FI130085B - Monitoring method and apparatus - Google Patents

Abstract

A monitoring method of a paper or board machine, which predicts a web break. Controlled variables are received from a plurality of process phases of the paper or board machine (20). Each process state is determined as a function of time. Controlled variables relating to a process state are compared with corresponding reference variables repeatedly for each process state. Deviations between them are distributed in at least three types with respect to at least two limits using absolute values in the following manner. A first absolute value is associated with a deviation, if the deviation is equal to or larger than a first limit. A second absolute value, which is different from or the same as the first absolute value, is associated with the deviation, if the deviation is equal to or smaller than a second limit, which is smaller than the first limit. Or alternatively, a third absolute value, which is different from the first and second absolute values, is associated with the deviation, if the deviation is between the first limit and the second limit. For each controlled variable compared with a corresponding reference variable an anomaly index as a moving sum of the first, second and third absolute values is formed within a time window. If the moving sum of any of the anomaly indices of the controlled variables is at or higher than an alarm threshold, alarm of a possibility of a consequent web break, and present of at least one of the controlled variables prior to and during a crossing over the alarm threshold in a time window indicating the possibility of the consequent web break.

Description

Monitoring method and apparatus

Field

The invention relates to a monitoring method and apparatus for a paper or board machine.

Background

Web breaks lead to considerable losses in paper or board machines. In the past, several methods, both for predicting occurrence and finding reasons for the web breaks, have been intensively studied. Measurements utilizing a variety of sensors and cameras have been used to detect the web breaks as they develop and arise, thus enabling an observation of an occurrence of a web break and a subsequent analysis thereof for an attempt to predict a new break. However, the analysis of the root causes of a web break has remained vague, and little is known why a web break happens, and hence practically nothing can be done prior to a new web break when it is going to happen.

Patent document US6405140 presents a system and a method for paper web time-break prediction. Patent document US20060070713 presents an apparatus and a method for monitoring the transfer of a material web. Patent document W020070964 presents a method for monitoring a rapidly-moving paper web and a corresponding system.

N All in all, a web break has remained unpredictable in real life despite all

N efforts. Hence, an improvement would be welcome.

N o Brief description z 25 a © The present invention seeks to provide an improvement to the web

N UN

O monitoring.

N The invention is defined by the independent claims. Embodiments are

N defined in the dependent claims.

List of drawings

Example embodiments of the present invention are described below, by way of example only, with reference to the accompanying drawings, in which

Figure 1 illustrates an example of a paper machine,

Figure 2A illustrates an example of process states, limits, variables following the limits and crossings of limits;

Figure 2B illustrates an example of process states, limits, variables following the limits and crossings of limits in more detail;

Figure 3 illustrates an example of limits;

Figure 4 illustrates an example of variables with a sudden anomaly;

Figure 5 illustrates an example of values on map as a function of a plurality of quantities;

Figure 6 shows coefficient of variation;

Figure 7 illustrates an example situation where several process states have been learned for future monitoring and control;

Figure 8 illustrates an example of the data processing apparatus; and

Figure 9 illustrates of an example of a flow chart of a monitoring method.

Description of embodiments

N The following embodiments are only examples. Although the

O specification may refer to “an” embodiment in several locations, this does not

N necessarily mean that each such reference is to the same embodiment(s), or that o 25 the feature only applies to a single embodiment. Single features of different

E embodiments may also be combined to provide other embodiments. Furthermore, © words "comprising" and "including" should be understood as not limiting the

D. described embodiments to consist of only those features that have been mentioned 3 and such embodiments may also contain features/structures that have not been specifically mentioned. All combinations of the embodiments are considered possible if their combination does not lead to structural or logical contradiction.

It should be noted that while Figures illustrate various embodiments, they are simplified diagrams that only show some structures and/or functional entities. The connections shown in the Figures may refer to logical or physical connections. It is apparent to a person skilled in the art that the described apparatus may also comprise other functions and structures than those described in Figures and text. It should be appreciated that details of some functions, structures, and the signalling used for monitoring and measuring and/or controlling are irrelevant to the actual invention. Therefore, they need not be discussed in more detail here.

