EP4225992A1 - Verfahren zur kategorisierung und verfahren zur vermeidung von störungen an einer papiermaschine, insbesondere von bahnabrissen - Google Patents
Verfahren zur kategorisierung und verfahren zur vermeidung von störungen an einer papiermaschine, insbesondere von bahnabrissenInfo
- Publication number
- EP4225992A1 EP4225992A1 EP21769321.7A EP21769321A EP4225992A1 EP 4225992 A1 EP4225992 A1 EP 4225992A1 EP 21769321 A EP21769321 A EP 21769321A EP 4225992 A1 EP4225992 A1 EP 4225992A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- category
- production
- disruption
- categories
- signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- D—TEXTILES; PAPER
- D21—PAPER-MAKING; PRODUCTION OF CELLULOSE
- D21G—CALENDERS; ACCESSORIES FOR PAPER-MAKING MACHINES
- D21G9/00—Other accessories for paper-making machines
- D21G9/0009—Paper-making control systems
Definitions
- the invention relates to a method for categorizing production disruptions in a paper machine, in particular web breaks, and it relates to a method for avoiding production disruptions in a paper machine, in particular web breaks, using a method for the above-mentioned categorization. Furthermore, it relates to a computer program product for executing one of these methods on a computer unit.
- All machines for the production and/or processing of fibrous webs are regarded here as paper machines, in addition to machines for the production of paper webs also, for example, so-called board machines, pulp machines, coating machines or paper processing machines.
- the object of the invention is now to develop a method that contributes or is suitable to avoiding disruptions in production, in particular web breaks on a paper machine, with a higher probability and thus leading to higher productivity of the paper machine. Furthermore, the task is to provide a corresponding computer program product for carrying out this method.
- the object is achieved by a method according to claim 1 or by a method according to claim 8. Furthermore, it is solved by a computer program product according to claim 7 or according to claim 13. Further advantageous features are specified in the respective dependent claims. According to the invention, the method according to claim 1 is characterized in that the following steps are carried out:
- DCS process control system
- QCS quality control system
- steps 1.2.1 and 1.2.2 can be repeated in parallel with the first execution and do not have to be executed one after the other.
- a significant advantage of the method according to the invention is that it allows production faults of the same type to be recognized as such and combined into one category.
- a separate predictive model is created for each category of production disruption, which can indicate a signifier for the category-specific disruption risk.
- the production control system (DCS) or the operator thus receives important information on how to improve the setting of the production parameters, since a separate risk of occurrence can be determined for each category of production disruption, in particular for each type of web break. This helps prevent future production disruptions by changing production settings in a timely manner.
- the method offers the possibility of using data from the production system without complex synchronization and without complex data cleansing. Not necessary for the categorization or not usable signal values are simply not reused. In particular, they can be assigned to a remaining amount.
- the signals that are provided in the first data set from the production system can come from sensors or measuring devices, for example, they can be control signals or status values from motors or valves, or they can be actual values or setpoint values from control loops. Furthermore, signals can also be state values that are available in the system and are discontinuously assigned to specific points in time, such as laboratory values for paper reels or the running time of fabrics. And the signals can also be image data or data determined from images, or acoustic data. These signal values can also originate from several process control systems (DCS) and from several quality control systems of the production system. Because signals are used not only from a DCS but also from a QCS, quality changes are also detected, which significantly improves segmentation into different categories and model building. For example, quality changes can indicate changes in raw material quality or incorrect production settings that increase the risk of disruption.
- DCS process control systems
- the data can be made available, for example, by reading in or transferring or storing the data.
- the paper machine with the associated stock preparation, raw material preparation and raw material supply and the existing further processing, as well as the existing ancillary units and laboratories, is regarded as a production system. Or part of it.
- the first data volume can include all or a large part of the available signals of the production system.
- the efficiency of the method can be improved if the number of signals provided in the first data volume is already reduced via a process model. For this purpose, for example, correlations between signals can be evaluated and/or it Only the signals that are relevant to the production disruptions to be categorized can be used.
