Classification and recognition of strip steel surface defects based on SOCP-RVM

In order to improve the accuracy of surface defect recognition of strip steel, an improved algorithm of multi- kernel relevance vector machine (RVM) is proposed. To solve the problem of large amount calculation and long training time of RVM, and then the RVM algorithm based on second order cone programming (SOCP) algorithm is proposed, which use SOCP optimize the parameters of RVM. The method includes the model parameter estimation, training and optimization of the model based on SOCP, classification test. The typical classification data set is simulated experimentally with SOCP-RVM algorithm, variety of strip steel surface defects from actual production are classified under SOCP-RVM algorithm, the results show that the classification method of SOCP-RVM algorithm has high classification accuracy, remarkable effect, strong practicability, and has wide application prospect.