This document, combined with papermaking expertise, teaches how to monitor and also prevent web breaks efficiently as explained below. By comparing recorded data of the web from at least one past running period and a live monitoring together it is possible detect trend of time dependent variables of the controlled variables and the manipulated variables. The recorded data indicates signals from previous web breaks, which can be used for developing the variables and the process. The results may be improved by keeping the operators in the development loop as well. After the machine learning from or otherwise analysing the recorded data, for example, it is possible to predict a next approaching web break with the live application, which identifies the signals that can cause web breaks. The presented method reveals process anomalies and thus helps to avoid unwanted incidents.

A process state may depend on sheet break log(s), levels, pressures

N and/or edge flows relating to headbox, stock, speed, web width, basis weight,

N suction width and/or suction box positions in on or more couch rolls, moisture of 7 25 web, consistency of suspension, grade, at least one nip pressure, variation in the at > least one nip pressure, at least one filler, at least one retention agent, at least one i process chemical, steam pressure, web monitoring system and/or coating paste or & the like without limiting to these. = Figure 1 illustrates an example of a paper machine that can be examined first. One or more stocks, which may include ash, cellulose and/or moisture, are fed onto a paper machine through a wire pit silo 100, which is usually preceded by a blending chest 132 for partial stocks and a machine chest 134. The machine stock is dispensed for a short circulation, for instance, controlled by a basis weight control, a grade change program or in general past, present and/or future process state(s). The blending chest 132 and the machine chest 134 may also be replaced by a separate mixing reactor (not shown in Figure 1), and the dispensing of the machine stock is controlled by feeding each partial stock separately by means of valves or another flow control means 130. In the wire pit silo 100, water is mixed into the machine stock to obtain a desired consistency for the short circulation (dashed line from a former 110 to the wire pit silo 100).

From the obtained stock it is possible to remove sand or the like (centrifugal cleaners), air (deculator) and other coarse material (pressure filter) using cleaning devices 102, and the stock is pumped with a pump 104 to a headbox 106. The sand or the like that avoids removal may form a part of ash in paper or board. Before the headbox 106, it is possible to add to the stock, in a desired manner, a filler TA, including e.g. gypsum, kaolin, calcium carbonate, talcum, chalk, titanium dioxide and diatomite etc. and/or a retention agent RA, such as inorganic, inartificial organic or synthetic water-soluble organic polymers. The filler TA and/or the retention agent RA may include ash component of the paper or board.

There may be sensors 116 in conjunction with different actuators making it then possible to measure the web or a quantity of the blending chest 132, machine stock, machine chest 134, mixing reactor, pit silo 100, cleaning devices 102, pump 104, headbox 106, for example.

N From the headbox 106 the stock is fed through a slice opening 108 of

N the headbox to a former 110, which may be a fourdrinier wire or a gap former. The 7 25 slice opening may be measured. In the former 110, water drains out of the web 10 > and additionally ash, fines and fibres are led to the short circulation. In the former

E 110, the stock is fed as a web 10 onto a wire, and the web 10 is preliminary dried & and pressed in a press 112, where a nip pressure may be measured. The web 10 is = actually dried in driers 114. In general, there is a plurality of sensors 116 that

N 30 performsavariety of measurements such as nip pressure, temperature, brightness, basis weight, moisture, dry stuff content of the web, stock types, filler, retention agent. The measurements may be performed before, at or after an actuator, and the measurements may be directed to the web or the actuator. A person skilled in the art is familiar with the measurements of the paper or board machine 20. The sensors 116 feed the measured data to a data processing apparatus 128 (see 5 continuous lines therebetween). Additionally, certain parameters may be set and adjusted by the data processing apparatus 128. Still additionally, the data processing apparatus 128 may receive data such as a sheet break log manually input by an operator of the paper or board making machine.

The paper or board machine 20 may also include a pre-calender 140, a coating section 142 and/or a finishing calender 144. It is not necessary to have the coating section 142, however, and therefore it is not necessary to have more calenders 140, 144 than one. In the coating section 142, coating paste may be spread onto paper. The coating paste may include gypsum, kaolin, talcum or carbonate, starch, latex or the like, for example. Their amount may be measured, or dosed in a predetermined manner controlled by the data processing apparatus.

