On-line process control neural network using data pointers

• US 5,167,009 A
• Filed: 08/03/1990
• Issued: 11/24/1992
• Est. Priority Date: 08/03/1990
• Status: Expired due to Term

• Current Assignee: Rockwell Automation Technologies Incorporated (Rockwell Automation, Inc.)
• Original Assignee: E. I. du Pont de Nemours and Company (Corteva, Inc.)
• Inventors: Skeirik, Richard D.
• Primary Examiner(s): MacDonald, Allen R.
• Application Number:
  US07/562,388
• Time in Patent Office: 844 Days
• Field of Search: 364/513, 364/148, 364/149, 364/150, 364/151, 364/164, 364/165, 364/807, 364/200, 364/474.15, 364/500, 364/501, 364/502, 364/503, 395/22, 395/26, 395/27, 395/906, 395/907, 395/914, 395/915, 395/11, 395/68
• US Class Current: 706/23
• CPC Class Codes: G05B 13/027 using neural networks only

Abstract

An on-line process control neural network using data pointers allows the neural network to be easily configured to use data in a process control environment. The inputs, outputs, training inputs and errors can be retrieved and/or stored from any available data source without programming. The user of the neural network specifies data pointers indicating the particular computer system in which the data resides or will be stored; the type of data to be retrieved and/or stored; and the specific data value or storage location to be used. The data pointers include maximum, minimum, and maximum change limits, which can also serve as scaling limits for the neural network. Data pointers indicating time-dependent data, such as time averages, also include time boundary specifiers. The data pointers are entered by the user of the neural network using pop-up menus and by completing fields in a template. An historical database provides both a source of input data and a storage function for output and error data.

Claims

1. A computer neural network process control system adapted for predicting output data provided to a controller used to control a process for producing a product having at least one product property, the computer neural network process control system comprising:

(a) an historical database comprising,

(1) storing means for storing a first training input data with an associated first timestamp, and for storing a first input data indicated by said associated first timestamp, and

(2) retrieving means, responsive to said storing means, for retrieving said first training input data and said first input data indicated by said associated first timestamp; and

(b) a trained neural network comprising,

(1) a first data pointer and a second data pointer, and

(2) predicting means, using said trained neural network, for predicting the output data which is stored in accordance with said first data pointer, the output data predicted in accordance with second input data specified by said second data pointer, wherein the output data is used to control the process.

2. The computer neural network process control system of claim 1, further comprising sensor means, responsive to said historical database, for providing said first input data.

3. The computer neural network process control system of claim 2, further comprising clock means an associated second timestamp to said historical database.

4. The computer neural network process control system of claim 1, wherein said second input data of said predicting means includes data indicative of past output of data.

5. The computer neural network process control system of claim 1, wherein said neural network comprises a software system adapted to run on a computer platform.

6. The computer neural network process control system of claim 1, further comprising output means for storing the output data in said historical database.

7. The computer neural network process control system of claim 1, wherein said first data pointer and said second data pointer are three-part pointers with additional qualifiers.

8. A computer neural network process control system adapted for predicting output data provided to a controller used to control a process for producing a product having at least one product property, the computer neural network process control system comprising:

(a) a database comprising,

(1) storing means for storing, in independently addressable storage, at least one data value,

(2) retrieving means, responsive to said storing means, for independently retrieving each of said at least one data value(s); and

(b) a neural network comprising,

(1) a data pointer specifying one of said at least one data value(s) in said database to be used as a first input data, and

(2) predicting means for predicting the output data, connected to request, from said retrieving means, said data value indicated by said data pointer, and connected to receive said first input data from said retrieving means.

9. The computer neural network process control system of claim 8, wherein said data pointer comprises means for storing a specification of a data location.

10. The computer neural network process control system of claim 8, wherein said data pointer comprises means for storing a specification of a data type.

11. The computer neural network process control system of claim 8, wherein said data pointer comprises means for storing a specification of a data item number.

12. The computer neural network process control system of claim 8, wherein said data pointer comprises means for storing a specification of an oldest time interval boundary.

13. The computer neural network process control system of claim 8, wherein said data pointer comprises means for storing a specification of a newest time interval boundary.

14. The computer neural network process control system of claim 8, wherein said data pointer comprises means for storing a specification of one specific point in time.

15. The computer neural network process control system of claim 8, wherein said data pointer comprises means for storing a specification(s) of a high limit value and/or a low limit value.

16. The computer neural network process control system of claim 15, wherein said predicting means further comprises means for scaling a value in accordance with said high limit value and/or said low limit value.

17. The computer neural network process control system of claim 15, wherein said predicting means further comprises means for clamping an output data in accordance with said high limit value and/or said low limit value.

18. The computer neural network process control system of claim 8, further comprising

(c) a user interface for presenting a template to a user for entering said data pointer.

19. The computer neural network process control system of claim 18, wherein

said template comprises user-alterable fields for entering a data system specification, a data type specification, and/or a data item specification, and

said data pointer comprises means for storing said data system specification, said data type specification, and/or said data item specification.

20. The computer neural network process control system of claim 8, further comprising:

(c) a controller, connected to use the output data as its controller input data, and connected to control the process in accordance with a controller parameter.

21. The computer neural network process control system of claim 20, wherein said data pointer comprises means for storing a specification(s) of a high limit value and/or a low limit value.

22. The computer neural network process control system of claim 21, further comprising comparing means for disabling the operation of said controller when the value of an input, training input, or output is above said high limit value and/or below said low limit value.

23. The computer neural network process control system of claim 8, wherein said neural network further comprises,

(3) a second data pointer specifying a storage location in said database to be used to store a first output data.

24. The computer neural network process control system of claim 8, herein said neural network further comprises,

(3) training means for training said neural network in accordance with training input data;

and

(4) a second data pointer specifying one of said at least one data value(s) in said database to be used as a first training input data.