Automated method for building a model

• US 6,243,696 B1
• Filed: 03/24/1998
• Issued: 06/05/2001
• Est. Priority Date: 11/24/1992
• Status: Expired due to Term

• Current Assignee: Rockwell Automation Technologies Incorporated (Allen-Bradley Co., Inc.)
• Original Assignee: Pavilion Technologies Incorporated (Rockwell Automation, Inc.)
• Inventors: Keeler, James D., Hartman, Eric J., Godbole, Devendra B., Ferguson, R. Bruce II, Telang, Jitendra W., Ellinger, Joshua, Krauskop, John, Kempf, Jill L., Piche, Steve, Arbila, Laura, Strauss, Audrey, OʼHara, Steven A.
• Primary Examiner(s): Davis, George B.
• Application Number:
  US09/047,853
• Time in Patent Office: 1,169 Days
• Field of Search: 706/21, 706/906, 706/907
• US Class Current: 706/21
• CPC Class Codes: G06N 3/047 Probabilistic or stochastic networks
• CPC Class Codes: G06N 3/049 Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
• CPC Class Codes: Y10S 706/906 Process plant
• CPC Class Codes: Y10S 706/907 Power plant

Abstract

A method for building a model of a system includes first extracting data from a historical database (310). Once the data is extracted, a dataset is then created, which dataset involves the steps of preprocessing the data. This dataset is then utilized to build a model. The model is defined as a plurality of transforms which can be utilized to run an on-line model. This on-line model is interfaced with the historical database such that the variable names associated therewith can be downloaded to the historical database. This historical database can then be interfaced with a control system to either directly operate the plant or to provide an operator an interface to various predicted data about the plant. The building operation will create the transform list and then a configuration step is performed in order to configure the model to interface with the historical database. When the dataset was extracted, it is unknown whether the variables names are still valid. It is therefore necessary to read and write the various variables to the database to determine if they are in fact valid. Further, the predicted output values, which may not have been a part of the historical database, need to be verified. Once these are verified, then the on-line model can be created to generate the predicted value for transfer to the control system. During this period of time, the control system must be disabled.

Claims

1. A method for creating a representation of a plant and incorporating it into a run time prediction system for generating predicted output values representing the operating parameters of the plant during operation thereof, comprising the steps of:

providing a historical database representing the operation of the plant and comprised of data associated with plant inputs and plant outputs;

extracting data from the historical database and creating a dataset of variables corresponding to the inputs and outputs from the historical database;

creating an off-line predictive model of the plant utilizing the created dataset to predict a plant output and defined by off-line model parameters;

creating an on-line model for generating predicted output values in real time during the operation the a plant and defined by on-line model parameters; and

replacing the on-line model parameters with the off-line model parameters after generation thereof.

2. The method of claim 1, wherein the historical database is integral with the operation of the plant in real time.

3. The method of claim 1, wherein the inputs to the plant are manipulatable variables for operating the plant.

4. The method of claim 1, wherein the outputs of the plant in the historical database constitute measurable variables which are measured outputs from the plant during operation thereof.

5. The method of claim 1, wherein the step of extracting data from the database and creating a dataset includes the step of preprocessing the data extracted from the historical database to modify the data in accordance with a predetermined preprocess routine.

6. The method of claim 1, wherein the step of creating an off-line predictive model comprises training a neural network with the data in the dataset to create a stored representation thereof with the neural network mapping the inputs thereto through the stored representation to provide the predicted output values.

7. The method of claim 1, wherein the predicted output values comprise an unmeasured output in the historical database and wherein the outputs in the historical database constitute measurable outputs.

8. The method of claim 1, and further comprising the step of testing the off-line predicted model with data in the historical database prior to the step of replacing.

9. The method of claim 8, wherein the step of testing comprises reading select ones of the variables defined by the off-line predictive model from the historical database to ensure that the variables in the created off-line predictive model can be read.

10. The method of claim 8, wherein the step of testing comprises writing to the historical database values that correspond to variables in the created off-line predictive model to ensure that data can be written to these variable locations in the historical database.

11. The method of claim 8, wherein the step of testing comprises reading from a location in the historical database corresponding to the predicted output value as defined by the off-line predictive model to ensure that the predicted output value name in the created off-line predictive model corresponds to a variable name in the historical database.

12. The method of claim 8, wherein the step of testing comprises writing the predicted output value to a corresponding location in the historical database to ensure that there is such a location in the historical database to ensure that the step of replacing will replace the on-line model parameters with the correct off-line model parameters.

13. The method of claim 8, wherein the operation of the on-line model will be paused during the step of testing.

14. A method for assisting a user in creating a representation of a plant and incorporating it into a run time prediction system for generating predicted output values representing operating parameters of the plant during operation thereof, comprising the steps of:

providing a display;

providing a historical database;

displaying an extraction graphical assist device with associated interactive instructions to assist the user in extracting data from a historical database representing the operation of the plant and comprised of data associated with plant inputs and plant outputs, and creating a dataset of variables corresponding to the inputs and outputs from the historical database when the user has completed the instructions;

displaying a model building graphical assist device with associated interactive instructions to assist the user in creating an off-line predictive model of the plant utilizing the created dataset to predict a plant output and defined by off-line model parameters;

displaying a graphical assist device with associated interactive instructions to assist the user in replacing on-line model parameters of an on-line model with the off-line model parameters after generation thereof, which the on-line model is operable to generate predicted output values in real time during the operation the a plant and defined by the on-line model parameters.

15. The method of claim 14, wherein the step of assisting the user in extracting data from the database and creating a dataset includes the step of assisting the user in preprocessing the data extracted from the historical database to modify the data in accordance with a predetermined preprocess routine.

16. The method of claim 14, wherein the step of assisting the user in creating an off-line predictive model comprises assisting the user in training a neural network with the data in the dataset to create a stored representation thereof with the neural network mapping the inputs thereto through the stored representation to provide the predicted output values.

17. The method of claim 14, and further comprising the step of displaying a test graphical assist device with associated interactive instructions to assist the user in testing the off-line predicted model with data in the historical database prior to the step of assisting to replace.

18. The method of claim 17, wherein the step of assisting the user to test the off-line model comprises assisting the user in reading select ones of the variables defined by the off-line predictive model from the historical database to ensure that the variables in the created off-line predictive model can be read.

19. The method of claim 17, wherein the step of assisting the user to test the off-line model comprises assisting the user in writing to the historical database values that correspond to variables in the created off-line predictive model to ensure that data can be written to these variable locations in the historical database.

20. The method of claim 17, wherein the step of assisting the user to test the off-line model comprises assisting the user in reading from a location in the historical database corresponding to the predicted output value as defined by the off-line predictive model to ensure that the predicted output value name in the created off-line predictive model corresponds to a variable name in the historical database.