In calenders 140, 144, where the uncoated or coated paper or board web runs between the rolls pressing with desired force, which may be set, it is possible to change the surface properties of the paper, such as smoothness, roughness, topography, gloss and the like, which may be measured. The calender 140, 144 may also affect the paper thickness and/or the basis weight or other mass per unit area of the paper or board, for example. These may be measured for the monitoring apparatus. In the calender 140, 144, the properties of the paper web

N may be changed by means of web moistening, temperature and nip pressure

N between the rolls. They may be set by and/or measured for the monitoring 7 25 apparatus. In addition to this, it is clear that the operation of a paper machine is > known, per se, to a person skilled in the art, and therefore, it needs not be presented i in greater detail in this context. & Figure 1 also shows a control arrangement of the paper machine. Then = the data processing apparatus 128 may also perform the data processing for the

N 30 control of the paper or board machine 20. Factors affecting the quality and grade change include, inter alia, the number and mutual proportion of partial stocks, the amount of filler, the amount of retention agent, machine speed, the amount of white water and drying capacity. All of them may be available to the monitoring apparatus. The data processing apparatus 128 may control the dispensing of partial stocks by means of valves 130, the dispensing of each filler TA by means of the valve 138, the dispensing of the retention agent RA by means of the valve 136, adjust the size of the slice opening 108, control the machine speed, control the amount of white water and the drying process in block 114 (see dashed lines). The data processing apparatus 128 utilizes the sensors 116 for measuring the web 10.

The data processing apparatus 128 may also measure the properties of the web 10 elsewhere (e.g. at the same locations where controls are performed). Additionally, the data processing apparatus 128 may measure the actuators of the paper or board machine 20.

The data processing apparatus 128 may be conceived as a paper machine's control arrangement, or part thereof, based on automatic data processing. The data processing apparatus 128 may receive digital signals or convert the received analog signals to digital ones. The data processing apparatus 128 may comprise at least one processor and at least one memory and execute the signal processing in accordance with an appropriate computer program. The operating principle of the data processing apparatus 128 may be, for instance, PID (Proportional-Integral-Derivative), MPC (Model Predictive Control) or GPC (General Predictive Control) control. In an embodiment, the data processing apparatus 128 is at least partly located in the cloud.

N The measurements of one or more paper or board machines generate a

N lot of data i.e. big data and the big data can be transmitted in a wired or wireless 7 25 manner to the data processing apparatus 128, which may perform machine > learning and/or artificial intelligence to the big data in order to analyse it. The big i data may be analysed in the cloud. Hence, the data processing apparatus 128 may & comprise one or more data processing units which may be located at the same site = as the paper or board machine 20 or at one or more separate sites from the paper

N 30 or board machine 20. Additionally, it is possible that one or a larger part of a plurality of data processing units of the data processing apparatus 128 is located at the site of the paper or board machine 20 while other data processing units of the data processing apparatus 128 are at separate sites from the paper or board machine 20.

A monitoring apparatus comprises or is comprised by the data processing apparatus 128 and a user interface 150. The monitoring apparatus may also comprise the sensors 116 or the monitoring apparatus may merely receive the measured data from the sensors 116. The monitoring apparatus can predict a web break in a manner explained next.

The monitoring apparatus determines a process state of a paper or board making machine as a function of time. Based on measurement with a variety of sensors such as the sensors 116 and the manipulated variables the monitoring apparatus may detect or determine the process state. The system may then be taught what at least one stable state of the process is based on data on recorded process history. Typically, there are a plurality of stable states. During operation the paper or board making machine 20 goes through a continuum of process states as a function time. Hence, the processes of the paper or board machine 20 may occupy a variety of states such as speed, grade, caliper, production, headbox pressure or the like.

The monitoring apparatus receives the plurality of controlled variables from a plurality of process phases of the paper or board making machine from the sensors 116. The controlled variables are the process outputs which the data processing apparatus 128 receives. Manipulated variables, in turn, are input

N directly or indirectly to the process by a controller which is shown for simplicity to

N be included in the data processing apparatus 128 in this document. = 25 In an embodiment, when a certain amount of data, which may be a > predetermined amount of data or an amount of data that can be considered a i sufficient amount of data or sufficiently representative data, has been collected & from the paper or board machine 20, the data may be filtered by removing data = that is unassociated with a web break. Such data that does not relate or is not

N 30 considered to relate to a web break may be removed from the data. Also data that keeps constant may be removed. The removal of the data unassociated with a web break may be performed automatically. Finally, data that is unlikely to be involved in or relate to a web break may be removed. This part of work may be manual or automatic. What remains is data that contains information on a web break with a high probability.

