Improved LSSVM to Predict the Elongation of Strip Steel in Annealing Furnace

Abstract

It is difficult to predict the elongation of strip steel in annealing furnace due to the influence of temperature, tension, roll speed and data noise. Thus, a strip elongation prediction method based on least squares support vector machine (LSSVM) optimized by artificial bee colony (ABC) algorithm is proposed. In order to improve the convergence speed and accuracy of the algorithm, the new adaptive step update formula, adaptive probability selection formula and global search factor were introduced to improve the standard artificial bee colony algorithm. The parameter of the LSSVM were optimized by improved artificial bee colony(IABC) algorithm which overcame the subjectivity of human selection and made the LSSVM get better generalization and prediction accuracy. Numerical simulation results of MATLAB show that the relative error and root mean square error predicted by IABC-LSSVM are better than those predicted by ABC-LSSVM and LSSVM which effectively improves the convergence speed and prediction accuracy of the algorithm. Under the real working conditions, IABC-LSSVM provides theoretical support for the prediction of strip extension in annealing furnace and has a certain engineering application value.