Boots – shoes – and leggings
Patent
1996-10-16
1999-09-07
Grant, William
Boots, shoes, and leggings
364149, 364150, 364151, G05B 1942
Patent
active
059496789
DESCRIPTION:
BRIEF SUMMARY
TECHNICAL FIELD
The present invention relates to method of monitoring an industrial process which is dependent on a large number of parameters, available through measured data, in a way which makes it possible to control the process to the desired conditions by allowing the relevant variables of the process to be represented by the axes in a linear space with as many dimensions as the number of variables, whereupon the process is projected onto a plane or a three-dimensional room, such that a calculated model of the process is obtained on-line and by comparing the model of the process with a reference model of the process such that a distance to the reference model is obtained, whereupon, when observing a drift of some parameter, the process can be restored to at least one norm range for the process by acting upon a deviating variable.
BACKGROUND ART
For obtaining, for instance, the desired quality of a manufactured produce in a manufacturing process with the best economy or otherwise monitoring an industrial process or industrial application, it is necessary to control the processes as efficiently and optimally as possible. A manufacturing process includes many important variable quantities (here only referred to as variables), the values of which are affected by the variations of the variables during the course of the process. The optimum result is achieved if the process-monitoring operator or the process-monitoring member is able to handle and control all the process-influencing variables in one and the same operation.
A conventional method of optimizing a process is to consider one variable at a time only, one-dimensional optimization. All the variables are fixed except one, whereupon the non-fixed variable is adjusted to an optimum result. Thereafter, the free variable is fixed and one of the other variables adjusted, and so on.
When the process variables have been set in this way one by one, it is supposed that the best working point of the process has been obtained. However, the fact is that this is not the whole truth. The process may still be far from its optimum working point, since the method does not take the mutual influence of the process variables into account. The difficulty of this method is to obtain a total overview of the process based on a number of mutually independent process variables as necessitated by such a view. It is only when the relationship between these variables can be interpreted correctly that the process operator gets a real overview and understanding of the process.
An operator is limited by his or her human ability to understand and control only a limited number of variables per unit of time. A process monitoring system measures up to hundreds of variables, of which perhaps some 20 more or less directly control the process. Such a monitoring system requires a computer which can continuously register if and when slight variations occur in any of the variables.
A model of a process is realized substantially by two different types of modelling techniques, mechanistic and empirical modelling. Mechanistic models are used, for example, in physics. Data are used to discard or verify the mechanistic model. A good mechanistic model has the advantage of being based on established theories and is usually very reliable over a wide range. However, the mechanistic model has its limitations and is only applicable for relatively small, simple systems, whereas it is insufficient, if even possible to use, for building an axiom around a complex industrial process. Many attempts have been made to model processes with the aid of mechanistic models based on differential equations. An important disadvantage of these models, however, is that they are greatly dependent on the dependence of certain parameters on each other. Such parameters with great dependence on each other must be determined for the model to function. In the majority of cases it is very difficult to quantify them in a reliable manner. A consequence of this is that it is very difficult to obtain mechanistic models that w
REFERENCES:
patent: 4965742 (1990-10-01), Keirik
patent: 5196997 (1993-03-01), Kurtzberg et al.
patent: 5249120 (1993-09-01), Foley
patent: 5402333 (1995-03-01), Cardner
patent: 5408405 (1995-04-01), Mozumder et al.
patent: 5457625 (1995-10-01), Lim et al.
patent: 5481465 (1996-01-01), Itoh et al.
patent: 5488561 (1996-01-01), Berkowitz et al.
patent: 5610843 (1997-03-01), Chou
patent: 5617321 (1997-04-01), Frizelle et al.
ABB tidning Apr. 1993, "Overvakning och styrning av processforlopp i flera dimensioner", Bert Skarberg et al.
Control and Instrumentation, B.R. Upadhyaya, Nuclear Safety, vol. 26, No. 1, pp. 32-43, Jan.-Feb. 1985.
Power Signal Validation, Lin et al., Nuclear Technology, vol. 84, pp. 7-13, Jan. 1989.
Time Variable Parameter and Trend Estimation, Peter C.Young, Juornal of Forecasting, vol. 13, 179-210, 1994.
Sundin Karl Olov Lasse
Wold Svante
Grant William
Patel Ramesh
Telefonaktiebolaget LM Ericsson
LandOfFree
Method for monitoring multivariate processes 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 for monitoring multivariate processes, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for monitoring multivariate processes will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-1810661