For calibration, a plurality of grades or as large variety of grades as possible may be run in the paper or board machine 20 for collecting the data. The data is then used for teaching the monitoring method and/or apparatus. If a grade that has not been taught to the monitoring method and/or apparatus is run in the paper or board machine 20, the limits and absolute values may be those of a closest grade. The teaching may utilize correlation of the state parameters (such as speed, caliber etc.) in relation to the input parameters i.e. manipulated variables.

Figure 2A illustrates an exemplary embodiment of controlled variables of a process state A, B, €, D and E or controlled variables relating to a process state

A, B, C, D and E. The monitoring apparatus compares the controlled variables with corresponding reference variables for each of at least one process state repeatedly.

Figure 2A also illustrates an exemplary embodiment of how the monitoring apparatus subseguently distribute deviations between the controlled variables and the reference variables in at least three types with respect to at least two limits using absolute values. The absolute value means a non-negative number i.e. the absolute value can be expressed using real or rational numbers having value that is zero or larger.

Figure 2B illustrates the same exemplary with in more detail. The states

N that result in a break free drive of the paper or board machine 20 may vary fairly

N frequently as states D, DA, DB, DC, DD and DE illustrate. The lengths of the lines of 7 25 the states have been drawn a little bit too long for making them easier to see. > The distribution of the deviations depends on the deviation, magnitude i of which may be measured as a quantity or a value. If a deviation of the said & deviations is equal to or larger than a first limit of a process state, the monitoring = apparatus associates at least one first absolute value with the deviation.

N 30 Alternatively, if a deviation of the said deviations is egual to or smaller than a second limit, which is smaller than the first limit, the monitoring apparatus associates at least one second absolute value, which is different from or the same as the at least one first absolute value, with the deviation.

Still alternatively, if a deviation of the said deviations is between the first limit and the second limit, the monitoring apparatus associates at least one third absolute value, which is different from any of the at least one first and second absolute values, with the deviation. In this manner, all deviations will be distributed effectively with respect to the limits that are based on an expected operation of the paper or board machine 20.

Therefore, the process state is first determined and in that process state deviations from references of said process state are determined and distributed.

The first and second limits of the process state C are marked in Figure 2. In other process states the limits are set similarly. Each process state may have unique limits like the examples of Figure 2A and 2B illustrate.

Based on these, the monitoring apparatus forms for each of the controlled variables compared with the corresponding reference variables an anomaly index as a moving sum of the first, second and third absolute values within a time window.

The moving sum is related to a moving average such that in the moving average the moving sum is divided by the number of absolute values of the sum. In the moving sum the division is not performed. The moving sum may also be called a rolling sum or a running sum. The moving sum continuously adds up a determined number of the absolute values which are processed in the summation

N according to a FIFO principle (FIFO = First In First Out) as new absolute values

N enters the summing operation. The summation may be performed utilizing a FIR- 7 25 filter (FIR = Finite Impulse Response). > The monitoring apparatus causes, if the moving sum of any of the i anomaly indices of the controlled variables is at or higher than an alarm threshold, & an alarm of a possibility of a consequent web break, the monitoring apparatus also = causes, based on the same conditions as for the alarm, presentation of at least one

N 30 of the controlled variables prior to and during crossing over the alarm threshold in a time window indicating the possible conseguent web break.

In an embodiment, the monitoring apparatus may present a warning, if the moving sum of any of the anomaly indices of the controlled variables is at or higher than a warning threshold. The warning threshold has a lower value than the alarm threshold. Additionally, the monitoring apparatus may present at least one of the controlled variables prior to and during crossing over the warning threshold in the at least one time window, if the moving sum of any of the anomaly indices of the controlled variables is at or higher than the warning threshold.

The alarm threshold may be based on learning of the previous web breaks. An initial setting of the alarm threshold may be a wild guess, based on simulation and/or experience of a user. The user may be a person or a machine.

In an embodiment an example of which is illustrated in Figure 3, the at least one third absolute value may have on more than one absolute value. Then absolute first middle value, which is different from the first and second absolute values, may be associated with a deviation of any of the controlled variables of the comparison, if the deviation is between the first limit and the first middle limit.

Alternatively, an absolute second middle value, which is different from the first and second values and the first absolute middle value, may be associated with a deviation of any of the controlled variables of the comparison, if the deviation is between the second limit and the second middle limit. A third absolute middle value may be associated with a deviation that remains at or between the first middle limit and the second middle limit.

