Camber Prediction Based on Fusion Method with Mechanism Model and Machine Learning in Plate Rolling

In order to realize the high-accuracy prediction of steel plate camber and accurate control roll gap tilt for straightness, a fusion method with mechanism model machine learning strip at exit side is investigated. Based on basic equation transverse asymmetric rolling, mathematical curvature radius entry derived. Meanwhile, vision measuring adopted in which subpixel coordinates rolled piece edges can be obtained, size plan view also settled indirectly carry out feedback defect. Tilt value controlled advance avoid occurrence predicted high accuracy. Prediction synthesis leveling based PSO-LSSVM algorithm used, relative error within ±5% both training set testing set. Combine adjustment prediction, different processes product specification calculated by predicting accurately obtain good straightness final product, range ±6% after being compensated algorithm. The research result reveals that this suitable optimization rolling process.