Process Model Re-Parameterization for Rolling Mill Control Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods in the raw material industry struggle to accurately update and track model parameters that are not measurable, leading to deviations in the actual mode of facilities from the target mode, particularly for parameters like modulus of elasticity and temperature influence, which affect output product quality such as profile, contour, and planarity in rolling processes.
Innovation Solution
An optimization method that compares actual variables of multiple output products to expected variables to re-ascertain and re-parameterize model parameters, using an optimizer to minimize deviations and adjust parameters like roller modulus of elasticity and temperature influence, ensuring better alignment with target modes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If model parameters are permanently specified or externally supplied, then the control system operates with a fixed model, but the model deviations accumulate and reduce control accuracy over time
Solution Approach 1:
The patent implements feedback by comparing actual output variables with expected variables and using the deviations to re-ascertain model parameters. The control unit continuously updates model parameters based on the difference between actual and expected facility modes, ensuring the model remains accurate over time without requiring permanent external specification.
Solution Approach 2:
The system performs self-service by automatically updating its own model parameters using operational data. The control unit re-ascertains model parameters independently based on comparisons between actual and expected variables, eliminating the need for continuous external parameter specification and enabling the system to maintain itself.
2Measurement precision
If model parameters are re-ascertained after each output product, then the model stays updated, but the computational load and processing time increase significantly
Solution Approach 1:
The patent applies periodic action by re-ascertaining model parameters after the production of a plurality of output products rather than after each individual product. This periodic update approach maintains model accuracy while significantly reducing the frequency of computational updates and associated processing time requirements.
3Manufacturing precision
If the facility mode is controlled to match target mode, then product quality improves, but unmeasurable parameters like modulus of elasticity cannot be accurately tracked
Solution Approach 1:
The patent uses an intermediary approach by inferring unmeasurable model parameters (such as modulus of elasticity and temperature influence) from measurable output variables. Instead of directly measuring these difficult-to-detect parameters, the system uses the relationship between facility mode and output quality to indirectly determine their values through optimization processes.
Solution Approach 2:
The system changes parameters by re-ascertaining model parameters based on optimization criteria that compare actual versus expected output variables. This allows the system to track and update parameters that would otherwise be difficult to measure, using mathematical optimization rather than direct physical measurement.
4Reliability
If an optimizer is introduced to minimize deviations, then model accuracy improves, but the device complexity increases
Solution Approach 1:
The control unit performs multiple functions: it controls the facility mode, compares actual versus expected variables, re-ascertains model parameters, and minimizes deviations. By consolidating these functions into a single control unit, the system achieves improved model accuracy without proportionally increasing overall device complexity.
Data Source
AI summary
A model (8) is based on mathematical-physical equations. The model models the production of a particular output product (1) from at least one input product (2) supplied in each case to an installation in the raw materials industry on the basis of operation (B) of the installation. During production of the output products (1), the installation is controlled by a control device (5) in such a manner that particular actual operation (B) of the installation corresponds as far as possible to particular desired operation (B*) of the installation. The desired operation (B*) is determined by the control device (5) using the model (8) of the installation. The model (8) is parameterized according to a number of first model parameters (P1) for the purpose of modelling the installation. After a multiplicity of output products (1) have been produced in each case, actual sizes (A) of the output products (1) in the particular multiplicity are compared with expected sizes (A′) of the output products (1) in the particular multiplicity. On the basis of the comparison, the first model parameters (P1) are newly determined and the model (8) in the control device (5) is newly parameterized according to the new values of the first model parameters (P1). After this time, the desired operation (B*) is determined by the control device (5) using the newly parameterized model (8) of the installation in the raw materials industry. The expected sizes (A′) are determined by means of the model (8), wherein the determination of the expected sizes (A′) is based on the actual operation (B) of the installation.