In an embodiment, the monitoring apparatus may vary any or some of

N the absolute values as function of the deviation such that an absolute value of the

N absolute values may be made larger in response to the larger deviations. The 7 25 monitoring apparatus may make the limits lower in response to smaller deviations. > This latter feature can be applied when the deviations are between the first and

E second limit or between the middle limits, for example. In this manner, the & monitoring apparatus may react quicker to large deviations because the sum more = easily crosses the first limit with larger absolute values. On the other hand, if the

N 30 deviations are small, the smaller deviations may be monitored more accurately with lowered limits by reacting also to smaller deviations because the sum more easily crosses the first limit that is lowered.

In an embodiment, the monitoring apparatus may update the absolute values based on the controlled variables and/or manipulated variables of one or more web breaks. The updating may change the value of any absolute value. Also the limits may be updated. The updating may be similar to that explained above.

In an embodiment, the monitoring apparatus may perform the presentation in an order of the anomaly indices such that an anomaly index i.e. the sum of a highest value may be presented first. Correspondingly, the monitoring apparatus may perform the presentation in an order of the anomaly indices such that a controlled variable and/or the manipulated variable of a highest value may be presented first.

In an embodiment, the monitoring apparatus may associate an absolute value of the absolute values with a deviation of the deviations by weighting the deviation with the absolute value. In this manner, an absolute value of the absolute values is multiplied with a weight and the weighted deviation is utilized in the summing operation. The weighted deviation then represents the absolute value in the moving sum. By adjusting a weight, the weighted deviation will also be adjusted, and hence the adjustment of an absolute value means here the adjustment of a weight.

In an embodiment, the monitoring apparatus may associate an absolute value of the absolute values with a deviation of the deviations by setting the

N absolute value for the deviation. In an embodiment, the absolute value may be

N retrieved from a table, for example. In this manner, the absolute value represents 7 25 the deviation in the summing operation. > In an embodiment, the monitoring apparatus may vary the absolute i values as a function of freguency of occurrence of the deviations by setting an & absolute value for a deviation occurrence of which is the most frequent. If a first = deviation occurs more frequently than a second deviation, the first deviation may

N 30 thushavealarger absolute value than the second deviation.

In an embodiment, the monitoring apparatus may vary the absolute values as a function of user input subsequent to a web break. In this manner, the user may decide the absolute values for the deviations. In general, the absolute values depend on the process state, the controlled variable and the magnitude of the deviation. Hence, the user has a liberty to optimize each of the absolute values and also the limits.

Figure 4 illustrates an example of measurements. Curve 400 shows wetend starch flow, which abruptly rises at the latter part of the curve 400, which is a sign of anomaly. Curve 402 shows a difference of the wetend starch flow and the abruptrise can be seen as a peak synchronously with the abrupt rise. Curve 404 shows a deviation from the standard starch flow. Curve 406 shows a flatbox pump operation where a sudden change can be seen. The sudden change is an anomaly which also causes the abrupt change in the process. Together with the starch flow change this could cause a web brake. The monitoring apparatus can detect the anomalous change in the flatbox pump and alarm the operator in order to avoid a larger damage.

Figure 5 illustrates an example basis weight and its variation. The rectangle presents a sliding monitoring window e.g. one hour, from which a representative deviation such as a standard deviation, and an average or expected value such as a mean value may be calculated. The signal in Figure 6 shows coefficient of variation that comes by dividing the representative deviation with the expected value multiplied by a desired number such as one hundred. This

N coefficient of variation explains the variation of the measured process variable

N within certain timespan and is thus an efficient way to find major changes. Any 7 25 other functions that describe variance and deviation may also used. > Figure 7 illustrates an example map of the process states of the paper i or board machine 20 with respect to three controlled variables: speed, basis weight & (BW) and pressure (nip pressure), for example. The speed is measured in meters = per a minute (m/min), the basis weight is measured in g/m?, and the pressure is

N 30 measured in bars or kilopascals. The monitoring apparatus has been taught or has learned that the normal operation includes the states that are marked with ellipses.

There is also an area where there are a plurality of break-free states marked with arrows. The clusters of break-free states are so densely packed that they could not be separately marked. Thus, driving the paper and board machine in these states within the limits results in web break free operation, the limits being those at least two limits explained earlier in this document and shown in Figure 3.