- Signal values that describe the production in a certain period of time in the past are referred to as historical signal values. For example, over several weeks or months.
- the data can originate from a data memory of the production system or can be specifically recorded over a certain period of time.
- the disruption times at which a production disruption, in particular a web break, occurred can be selected by manual input or by entering lists of disruption times. This selection can preferably take place automatically in the method by using a single signal or a plurality of signals from the first data volume as a detection for a fault.
- a fault sensor in particular a breakage sensor or a quality sensor, can be used for this.
- Signals that are indirectly coupled to the production disruption can also be used, such as the speeds of certain drives or pumps or spray pipes on a pulper, which are activated in the event of a break.
- the characteristic numerical value can be assigned to a signal change using an algorithm, supported in particular by methods of artificial intelligence.
- time intervals of different lengths before the observed disturbance time can also be used for different signals.
- the gradient of the change or other variables that can be derived from the course of time are taken into account in the assignment.
- a weighting factor for a signal can be used when assigning the characteristic numerical values.
- the weighting factors can preferably be generated from a process model that takes into account the correlations between different signals.
- Each characteristic numerical value represents the change in a specific signal and is also uniquely assigned to it.
- the series of characteristic numerical values each describe the state change of the production system before a disruption.
- the rows can be some kind of vectors or ordered lists or matrices.
- Various match criteria can be used to determine the essential match of the characteristic numerical values in rows (R1...Rn).
- mathematical methods for calculating the distance between the rows are used for this purpose. Such mathematical methods are well known. If the distance is small, i.e. if there is a substantial match, the rows are assigned to the same category. Similar changes in certain signals and/or changes in the same signals thus result in rows being combined into one category. This means that faults are grouped together in which similar status changes occurred before the fault.
- the similarity between rows of characteristic numerical values can preferably be determined using methods of artificial intelligence for pattern recognition or for segmentation.
- the number of categories into which the segmentation is to be made can be specified.
- the criterion for the match can be adjusted in such a way that the desired number of categories is created.
- preferably only part of the characteristic numerical values i.e. only part of the underlying signals, can be used when combining the characteristic numerical values into rows and/or when determining the similarity between rows.
- Which characteristic numerical values and thus which signals are used can be specified via a process model, as described above.
- one or more measures are preferably assigned to each category (K1 ... Km) of production disruption and stored, with the measures each representing a change in the production settings such that a reduction in the corresponding risk of disruption in the category is to be expected.
- measures each representing a change in the production settings such that a reduction in the corresponding risk of disruption in the category is to be expected.
- Any display device or storage device can serve as the end device for the output.
- the output can also take place in the DCS and/or in the QCS.
- the output can be displayed directly or stored for later display or use. It is advantageous if the proposed measures for reducing the risk of occurrence associated with a particular category are also output.
- the formation of the predictive models takes place with mathematical methods.
- the target value for these models is a so-called signifier.
- the signifier represents a value that describes the risk of disruption in the corresponding category.
- the signifier can be a single numerical value and in particular can be between 0 and 1. The closer it comes to 1, the more likely it is that a production disruption will occur, in particular a web break, in this category.
- the predictive models are preferably created using machine learning methods, in particular neural networks.
- quality deviations i.e. that certain quality parameters are outside of the specification.
- quality deviations can include basis weight fluctuations, moisture fluctuations, optical or paper-related quality parameters.
- proper production is interrupted if the relevant quality parameters are outside the permissible limits for the product.
- the method for avoiding production disruptions according to claim 8 is characterized in that the following steps are carried out:
- An essential advantage of the implementation of the method according to the invention is that the current, ongoing production is monitored and imminent production disruptions are recognized in good time and clearly assigned to a category of production disruptions. This is possible through the category-specific output of the respective risk of disruption.
