly,the main source of straightening stroke error is analyzed,and BP neural network is used to establish the compensation model of pressure straightening stroke error;Secondly,the compensation model is simulated by the straightening model based on elastic-plastic theory;Finally,an error compensation experiment is conduct to verify the accuracy of the compensation model of straightening stroke error.(3)A straightening stroke prediction system is established and the straightening stroke prediction software is developed.Firstly,the overall scheme of straightening stroke prediction system based on the straightening stroke model and the compensation algorithm of straightening stroke error are designed in detail,and then straightening processing database is established and straightening stroke prediction software is developed.This system has self-learning ability,and it can acquire the straightening data automatically and store it in the database,and then the data is used to predict straightening stroke and compensate straightening stroke error.(4)The straightening stroke prediction system is tested on self-developed straightening machine.Firstly,an experiment is conducted to compare the accuracy of straightening process with and without compensation algorithm of straightening stroke error,and then a three-coordinate measuring machine is used to measure the straightness of the workpiece before and after straightening.The accuracy of straightening is further verified.