Model learning method applicable to cross rolling

Abstract

The invention discloses a model learning method applicable to cross rolling, comprising the following steps: defining a fuzzy set theory domain U by a mathematical method, building a subordinate function of a fuzzy subset A, representing the fuzzy set by a ZADEH representation method, wherein the subordinate function in the theory domain U is a real-valued function, U is an element of a set [0, 1], determining the subordinate function by a fuzzy distribution method, carrying out clustering analysis on similarity level of steel types and specifications by an MINKOWSKI distance close degree method; determining cut set level of a satisfied fuzzy similar matrix by the golden mean method, determining a fuzzy rule number corresponding to the cut set level by a maximal tree method, thus building a rolling line forward slip learning fuzzy matrix, wherein with respect to the quantity with matrix value of 1, the learning can be completely inherited, with respect to the quantity with matrix value of 0, the learning is not inherited. According to the invention, the stability of the production process can be improved, a sleeve rising steel blocking accident resulted from model learning trend errors in the cross rolling process are effectively prevented, the precision of the product quality is improved, and free rolling is realized.

Description

technical field [0001] The invention belongs to the field of steel rolling automatic control, in particular to a model learning method suitable for cross rolling of hot-rolled strip steel. Background technique [0002] At present, the mathematical models adopted by domestic and foreign hot rolling mills basically include GE model, Siemens model and Japan Mitsubishi model. The Mitsubishi model uses the classic metal pressure processing formula, while the GE and Siemens models use the empirical formula, and their learning methods use the actual data of the last rolling for learning. No matter which model is suitable for batch rolling, the larger the batch size, the larger the batch size. The more rolling of the same steel type and specification, the more stable the rolling process and the higher the precision of product quality. On the contrary, when cross-rolling, not only the specifications and steel types are crossed, but also the furnace sequence is crossed, and there are...

AI Technical Summary

Problems solved by technology

On the contrary, when cross-rolling, not only the specifications and steel types are crossed, but also the furnace sequence is crossed, and there are problems of rolling instability and low product quality precision to varying degrees.

Images