- the current signal values mean the data from the current production. These are not only the actual values at the current time, but also the signal values from production that was a little earlier, i.e. from a period of a few hours or a few days before the current time. This is the only way to record the signal values as time series and thus evaluate changes in the system's status.
- a process model for reducing the number of required signals can preferably be used for the provision of the second volume of data.
- the second data volume particularly preferably contains the signal values of the signals that are used at least in a predictive model of one of the categories formed.
- the evaluation of the second data volume using the respective predictive models can in turn be carried out using machine learning methods.
- the target value of the evaluations is a signifier for each category.
- the signifier indicates the current risk of disruption for a certain category, based on the second data set with signal values from the current production. The higher the corresponding current disruption risk, the more likely it is that a production disruption will occur in the respective category.
- the signifier is exactly one value.
- the signifiers are advantageously normalized in such a way that they assume values between 0 and 1. The closer the signifier is to 1, the higher the risk of interference in this category.
- a threshold value can be defined for each signifier. Different threshold values can be defined for different categories. If the signifier has reached or exceeded the threshold value of the associated category, the signifier is output. This indicates an increased risk of interference in the corresponding category. All signifiers can also be output, and, for example, the signifiers that have reached the threshold value can be highlighted. Furthermore, the development of the signifiers over time can be output, so the development of the disruption risks for the various categories of the current production can be observed.
- steps 8.2 and 8.3 not only the time series of the signal values but also signal changes in a time interval and/or the gradient and/or other derived variables of the signal value time series are used. It is also advantageous if, in step 8.4, additional measures that are assigned to the category for which a signifier has reached or exceeded the threshold value are output.
- the additional advantage is that the operator receives suitable suggestions for changing the current production directly for the respective situation.
- the current risk of disruption and suitable remedial measures are given separately for each category of production disruption.
- step 8.1 In order to improve the method and to be able to adapt it to changed production conditions, it is advantageous if one or more repetitions of step 8.1 are provided in order to update the categories and/or the predictive models and/or the measures after a certain production time.
- This computer unit can be an independent unit or a separate computer, it can also be part of a process control system (DCS) or a quality control system (QCS).
- DCS process control system
- QCS quality control system
- the paper machine 1 is a machine for the production and/or processing of fibrous webs, in particular a machine for the production of paper or cardboard webs, for example a so-called cardboard machine, or a pulp machine, a coating machine or a paper processing machine.
- the further processing 3 can be a reeling, a re-reeling or a slitting machine or a packaging machine. In particular, it can also be a coating machine or paper processing machine if the paper machine is a machine for producing paper or cardboard webs.
- Stock preparation 2, paper machine 1 and further processing 3 can also consist of several subsystems.
- the production system 10 includes the ancillary units 4.5 and possibly laboratories, as well as at least the process control system (DCS) 20 and at least the quality control system (QCS) 21.
- DCS process control system
- QCS quality control system
- the various signals Sx of the production system 10 are received, processed and regulated in the process control system (DCS) 20 and in the quality control system (QCS) 21 in order to control the production system as is known in the prior art.
- DCS process control system
- QCS quality control system
- the first data volume D1 comprising signal values of different signals Sx as time series Sx(t), is transmitted to a computer unit 22. If necessary, these signal values can be stored beforehand on a data memory 23 and can be retrieved as required.
- the first data set D1 includes historical signal values that describe a period of time in which at least a number of production disruptions occurred. Signal values from a period of several weeks or several months are usually used.
- the computer unit 22 can be a separate computer or part of the DCS or the QCS.
- the process for categorizing production disturbances is carried out on the computer unit 22 and the results are output.
- Results here are in particular the determined categories K1 . . . Km and/or the predictive models P1 . . . Pm formed for the various categories.
- the output can be done on different devices.
- the output A1 can be transmitted to the DCS 20 and/or the output A2 to the QCS and/or the output A3 to the terminal 24.
- the terminal 24 can be a screen or another computer, for example a tablet.
- a further output A4 to the QCS or also an output to the DCS can take place from the terminal 24 .