The monitoring apparatus may receive measured data on one or more of the following: levels, pressures and/or edge flows relating to headbox, stock, speed, web width, basis weight, suction width and/or suction box positions in on or more couch rolls, moisture of web, consistency of suspension, grade, at least one nip pressure, variation in the at least one nip pressure, at least one filler, at least one retention agent, at least one process chemical, stock type, steam pressure (relating drying), web monitoring system of quality of web, coating paste or the like without limiting to these. That is, any property or feature that may affect a web break may be measured. The measured properties of paper or board or the paper or board machine 20 of the web are typically known, per se, to a person skilled in the art although new one(s) may be used as the technology is developing.

The web monitoring system may be based on a camera that forms repeatedly images of the moving web and the web inspection system utilizes image processing for detecting holes, dirt, intensity variation of the web and/or color variation of the web, certain variation of which may relate to a web break. For example, a large number of holes and/or dirt in the web cause the web break which can be learned from observing real operation of the paper or board machine 20.

N Too large intensity or color variation may also indicate that a web break may occur.

N The stock type may refer to recycled or virgin paper or the great variety of stock 7 25 types are known, per se, to a person skilled in the art. > The monitoring apparatus may determine a normal operation of the i paper or board machine 20 by excluding from the normal operation data relating & to at least one of the following: recovery time after the web break, oscillation of the = manipulated variables related to and/or caused by a grade change, unstability

N 30 before and leading to the web break and period of web break. This is a way to determine operation which does not lead to web break because by monitoring and controlling keeps the web in a stable condition and/or guides the web back to a reliable condition if signs of anomaly starts to appear.

In an embodiment an example of which is illustrated in Figure 8, the monitoring apparatus may comprise one or more processors 700 and one or more memories 702 including computer program code. The one or more memories 702 and the computer program code configured to, with the one or more processors 700, may cause apparatus to perform the method steps as explained above and shown in Figure 9.

Figure 8 is a flow chart of the monitoring method. In step 800, a plurality of controlled variables are received from a plurality of process phases of the paper or board making machine.

In step 802, each process state of at least one process state of a paper or board making machine is determined as a function of time.

In step 804 controlled variables relating to a process state are compared with corresponding reference variables repeatedly for each of the at least one process state.

In step 806, distributing deviations between the controlled variables and the references are distributed in at least three types with respect to at least two limits using absolute values such that: associating 806A at least one first absolute value with a deviation, if the deviation is equal to or larger than a first limit, associating 806B at least one second absolute value, which is

N different from or the same as the at least one first absolute value,

N with the deviation, if the deviation is egual to or smaller than a = 25 second limit, which is smaller than the first limit, or > associating 806C at least one third absolute value, which is i different from any of the at least one first and second absolute & values, with the deviation, if the deviation is between the first limit = and the second limit.

N

In step 808 forming an anomaly index is formed for each of the controlled variables compared with the corresponding reference variables as a moving sum of the first, second and third absolute values within a time window.

In step 810, an alarm of a possibility of a consequent web break is caused if the moving sum of any of the anomaly indices of the controlled variables is at or higher than the alarm threshold. Simultaneously, presentation of at least one of the controlled variables prior to and during crossing over the alarm threshold in a time window indicating the possibility of the consequent web break is caused.

The method shown in Figure 7 may be implemented as a logic circuit solution or computer program. The computer program may be placed on a computer program distribution means for the distribution thereof. The computer program distribution means is readable by a data processing device, and it encodes the computer program commands, carries out the measurements and monitors and optionally controls the processes on the basis of the measurements.

The computer program may be distributed using a distribution medium which may be any medium readable by the controller. The medium may be a program storage medium, a memory, a software distribution package, or a compressed software package. In some cases, the distribution may be performed using at least one of the following: a near field communication signal, a short distance signal, and a telecommunications signal.

It will be obvious to a person skilled in the art that, as technology

N advances, the inventive concept can be implemented in various ways. The

O

N invention and its embodiments are not limited to the example embodiments

N

N 25 described above but may vary within the scope of the claims. o

I a a ©

N

LO

LO

N

O

N

Claims (13)

