Patent US6255167B1 successfully refreshed from Google Patents.
US5778151A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Method and control device for controlling a material-processing process
Assignee: Siemens AG
•
Filing: 1994-05-16
•
Granted: 1998-07-07
•
Expires: 2014-05-16
Abstract
Abstract In the control of a material-processing process in a regulated system, a preliminary adjustment of the system takes place at the beginning of each process cycle as a function of a precalculated process parameter. A material characteristic which is relevant for the process and which in turn is dependent on state variables (such as the composition of the material and its temperature), is included in an advance calculation of the process parameter. The relationship between the state variables and the material property is modelled in a neural network which forms a prediction value for the material property on its output side. As a function of the deviation between the prediction value and an actual value for the material property which is determined based on measuring the process parameter during the process cycle, an adaptive change of the network parameters takes place in the sense of reducing this deviation.
Inventors
- 👤 Otto Gramckow
- 👤 Thomas Martinetz
- 👤 Thomas Poppe
- 👤 Gunter Sorgel
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-31 08:37
CPC Classifications
G; G05; G05B; G05B13/00; G05B13/02; G05B13/0265; G05B13/027; B; B21; B21B; B21B37/00; B21B37/16; B21B2261/00; B21B2261/22; B21B2265/00; B21B2265/12
Claims Analysis
Total Claims: 25
Claim 1
Independent Claim
175 words
A method for controlling the processing of a material in a regulated system having at least one process cycle, the method comprising: performing a preliminary adjustment of the regulated system at a beginning of each process cycle of the material-processing process as a function of a precalculated process parameter, wherein precalculation of said process parameter includes a material property that is relevant to said material-processing process and is also dependent on a plurality of state variables of said material; applying said plurality of state variables to a neural network prior to the beginning of the process cycle, said neural network including variable network parameters; forming in said neural network a prediction value for said material property at an output of said neural network; measuring said process parameter during said process cycle; determining an actual value of said material property after said process cycle by recalculation; adaptively changing said variable network parameters as a function of a deviation between said prediction value and the actual value of the material property in order to reduce said deviation.
Claim 2
Independent Claim
175 words
A method for controlling the processing of a material in a regulated system having at least one process cycle, the method comprising: performing a preliminary adjustment of the regulated system at a beginning of each process cycle of the material-processing process as a function of a precalculated process parameter, wherein precalculation of said process parameter includes a material property that is relevant to said material-processing process and is also dependent on a plurality of state variables of said material; applying said plurality of state variables to a neural network prior to the beginning of the process cycle, said neural network including variable network parameters; forming in said neural network a prediction value for said material property at an output of said neural network; measuring said process parameter during said process cycle; determining an actual value of said material property after said process cycle by recalculation; adaptively changing said variable network parameters as a function of a deviation between said prediction value and the actual value of the material property in order to reduce said deviation.
Claim 3
Dependent Claim
72 words
The method of claim 1, wherein after said process cycle said plurality of state variables are recalculated based on measurements taken during said process cycle and said recalculated state variables are passed to said neural network to generate a subsequent value as a network response, wherein a deviation between said network response of said neural network and the actual value of the material property is used to adapt said variable network parameters.
Claim 4
Dependent Claim
33 words
The method of claim 2, wherein adaptation of the variable network parameters of said neural network takes place on-line, such that adaptation of the variable network parameters takes place after every process cycle.
Claim 5
Dependent Claim
45 words
The method of claim 3, wherein deviations from prior process cycles are used to adapt the network parameters of said neural network after each individual process cycle, in addition to a current deviation between the network response and the actual value for the material property.
Claim 6
Dependent Claim
28 words
The method of claim 4 wherein said neural network is pre-adapted with simulated process values for the state variables and the material property before a first process cycle.
Claim 7
Dependent Claim
28 words
The method of claim 4 wherein said neural network is pre-adapted with operational process values for the state variables and the material property before the first process cycle.
Claim 8
Dependent Claim
64 words
The method of claim 5 wherein in a rolling process, said material characteristic is a material strength of a rolling stock in said rolling process and said process parameter is a rolling force of said rolling stock, such that a prediction of the material strength of the rolling stock takes place based on an advance calculation of the rolling force by said neural network.
Claim 9
Dependent Claim
56 words
The method of claim 7 wherein the state variables of said rolling stock are concentrations of chemical elements selected from the group of Carbon (C), Silicon (Si), Manganese (Mn), Phosphorus (P), Sulfur (S), Aluminum (Al), Nitrogen (N), Copper (Cu), Chromium (Cr), Nickel (Ni), Tin (Sn), Vanadium (V), Molybdenum (Mo), Titanium (Ti), Niobium (Nb), and Boron (B).
Claim 10
Dependent Claim
20 words
The method of claim 8, wherein said state variables include temperature, geometric dimensions of the rolling stock, and rolling speed.
Claim 11
Dependent Claim
88 words
The method of claim 9 wherein said neural network includes an input layer having an input element for each of said state variables, a hidden layer totaling said state variables in totaling elements with individual weight factors to form an answer at an output of said hidden layer according to a function with a sigmoid progression, and an output layer following said hidden layer having an output element totaling answers of the elements of the hidden layer, each with an individual weight factor to yield said prediction value.