- a second data volume D2 with current signal values of the signals Sx is transmitted to the computer unit 22 as a time series Sx(t).
- the computer unit 22 applies the predictive models P1 . . . Pm and evaluates the second data set D2 accordingly.
- the results of the method are again output as A1, A2, A3 and/or A4 as previously described.
- the results of this procedure are the signifiers XI ...Xm, each of which indicates the category-specific risk of disruption.
- the sequence of a method for categorization according to the invention is shown schematically in FIGS. 2a and 2b.
- 2a describes the formation of the rows R1...Rn from the characteristic numerical values ZW1...ZWq.
- the first data volume D1 includes the signal values of numerous signals Sx of the production system 10 as time series Sx(t).
- Various disruption times t1, t2, t3, . . . tn are selected and/or determined automatically.
- the signal changes ASx(tn) that occurred in a time interval before the fault time tn are determined for the different signals. Different time intervals can also be used for different signals.
- shorter time intervals can be relevant for drive data, while time intervals of several hours may have an influence for raw material data, and several days can even be used for clothing data, such as felt age.
- a selection can be made of signals Sx that are actually provided in the first data set D1 and/or for which the signal changes ASx(tn) are determined.
- Characteristic numerical values ZW are assigned to the signal changes ASx(tn) and combined to form rows, in particular vectors.
- Each numerical value ZW1...ZWq belongs to a specific signal S1...Sx.
- each row belongs to a certain disturbance time t1...tn.
- the series R1 formed here from ZW2, ZW1, ZW3, ZWx..., for the disruption time t2, the series R2 and so on up to the series Rn for the disruption time tn.
- the rows can be formed as some kind of vectors or ordered rows or matrices.
- the change in the signal value in a time interval is used to determine the respective characteristic numerical value ZWx for a signal Sx. Not only the absolute change, but also the dynamics of the change, the gradient and other mathematical characteristics of the time course are taken into account. In addition, a weighting factor can also be used for different signals. For example, already known relationships can be taken into account.
- 2b describes the segmentation of the rows into different categories and the formation of the respective predictive models.
- the rows R1 ... Rn are examined for similarity.
- the series are assumed to be similar if there is a substantial agreement between the characteristic numerical values ZW1 . . . ZWq contained therein.
- This essential agreement can be achieved in particular by mathematically determining the distance. If the distance is less than a specified value, the rows become the same assigned category. The distance is determined using known mathematical methods.
- mathematical methods for pattern recognition or segmentation can be used for segmenting the rows into different categories K1 . . . Km.
- Rows R1...Rn recognized as similar are assigned to the same category K1...Km.
- the number of categories K1 . . . Km can be predetermined. However, it can also be created accordingly by the evaluation algorithm. Rows that are not similar to other rows are assigned a remaining quantity L and are not further considered for the evaluation. This can be, for example, production disruptions caused by external influences that cannot be detected in the signal values. For example, when impurities come loose on the paper machine and a chunk falls into the paper web, causing a tear.
- a predictive model P1 . . . Pm is then formed within each category by evaluating the rows assigned to the category. Artificial intelligence methods, in particular for machine learning, such as neural networks, can be used for this purpose.
- a separate predictive model P1 ... Pm is created for each category K1 ... Km, which is a major advantage of the method and offers greater reliability.
- the predictive models P1 . . . Pm are designed in such a way that they can each output a signifier X1 . So how high is the risk for the second data volume D2 that a production disruption of the respective category will occur.
- the segmentation can then distinguish web breaks that are announced by fluctuations in the stock preparation or by noticeable changes in the state of the drive data of the paper machine or by changes in the moisture cross profile or by other special abnormalities in the data.
- the schematic representation in FIG. 3 shows the course of a method for avoiding production disruptions.
- the second data set D2 includes numerous signal values as time series Sx(t) from the current production operation. If necessary, the number of included signals can be reduced by using a process model to select only the relevant signals for the types of production disturbance that are considered.