Claims

1. A monitoring method of a paper or board making machine, characterized in that the monitoring method predicts a web break by: receiving (800) a plurality of controlled variables from a plurality of process phases of the paper or board making machine (20); determining (802) each process state of at least one process state of a paper or board making machine (20) as a function of time; comparing (804) controlled variables relating to a process state with corresponding reference variables repeatedly for each of the at least one process state; distributing (806) deviations between them in at least three types with respect to at least two limits using absolute values such that: associating (806A) at least one first absolute value with a deviation, if the deviation is equal to or larger than a first limit, associating (806B) at least one second absolute value, which is different from or the same as the at least one first absolute value, with the deviation, if the deviation is equal to or smaller than a second limit, which is smaller than the first limit, or associating (806C) at least one third absolute value, which is different from any of the at least one first and second absolute values, with the deviation, if the deviation is between the first limit N and the second limit; N forming (808) for each of the controlled variables compared with the 7 25 corresponding reference variables an anomaly index as a moving sum of the first, > second and third absolute values within a time window; i causing (810), if the moving sum of any of the anomaly indices of the & controlled variables is at or higher than an alarm threshold, an alarm of a = possibility of a consequent web break, and presentation of at least one of the N 30 controlled variables prior to and during a crossing over the alarm threshold in a time window indicating the possibility of the conseguent web break.

2. The method of claim 1, characterized by presenting, if the moving sum of any of the anomaly indices of the controlled variables is at or higher than a warning threshold that has a lower value than the alarm threshold, a warning and at least one of the controlled variables prior to and during a crossing over the warning threshold in at least one time window.

3. The method of claim 1, characterized by varying the absolute value as function of the deviation such that the absolute values are made larger in response to the larger deviations

4. The method of claim 1, characterized by updating the absolute values based on the controlled variables and/or manipulated variables of one or more web breaks.

5. The method of claim 1 or 2, characterized by performing the presentation in an order of the anomaly indices such that an anomaly index of a highest value, manipulated variable and/or a controlled variable of a highest value is presented first.

6. The method of claim 1, characterized by associating an absolute value of the absolute values with a deviation of the deviations by weighting the deviation with the absolute value.

7. The method of claim 1, characterized by associating an N 20 absolute value of the absolute values with a deviation of the deviations by setting N the absolute value for the deviation. N o

8. The method of claim 1, 6 or 7, characterized by varying the I absolute values as a function of frequency of occurrence of the deviations by setting c a highest absolute value to a deviation the occurrence of which is the most freguent. N LO = 25

9. The method of claim 1, 6 or 7, characterized by varying the NN absolute values as a function of user input subsequent to a web break.

10. The method of claim 1, characterized in that the process state depends on levels, pressures and/or edge flows relating to headbox, stock, speed, web width, basis weight, suction width and/or suction box positions in on or more couch rolls, gloss, moisture of web, consistency of suspension, grade, at least one nip pressure, variation in the at least one nip pressure, at least one filler, at least one retention agent, at least one process chemical, steam pressure, web monitoring system and/or coating paste.

11. The method of claim 1, characterized in that determininga normal operation of the paper or board machine (20) by excluding from the normal operation data relating to at least one of the following: recovery time after the web break, oscillation of the manipulated variables related to and/or caused by a grade change, unstability before and leading to the web break and period of web break.

12. A monitoring apparatus, characterized in thatthe apparatus comprises one or more processors (700); and one or more memories (702) including computer program code; the one or more memories (702) and the computer program code configured to, with the one or more processors (702), cause apparatus at least to: perform the method steps of any one of the claims 1 to 11.

13. A monitoring apparatus for a paper or board making machine, characterized in that the monitoring apparatus is configured to predict a N web break by being configured to: N receive a plurality of controlled variables from a plurality of process a phases of the paper or board making machine (20); > 25 determine a process state of a paper or board making machine (20) as & a function of time; & compare controlled variables of a process state with corresponding = reference variables for each of at least one process state repeatedly, and distribute N deviations between them in atleast three types with respect to at least two limits using absolute values such that if a deviation of the deviations is:

equal to or larger than a first limit, associate at least one first absolute value with the deviation,

equal to or smaller than a second limit, which is smaller than the first limit, associate at least one second absolute value, which is different from or the same as the at least one first absolute value,

with the deviation, or between the first limit and the second limit, associate at least one third absolute value, which is different from any of the at least one first and second absolute values, with the deviation;

form for each of the controlled variables compared with the corresponding reference variables an anomaly index as a moving sum of the first, second and third absolute values within a time window;

cause, if the moving sum of any of the anomaly indices of the controlled variables is at or higher than an alarm threshold that has a higher value than the warning threshold, an alarm of a possibility of a conseguent web break, and presentation of at least one of the controlled variables prior to and during a crossing over the alarm threshold in a time window indicating the possibility of the conseguent web break.

N N O N N o I a a © N LO LO N O N