Straightening method based on support vector machine for workpiece deformation

Abstract

The invention relates to a straightening method based on a support vector machine for workpiece deformation. The method includes steps of (1) collecting test data of workpiece deformation straightening; (2) building up a training sample set; (3) building up a straightening parameter predicting model based on computing method of the support vector machine; and (4) performing workpiece straightening process: inputting actual workpiece deformation data into the predicting model, automatically calculating straightening position and straightening force parameters, and inputting the calculated data into a straightening device to perform automatic workpiece straightening. By the aid of a small amount of workpiece straightening test data, the straightening method is capable of realizing intelligent predicting of the straightening parameter, and accordingly, a large amount of test cost and time is saved, and efficiency and precision of the workpiece straightening are enhanced.

Description

Technical field [0001] The invention relates to the technical field of workpiece correction, in particular to a method for correcting workpiece deformation based on a support vector machine. Background technique [0002] During the processing of metal workpieces such as casting, forging, heat treatment, etc., due to uneven stress, uneven heating and cooling, transportation and stacking, etc., residual stresses are usually generated inside, resulting in varying degrees of residual deformation. These deformations seriously affect the performance of the workpiece and must be corrected. [0003] At present, most of the domestic workpiece correction equipment is a manually operated machine with a simple structure. It mainly depends on the experience of workers to determine the correction process parameters. It can realize automatic operation according to different correction requirements. It has disadvantages such as low work efficiency and poor stability, and it is difficult to meet la...

AI Technical Summary

Problems solved by technology

[0006] In practice, using neural network and other methods to determine the correction process parameters and determine the correction effect, it is necessary to provide a large and reliable training data to build a calculation model, which is difficult to achieve in actual operation

Moreover, problems such as over-learning and under-learning are prone to occur during the neural network calculation process, which affects the accuracy of deformation correction.

Images