- Each predictive model P1...Px supplies a signifier X1...Xm, which describes the risk of failure of the respective category.
- the signifier is a single value and is preferably between 0 and 1. The closer it is to 1, the higher the risk that a production disruption of this category will occur.
- other standardizations for the signifier are also conceivable.
- the particular advantage of the method according to the invention is, among other things, that a separate signifier is output for each category. This allows very good monitoring of the production process and better conclusions in the event of imminent faults.
- Each signifier X1...Xm is now compared with the associated threshold value G1...Gm. If the threshold value G1 . . . Gm is reached or exceeded, there is a high risk that the corresponding production disruption can occur. At least the signifiers that exceed the respective threshold value are output. All signifiers X1...Xm can also be output, and those that have reached the threshold value are specially marked or highlighted.
- One or more measures to reduce the risk of disruption can be assigned to each category of production disruption.
- Measures M1 and M2 are assigned to category K1 and measure M2 to category K2.
- these are proposed changes to production settings. For example, the web tension can be increased if a category with fluttering web edges has been formed. Or a change in the drainage settings in the forming section can be assigned if a category with poorer drainage of the stock has been created.
- the measures M1 . . . Mq that are assigned to the respective category are preferably also output, at least for the signifiers that have reached the threshold value. In this way, the operator or the process control system can initiate category-specific remedial measures in a timely and targeted manner and thus more reliably avoid impending production disruptions.
- DCS process control system
- QCS Quality Control System
Landscapes
- General Factory Administration (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
Description
Claims
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| DE102020126072.5A DE102020126072A1 (de) | 2020-10-06 | 2020-10-06 | Verfahren zur Kategorisierung und Verfahren zur Vermeidung von Störungen an einer Papiermaschine, insbesondere von Bahnabrissen |
| PCT/EP2021/074080 WO2022073699A1 (de) | 2020-10-06 | 2021-09-01 | Verfahren zur kategorisierung und verfahren zur vermeidung von störungen an einer papiermaschine, insbesondere von bahnabrissen |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| EP4225992A1 true EP4225992A1 (de) | 2023-08-16 |
Family
ID=77821722
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| EP21769321.7A Pending EP4225992A1 (de) | 2020-10-06 | 2021-09-01 | Verfahren zur kategorisierung und verfahren zur vermeidung von störungen an einer papiermaschine, insbesondere von bahnabrissen |
Country Status (3)
| Country | Link |
|---|---|
| EP (1) | EP4225992A1 (de) |
| DE (1) | DE102020126072A1 (de) |
| WO (1) | WO2022073699A1 (de) |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6405140B1 (en) | 1999-09-15 | 2002-06-11 | General Electric Company | System and method for paper web time-break prediction |
| US6542852B2 (en) | 1999-09-15 | 2003-04-01 | General Electric Company | System and method for paper web time-to-break prediction |
| US6519534B2 (en) | 2000-05-30 | 2003-02-11 | General Electric Company | Paper web breakage prediction using bootstrap aggregation of classification and regression trees |
| JP4560772B2 (ja) * | 2004-04-13 | 2010-10-13 | 王子製紙株式会社 | 抄紙機紙切れ予測装置、抄紙機紙切れ予測方法およびコンピュータプログラム |
| EP3699356A1 (de) | 2019-02-25 | 2020-08-26 | Siemens Aktiengesellschaft | Verfahren und vorrichtung zur erkennung eines bahnabrisses einer faserbahn, industrielle anlage sowie computerprogrammprodukt |
-
2020