Claim 12
Dependent Claim
52 words
The method according to claim 10 wherein the input layer and the hidden layer of said neural network each has an additional element for receiving a constant value on an input side for computation of the answers of the elements of said hidden layer and the prediction value of said output element.
Claim 13
Dependent Claim
97 words
A control device for controlling a material-processing process in a regulated system having at least one program cycle, comprising: a device for preliminary adjustment of the system at a beginning of each process cycle as a function of a precalculated process parameter, where a material property which is dependent on state variables of the material and which is relevant for the material-processing process, is included in the precalculation; and a neural network coupled to said device for preliminary adjustment having variable network parameters for adaptive modelling of a dependence between the material property and the state variables.
Claim 14
Dependent Claim
97 words
A control device for controlling a material-processing process in a regulated system having at least one program cycle, comprising: a device for preliminary adjustment of the system at a beginning of each process cycle as a function of a precalculated process parameter, where a material property which is dependent on state variables of the material and which is relevant for the material-processing process, is included in the precalculation; and a neural network coupled to said device for preliminary adjustment having variable network parameters for adaptive modelling of a dependence between the material property and the state variables.
Claim 15
Dependent Claim
84 words
The control device of claim 12, wherein in said neural network a prediction value is formed for said material property at an output of said neural network, such that an actual value of said material property is determined by recalculation based on process parameters measured during said process cycle and said network parameters of said neural network are adaptively changed as a function of a deviation between the prediction value and the actual value of the material property in order to reduce said deviation.
Claim 16
Dependent Claim
72 words
The control device of claim 13, wherein after said process cycle said state variables are recalculated based on measurements taken during said process cycle, such that said recalculated state variables are passed to said neural network to generate a subsequent value as a network response, wherein a deviation between said network response of said neural network and the actual value of the material property is used to adapt said variable network parameters.
Claim 17
Dependent Claim
34 words
The control device of claim 14, wherein adaptation of the variable network parameters of said neural network takes place on-line, such that adaptation of the variable network parameters takes place after every process cycle.
Claim 18
Dependent Claim
46 words
The control device of claim 15 wherein deviations from prior process cycles are used to adapt the network parameters of said neural network after each individual process cycle, in addition to a current deviation between the network response and the actual value for the material property.
Claim 19
Dependent Claim
29 words
The control device of claim 16 wherein said neural network is pre-adapted with simulated process values for the state variables and the material property before a first process cycle.
Claim 20
Dependent Claim
29 words
The control device of claim 16 wherein said neural network is pre-adapted with operational process values for the state variables and the material property before the first process cycle.
Claim 21
Dependent Claim
71 words
The control device of claim 17 further comprising: a rolling process including a rolling stock, such that said material characteristic is a material strength of said rolling stock in said rolling process and said process parameter is a rolling force of said rolling stock, such that a prediction of the material strength of the rolling stock takes place based on an advance calculation of the rolling force by said neural network.
Claim 22
Dependent Claim
57 words
The control device of claim 19 wherein the state variables of said rolling stock are concentrations of chemical elements selected from the group of Carbon (C), Silicon (Si), Manganese (Mn), Phosphorus (P), Sulfur (S), Aluminum (Al), Nitrogen (N), Copper (Cu), Chromium (Cr), Nickel (Ni), Tin (Sn), Vanadium (V), Molybdenum (Mo), Titanium (Ti), Niobium (Nb), and Boron (B).
Claim 23
Dependent Claim
21 words
The control device of claim 20 wherein said state variables include temperature, geometric dimensions of the rolling stock, and rolling speed.
Claim 24
Dependent Claim
102 words
The control device of claim 21 wherein said neural network further comprises: an input layer having an input element for each of said state variables; a hidden layer, such that said state variables from said input layer are totaled in totaling elements of said hidden layer with individual weight factors to form an answer at an output of said hidden layer according to a function with a sigmoid progression; and an output layer, following said hidden layer, having an output element which totals answers of the elements of the hidden layer, each with an individual weight factor to yield said prediction value.
Claim 25
Dependent Claim
52 words
The control device of claim 22, wherein the input layer and the hidden layer of said neural network each has an additional element for receiving a constant value on an input side for computation of the answers of the elements of said hidden layer and the prediction value of said output element.
References Cited (Backward Citations)
DE534221C
Cited
US5119468A
Cited
DE4008510A1
Cited
US5303385A
Cited
US5305230A
Cited
US5311421A
Cited
EP0460892A2
Cited
US5159660A
Cited
US5175678A
Cited
US5479358A
Cited
US5581459A
Cited
US5402519A
Cited
US5406581A
Cited
US5268834A
Cited
EP0531712A2
Cited
DE4131765A1
Cited
US5444819A
Cited
US5559690A
Cited
US5513097A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 2001-07-06 | 2002-01-05 | 2002-07-07 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2005-07-06 | 2006-01-05 | 2006-07-07 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2009-07-06 | 2010-01-05 | 2010-07-07 | $7700.00 | Window Closed / Lapsed |