Patent US5388186A successfully refreshed from Google Patents.
US5388186A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Differential process controller using artificial neural networks
Assignee: AT&T Inc
•
Filing: 1993-02-01
•
Granted: 1995-02-07
•
Expires: 2013-02-01
Abstract
Abstract Control of a process in accordance with both optimal process values (t d ), which may be fixed or slowly varying, and actual process output values (t r-1 ) generated during a previous interval (r-1) is accomplished by a differential process controller (10). The controller (10) employs two artificial neural networks (36 and 38), each generating a separate intermediate control vector for controlling the process in accordance with a separate one of the vectors t d and t r-1 . A first summing amplifier (42) computes the difference between the intermediate control vectors and generates a differential control vector which varies accordingly. A second summing amplifier (44) sums the differential control vector, together with the output signal of the summing amplifier generated during the immediately previous interval (r-1), to generate a control signal c r for controlling the process.
Inventors
- 👤 Chinmoy B. Bose
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-04 18:21
CPC Classifications
G; G05; G05B; G05B13/00; G05B13/02; G05B13/0265; G05B13/027; Y; Y10; Y10S; Y10S706/00; Y10S706/902; Y10S706/903
Claims Analysis
Total Claims: 10
Claim 1
Independent Claim
263 words
A differential process controller for controlling a process at a time r in accordance with both an optimal process output value vector (t d ), whose components represent desired process values, and an actual process value vector (t r-1 ), whose components represent actual process values measured during an immediately previous interval (r-1), comprising: first processing means supplied with the optimal process output value vector t d for generating a first vector varying in accordance therewith; a first artificial neural network for processing the first vector in accordance with past actual process values to yield a first intermediate control vector for controlling the process; second processing means supplied with the actual process value vector t r-1 for generating a second vector varying in accordance therewith; a second artificial neural network for processing the second vector in accordance with past actual process values to yield a second intermediate control vector for controlling the process; a first summing amplifier for producing an output vector varying in accordance with the difference between the first and second intermediate control vectors; a second summing amplifier having a first input supplied with the output vector of the first summing amplifier, and a second input, for generating a control vector for controlling the process in accordance with the sum of the signals at the first and second summing amplifier inputs; and a store and forward means supplied with the control vector from the second summing amplifier for storing said control vector and thereafter supplying said control vector to said second input of said second summing amplifier after a prescribed interval.
Claim 2
Dependent Claim
263 words
A differential process controller for controlling a process at a time r in accordance with both an optimal process output value vector (t d ), whose components represent desired process values, and an actual process value vector (t r-1 ), whose components represent actual process values measured during an immediately previous interval (r-1), comprising: first processing means supplied with the optimal process output value vector t d for generating a first vector varying in accordance therewith; a first artificial neural network for processing the first vector in accordance with past actual process values to yield a first intermediate control vector for controlling the process; second processing means supplied with the actual process value vector t r-1 for generating a second vector varying in accordance therewith; a second artificial neural network for processing the second vector in accordance with past actual process values to yield a second intermediate control vector for controlling the process; a first summing amplifier for producing an output vector varying in accordance with the difference between the first and second intermediate control vectors; a second summing amplifier having a first input supplied with the output vector of the first summing amplifier, and a second input, for generating a control vector for controlling the process in accordance with the sum of the signals at the first and second summing amplifier inputs; and a store and forward means supplied with the control vector from the second summing amplifier for storing said control vector and thereafter supplying said control vector to said second input of said second summing amplifier after a prescribed interval.
Claim 3
Dependent Claim
113 words
The apparatus according to claim 1 wherein each of the first and second artificial neural networks comprises: a first layer of nodes, each receiving a component of an input vector supplied to the artificial neural network and distributing the component of the input vector supplied thereto; a second layer of nodes, each supplied with the vector component distributed by each of the nodes in the first layer for transforming the vector component supplied thereto; and a third layer of nodes, each supplied with the vector component produced by each of the nodes in the second layer for further transforming the vector components supplied thereto to yield a component of an intermediate control vector.
Claim 4
Dependent Claim
116 words
The apparatus according to claim 1 wherein said first processing means comprises: a processing element for normalizing the optimal process output vector t d and the actual process value vector t r-1 to a mean value and to yield first and second output vectors, respectively, which vary accordingly; a third summing amplifier supplied with the first and second output vectors and generating a third output vector varying in accordance with the difference therebetween; scaling means for scaling the third output vector produced by the third summing amplifier by a constant value to yield a fourth output vector; a fourth summing amplifier for summing the fourth output vector of the scaling means with the second output vector.
Claim 5
Dependent Claim
50 words
The apparatus according to claim 3 further including: means for comparing the second output vector to the first output vector; and a signal generator for generating a fixed-value control vector for controlling the process when the difference between the first and second output vectors is within a prescribed tolerance value.
Claim 6
Dependent Claim
26 words
The apparatus according to claim 1 wherein the second processing means includes a processing element for normalizing the actual process value vector to a mean value.
Claim 7
Independent Claim
201 words
A method for controlling a process at a time r in accordance with both an optimal process output value vector (t d ), whose components represent desired process values, and an actual process value vector (t r-1 ), whose components represent actual process values measured during a previous interval (r-1), comprising the steps of: processing the optimal output value vector t d to yield a first vector varying in accordance therewith; processing the actual process value vector t r-1 to yield a second vector varying in accordance therewith; applying the first vector to an artificial neural network which sums and transforms the components of the first vector in accordance with past actual process output values to yield a first intermediate control vector; applying the second vector to an artificial neural network which sums and transforms the components of the second vector in accordance with past actual process output values to yield a second intermediate control vector; generating a differential control vector varying in accordance with the difference between the first and second intermediate control vectors; successively generating a final control vector in accordance with the sum of the differential control vector and the final control vector generated during a previous interval.
Claim 8
Independent Claim
201 words
A method for controlling a process at a time r in accordance with both an optimal process output value vector (t d ), whose components represent desired process values, and an actual process value vector (t r-1 ), whose components represent actual process values measured during a previous interval (r-1), comprising the steps of: processing the optimal output value vector t d to yield a first vector varying in accordance therewith; processing the actual process value vector t r-1 to yield a second vector varying in accordance therewith; applying the first vector to an artificial neural network which sums and transforms the components of the first vector in accordance with past actual process output values to yield a first intermediate control vector; applying the second vector to an artificial neural network which sums and transforms the components of the second vector in accordance with past actual process output values to yield a second intermediate control vector; generating a differential control vector varying in accordance with the difference between the first and second intermediate control vectors; successively generating a final control vector in accordance with the sum of the differential control vector and the final control vector generated during a previous interval.
Claim 9
Dependent Claim
85 words
The method according to claim 6 wherein the optimal process value vector is processed by the steps of: normalizing the optimal process value vector t d to a mean value; normalizing the actual process output value vector t r-1 to a mean value; generating a difference vector in accordance with the normalized-to-mean values of the vectors t d and t r-1 ; scaling the difference vector by a constant; and summing the scaled difference vector with the mean value of the vector t r-1 .
Claim 10
Dependent Claim
24 words
The method according to claim 6 wherein the actual process output value vector is processed by normalizing it the vector to a mean value.
References Cited (Backward Citations)
US5159660A
Cited
US5285377A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1998-02-06 | 1998-08-08 | 1999-02-07 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2002-02-06 | 2002-08-08 | 2003-02-07 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2006-02-06 | 2006-08-08 | 2007-02-07 | $7700.00 | Window Closed / Lapsed |