- 2020-10-06 DE DE102020126072.5A patent/DE102020126072A1/de active Pending
-
2021
- 2021-09-01 EP EP21769321.7A patent/EP4225992A1/de active Pending
- 2021-09-01 WO PCT/EP2021/074080 patent/WO2022073699A1/de not_active Ceased
Also Published As
| Publication number | Publication date |
|---|---|
| WO2022073699A1 (de) | 2022-04-14 |
| DE102020126072A1 (de) | 2022-04-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3671632B1 (de) | Bildbasierte wartungsvorhersage und detektion von fehlbedienungen | |
| EP3737943B1 (de) | Verfahren und vorrichtung zur überwachung eines texturierprozesses | |
| EP4165484B1 (de) | Überwachungsvorrichtung und verfahren zur anomaliedetektion | |
| EP3634896B1 (de) | Verfahren und vorrichtung zur überwachung einer fadenspannung an einem laufenden faden | |
| EP3876060B1 (de) | Verfahren und recheneinheit zur ursachenanalyse eines anomalen zustandes einer maschine | |
| EP3672895B1 (de) | Verfahren und vorrichtung zum texturieren eines synthetischen fadens | |
| EP3282399A1 (de) | Verfahren zur verbesserten erkennung von prozessanomalien einer technischen anlage sowie entsprechendes diagnosesystem | |
| EP3825796A1 (de) | Verfahren und vorrichtung zum ki-basierten betreiben eines automatisierungssystems | |
| DE102018002781B4 (de) | Schaltungskonfigurations-Optimierungsvorrichtung und maschinelle Lernvorrichtung | |
| EP4295202B1 (de) | Verfahren und system zur vermessung von bauteilen sowie programm | |
| EP3796117B1 (de) | Diagnoseverfahren und diagnosesystem für eine verfahrenstechnische anlage | |
| EP4058857A1 (de) | Steuern und/oder überwachen einer maschinenanordnung | |
| EP4225992A1 (de) | Verfahren zur kategorisierung und verfahren zur vermeidung von störungen an einer papiermaschine, insbesondere von bahnabrissen | |
| AT519777B1 (de) | Verfahren zur Erkennung des normalen Betriebszustands eines Arbeitsprozesses | |
| DE102020004467A1 (de) | Verfahren und Vorrichtung zur Überwachung einer Maschinenanlage zur Herstellung oder Behandlung von synthetischen Fasern | |
| DE102021002646A1 (de) | Verfahren zur Überwachung einer Vielzahl von Bearbeitungsstellen für synthetische Fäden | |
| DE102022121211B4 (de) | Verfahren zum Betreiben einer Sensor-Anordnung und Sensor-Anordnung sowie Einrichtung zur Datenverarbeitung und Vorrichtung | |
| EP3851181A1 (de) | Verfahren zum überwachen einer membrantechnischen vorrichtung | |
| EP4115009B1 (de) | Verfahren und vorrichtung zur überwachung einer maschinenanlage zur herstellung oder behandlung von synthetischen fasern | |
| DE102005038889A1 (de) | Verfahren zur Prüfung eines Fertigungsgutes | |
| EP3699701A1 (de) | Verfahren und vorrichtung zur bestimmung oder vorhersage einer position eines bahnabrisses, computerprogramm und industrielle anlage | |
| DE102021116520A1 (de) | Verfahren zur Qualitätsüberwachung von elektrischen Antrieben | |
| WO2025242390A1 (de) | Verfahren zum erzeugen eines fehlerrepräsentierenden signals frs für eine industrieanlage | |
| DE102021134315A1 (de) | Verfahren zum Trainieren einer künstlichen Intelligenz sowie ein Feldgerätesystem und ein Verfahren zur Detektion von zumindest einem Feldgerät aus zumindest einem Cluster von Feldgeräten | |
| DE102022207447A1 (de) | Verfahren und Vorrichtung zur Zustandsüberwachung von Wickelmaschinen |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
| PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
| 17P | Request for examination filed |
Effective date: 20230508 |
|
| AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |
|
| DAV | Request for validation of the european patent (deleted) | ||
| DAX | Request for extension of the european patent (deleted) | ||
| GRAP | Despatch of communication of intention to grant a patent |
Free format text: ORIGINAL CODE: EPIDOSNIGR1 |
|
| STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: GRANT OF PATENT IS INTENDED |