Brief introduction of self-adaptive characteristics of packaging steel belt model

Brief introduction of self-adaptive characteristics of packaging steel strip model
　　 The mathematical model adopts various assumptions and the average nature of the environment during modeling limits the accuracy of the model. The use of measured data in the current environment to adapt and self-learn the model is to improve the prediction of the model in a specific environment. An important method of setting accuracy.
The 　　 flatness (convex) model involves a number of factors that affect the shape of the loaded roll with complex mechanisms, such as the hot roll shape and its change under segmented cooling, the worn roll shape, and the rolling force related and related factors. The lateral stiffness related to the bending force, etc., therefore complicates the calculation of the shape of the loaded roll gap. If the cross-sectional shape of the strip steel at the exit of each rack can be measured, even the measurement of the two-point thickness (the center and the marking point on the side of the strip) will make the self-adaptation and self-learning of the convexity model easy.
　　 At present, most of the current cold-linked Zamak rack convexity meters have just begun to enter practical applications. Therefore, the flatness control is far behind the thickness control in the convexity detection instrument, which brings great difficulties to the self-adaptation and self-learning of the flatness model.
　　 The commonly used 'self-learning' at present is: when the final frame shape feedback control is used to improve the shape of the finished product, the roll bending force and roll movement position of the feedback control are recorded for self-learning. , In order to improve the lower coil steel.
　　Because the flatness defect of the finished strip is caused by the unequal accumulation of the relative convexity of the export of each frame, the ideal method should first ensure that the relative convexity of the outlet of each frame is equal to the relative convexity of the hot milk incoming material. The convexity is then fine-tuned by the feedback control of the shape of the last frame. If the deviation of the previous stands is corrected by the last stand alone, that is, only the roll bending force and the CVC (or HC rolling mill) roll shift position after the feedback control of the last stand are recorded. Identity) is wrong (only learn the feedback control result of the last rack).
　　 Of course, if the roll bending force and roll shifting of each stand of cold tandem rolling are set to ensure that the relative crown of each stand is close to constant or to ensure that the flatness of the strip between the stands is better, It is beneficial to continuously improve the shape of the plate by recording the roll bending force and roll shifting value after the feedback control of the last frame is used for self-learning.
　　Model self-learning also uses exponential smoothing, which is similar to the self-learning method of thickness model.