Data processing: generic control systems or specific application – Specific application – apparatus or process – Product assembly or manufacturing
Reexamination Certificate
1995-08-22
2001-07-17
Cuchlinski, Jr., William A. (Department: 3661)
Data processing: generic control systems or specific application
Specific application, apparatus or process
Product assembly or manufacturing
C700S048000, C700S031000, C700S150000, C700S037000, C700S023000, C700S047000, C700S044000, C700S130000, C700S140000, C700S143000, C700S144000, C706S044000
Reexamination Certificate
active
06263257
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method and a device for predicting properties of products of a textile processing process on the basis of a process model, proceeding from material properties which are relevant to the processing of a specific input product, as well as predicting configuration and setting variables which influence the course of the processing process.
BACKGROUND OF THE INVENTION
European Patent Application No. 93120587.6 (Shofner et al.) discloses a method for the optimum control of textile machines by means of a process model. The method proceeds from the fact that the process model is represented as a system of linear mathematical equations whose coefficients are determined by means of the method of multiple linear regression on the basis of results from trials which are carried out systematically.
One disadvantage of this method consists of the fact that the restrictions which are produced as a result of the presupposition of a system of linear equations as a process model do not satisfy in many ways the practical requirements as to the accuracy of the simulation of a process by means of such a process model. Since the functional relationships between input data, output data and configuration and setting parameters of a process are as a rule not linear, the simulation of a process by means of a linear model yields results which are usable to a certain extent at most in a narrow operating range.
A further disadvantage of this method consists of the fact that the practical implementation of the method of multiple linear regression for determining the coefficients of the equations of a system of linear equations requires a very good mathematical understanding on the part of the person carrying it out, if a model is to be constructed which is as practically suitable as possible in spite of these fundamental restrictions.
It is the object of the invention, as it is characterized in the claims, to produce a method and a device with which, using a process model, reliable predictions about the interactions of input variables, output variables and setting variables can be made possible in the processing of textile structures without the above mentioned disadvantages.
SUMMARY OF THE INVENTION
This object is achieved in that the relationships between the input variables, the output variables and the configuration and setting variables are prescribed by a regular system, generally designated as a neural network, of nonlinear equations which are at least approximately mutually identical.
This object is further achieved in that the coefficients which the structurally defined equations of such a neural network express are determined by means of an algorithm of the so-called “training method”, known as “back propagation”, for neural networks on the basis of results from trials which are carried out systematically.
This object is further achieved in that a given basic model in the area of the specific experience of a spinning concern can be matched using simple means in such a way that, in the area of this specific experience, it is reproduced exactly.
This object is further achieved in that at least one spinning machine is provided with measuring devices for determining the input variables, with a computer having an input and an output unit for data, with means for the transmission of data and signals between the computer and the measuring devices, with a program memory in the computer having a model structure which represents a neural network and with a data memory having data about the input products, the output products and configuration and setting parameters.
Significant advantages result from this which, for example, consist in the fact that the process model, because of the great adaptability of a neural network to nonlinear properties of a process to be modelled, can be matched over a greater range with good accuracy to the effective behavior of such a process.
A further advantage lies in the fact that, because of the great adaptability of a neural network to nonlinear properties of a process to be modelled, during the determination of the coefficients which are expressed by the equations on which the model is based, no specific mathematical knowledge and/or insights into the mathematical relationships are necessary and the operations which are necessary for this can also be undertaken by personnel who do not have any special mathematical training.
A further advantage resides in the fact that, in this way, not only can predictions be made about closed, that is to say unitary, processing operations such as, for example, a spinning process from which a yarn is produced. Two or more processing operations which take place one after another can equally well be brought together into a combined process model. This is even possible when, during this processing, processes take place which are distinguished by pronounced non-linearities.
REFERENCES:
patent: 5282261 (1994-01-01), Skeirik
patent: 5285377 (1994-02-01), Sugasaka et al.
patent: 5414619 (1995-05-01), Katayama et al.
patent: 5477444 (1995-12-01), Bhat et al.
patent: 5553196 (1996-09-01), Takatori et al.
patent: 0573656 (1993-12-01), None
patent: 0604876 (1994-07-01), None
European Search Report of Jan. 11, 1996.
Cuchlinski Jr. William A.
McDieunel Marc
LandOfFree
Method and device for determining relevant variables in the... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and device for determining relevant variables in the..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and device for determining relevant variables in the... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2530529