Method, system and paperboard production machine for estimating paperboard quality parameters

Patent metadata

Field	Value
Publication number	US-12282012-B2
Application number	US-202017636865-A
Country	US
Kind code	B2
Filing date	Aug 19, 2020
Priority date	Aug 28, 2019
Publication date	Apr 22, 2025
Grant date	Apr 22, 2025

Abstract

A computer-implemented method for estimating at least one quality parameter of paperboard produced in a paperboard production subprocess of a paperboard processing pipeline by a data-driven module having a preprocessing module and a machine-learning module. Sensor data is acquired along the processing pipeline. Features are extracted from the sensor data by the preprocessing module. The machine-learning module is trained to reproduce target quality values from historical features. After training, the machine-learning module processes real-time features and estimates at least one quality parameter. A system implements the computer-implemented method and to a paperboard production machine includes such a system.

First claim

The invention claimed is:

1. A computer-implemented method for estimating at least one quality parameter of paperboard produced in a paperboard production subprocess of a paperboard processing pipeline during the production by a data-driven module comprising a preprocessing module and a machine-learning module, the method comprising:
acquiring sensor data from at least one processing step of the processing pipeline and transferring the sensor data to a data repository,
determining at least one historical feature by the preprocessing module by at least partially evaluating historical sensor data that was acquired during at least one previously produced batch of paperboard and retrieved from the data repository,
training the machine-learning module to reproduce from the at least one historical feature a target value of the at least one quality parameter, wherein the target value is determined for a previously produced batch of paperboard corresponding to the historical sensor data from which the historical feature was determined,
determining at least one real-time feature by the preprocessing module from a stream of current sensor data being acquired from a currently produced batch of paperboard and retrieved from the data repository,
determining an estimate for the at least one quality parameter from the at least one real-time feature by the trained machine-learning module and providing the estimate as an output value,
adjusting the paperboard production subprocess in response to the output value not being within a predetermined range,
wherein at least one range for the at least one quality parameter is provided as additional input to the data-driven module, wherein for each range a probability value for the respective quality parameter to fall within the respective range is determined as an output value,
wherein for one or more of the probability values determined by the data-driven module for the range of the at least one quality parameter, each probability value is compared with a probability limit assigned to the respective range and the respective quality parameter, resulting in one Boolean comparison value per range and quality parameter, and
wherein a predetermined Boolean expression depending on the one Boolean comparison value is evaluated, wherein an alarm is triggered, when the predetermined Boolean expression returns logical true.

2. The computer-implemented method according to claim 1 ,
wherein historical sensor data is split into a plurality of intervals, wherein episodes of historical sensor data that are associated with error messages or inconsistencies are omitted.

3. The computer-implemented method according to claim 2 ,
wherein historical sensor data is split into a plurality of intervals, comprising 10000 intervals or more.

4. The computer-implemented method according to claim 1 ,
wherein a delay group is determined for a processing step as the minimum latency of a parameter change in that processing step to affect the outcome of the production subprocess, wherein current sensor data acquired within the latency of the respective delay group backwards from a current production time stamp is excluded from evaluation by the data-driven module.

5. The computer-implemented method according to claim 1 ,
wherein at least one feature comprises at least one time series describing a variation of at least one parameter of the sensor data along discrete time windows, wherein the machine-learning module performs a one-dimensional convolution of the at least one time series along a time axis.

6. The computer-implemented method according to claim 5 ,
wherein for a discrete time window at least one statistical parameter is determined as sample value of the at least one time series.

7. The computer-implemented method according to claim 6 ,
wherein the at least one statistical parameter comprises a mean value and/or a standard deviation value.

8. The computer-implemented method according to claim 1 ,
wherein validity and/or consistency of current sensor data is evaluated and optionally logged in the data repository by the preprocessing module.

9. The computer-implemented method according to claim 1 ,
wherein one quality parameter is a Z-strength of the paperboard.

10. The computer-implemented method according to claim 1 ,
wherein one quality parameter is a Scott bond of the paperboard.

11. The computer-implemented method according to claim 1 ,
wherein for the at least one quality parameter a first range is a set of invalid values lower than a lower valid limit, a second range is a set of valid values between the lower valid limit and an upper valid limit and a third range is a set of invalid values higher than the upper valid limit, wherein an alarm is triggered, when the probability value associated with the first range exceeds a predetermined first probability limit or when the probability value associated with the second range falls below a predetermined second probability limit or when the probability value associated with the third range exceeds a predetermined third probability limit.

12. A system, comprising:
at least one sensor,
a data repository, and
a computer,
wherein the at least one sensor is designed to acquire sensor data in a processing step of a processing pipeline comprising a paperboard production subprocess,
wherein the data repository is designed to receive and persistently store sensor data from the at least one sensor and to transfer sensor data towards the computer, and
wherein the computer implements the computer-implemented method according to claim 1 .

13. The system according to claim 12 ,
wherein the data repository is formed as a cloud storage that can be accessed by an internet protocol based communication protocol stack.

14. The system according to claim 12 , further comprising:
an alarm, wherein the alarm is designed to be triggered by the computer and to indicate a need of manual interaction with the paperboard production subprocess when being triggered.

15. A paperboard production machine designed to produce paperboard or a semi-product of paperboard, comprising:
the system according to claim 12 .

Assignees

• Siemens Ag
• Siemens Energy Global Gmbh & Co Kg

Inventors

• Alpire Adam
• Eisenstecken Thomas
• Hildebrand Michael
• Kisslinger Ferdinand

Classifications

• D21J1/00

  Fibreboard (preparation of pulp compositions or addition of chemical agents D21B, D21C, D21H; formation of the wet web D21F) · CPC title

• D21F7/00

  Other details of machines for making continuous webs of paper · CPC title

• G06N20/00

  Machine learning · CPC title

• G01N33/34 Primary

  Paper · CPC title

• G01N33/346 Primary

  Paper sheets · CPC title