EP4361328A1 - Method for operating a carding machine, carding machine and preparation device for spinning - Google Patents

Abstract

The invention relates to a method for operating a card in a spinning preparation plant, in which fiber flocks are broken down to the individual fiber between a garnished rotating drum (4) and fixed and rotating carding elements and the resulting fiber web is transferred from the drum (4) to a doffer (5) and subsequently formed into a fiber band, having a card control with an operating unit (18), and that after an operator has entered a raw material and a production quantity, the card control determines the reference drum speed (n_R).The invention is characterized in that the operator starts an optimization program that enables a selection between a neps and/or energy-optimized mode of operation, wherein the card control carries out an automatic series of measurements in which several sensors (30) detect the number of neps in the fiber web at different drum speeds over a respective predetermined period of time (T) and at the same time the drive power of the card is determined, wherein the data from the sensors (30) and the determined Drive power is transmitted to a higher-level control system (43) of the spinning preparation, which, on the basis of at least these data and by means of a mathematical algorithm, suggests to the operator an operating mode of the card in different quality categories.

Description

Description

The present invention relates to a method for operating a carding machine, a carding machine and a spinning preparation plant according to the preamble of the independent patent claims.

When producing a yarn, the number of neps in the fiber sliver of the spinning preparation is a decisive quality criterion. A nep is understood to be a collection of knotted fibers that arises during the cotton harvest, the ginning process and the mechanical cotton processing. A distinction is made between husk neps and fiber neps. Husk neps arise when a piece of a seed husk is connected to the fibers. Fiber neps arise from a collection of short fibers that contract to form a knot or compaction after the carding process in the draw frame or flyer. Neps lead to an uneven color distribution in the yarn in the subsequent spinning process, which is undesirable. Furthermore, fiber knots also lead to a yarn of inferior quality, in which individual defects have to be removed. In the following, the term neps is used to summarize husk neps, short fiber collections and fiber knots.

According to the state of the art, it is known to change the settings of the carding machine in such a way that the number of neps is minimized. However, the settings also vary depending on the raw material, so the operator must be very experienced in order to achieve an optimal result.

The EP0409772A1 discloses a method for determining the number of neps at the exit of the card, whereby the pile of the fibers is measured at the same time. If the measured values deviate from the specified target values, an attempt is first made to improve the corresponding values by resetting the fine cleaning machine. If this is not successful, it is necessary to change the mixing ratios, which must be done by controlling the bale removal machine and ultimately also has an impact on the bale storage. In this method, the individual variables of nep number and pile are measured, and a new card setting is made that is individually assigned to the individual variables of nep number and pile. A certain card setting can either reduce the number of neps or change the pile by using a different card setting. It can happen that a change in the control variable, e.g. a card setting (carding speed, carding gap), a significant reduction in the number of neps occurs, but at the same time a significant negative change in the pile occurs, or vice versa. In addition, the known method assumes that an improvement in the two individual measured variables, namely the reduction in the number of neps and an improvement in the pile, which cannot be achieved on the card itself, but rather a change in the mixing ratios must be made.

The DE 19651893 B4 provides for the measured values for the number of neps and the measured values for the fiber length or shortening to be linked with one another, so that, in contrast to the known method, the greatest possible reduction in the number of neps and at the same time the least possible fiber damage (fiber shortening) is achieved. The measured values for the number of neps and the fiber lengths are combined and used for control intervention. In this way, optimization is achieved in a particularly advantageous way. The disadvantage is that a partial amount of fibers has to be manually removed from the carded fiber sliver, which makes the process complex. By the time the partial amount has been analyzed and the carding machine settings have been changed, too much fiber material, possibly of poor quality, has been produced.

The object of the invention is to further develop a method for operating a carding machine or a carding machine in such a way that the operator receives an optimal setting of the carding machine depending on the raw material, with which a minimum of neps and/or an energy-efficient mode of operation is possible. The term neps covers the shell neps, the short fiber accumulations and the fiber knots.

The invention solves the problem by a method having the features specified in claim 1. The card according to the invention is further improved by the features of claim 11. The features of claim 16 relate to the design of a spinning preparation system with several cards. Advantageous further developments of the invention are defined in the dependent claims.

The invention relates to a method for operating a carding machine in which fiber flocks are broken down into individual fibers, aligned and cleaned between a clothed rotating drum and stationary or rotating carding elements. The resulting fiber web is transferred from the drum to a doffer and subsequently formed into a fiber sliver.

The invention includes the technical teaching that the raw material and a production quantity are entered into a card control with an operating unit by an operator and the card control determines the reference drum speed. The operator can start an optimization program with which the operation of the card can be switched between a neps and/or energy-optimized mode of operation is possible. The card control system carries out an automatic series of measurements in which several sensors detect the number of neps in the fiber web at different drum speeds over a predetermined period of time and at the same time the drive power of the card is determined, with the data from the sensors and the drive power determined being transmitted to a higher-level control system for spinning preparation. Based on this data, a mathematical algorithm within the higher-level control system suggests to the operator how the card should operate in different quality categories.

Based on different quality categories, the operator can select whether he wants to operate the carding machine with nep optimization, energy optimization, or a mixture of both modes of operation. The advantage of the process is that the higher-level control system can process a large amount of data from individual sensors using the mathematical algorithm. The sensors' computers, on the other hand, evaluate the image data in parallel for their respective track width and forward this to the higher-level control system of the carding machine, which then forwards it to the control system of the spinning preparation system. The processing speed increases when a large amount of data is processed in parallel, and the carding machine is adjusted more quickly without too many fibers with an increased nep value being processed or the carding machine consuming too much energy. If it is technologically sensible and the reference drum speed is already at a limit, the automatic series of measurements can only be carried out above or below this reference drum speed. For example, at high operating temperatures due to the fiber material to be processed or due to hot and humid climatic conditions in the spinning mill, it may be useful to carry out the series of measurements only at speeds below the reference drum speed, since an increased drum speed leads to a further increase in temperature and the carding gap must be regulated back for this purpose.

Alternatively, the different drum speeds can be above and below the reference drum speed, so that the number of neps and the energy consumption are determined for different speed ranges around the reference drum speed.

The different drum speeds can be at equal drum speed intervals above and below the reference drum speed. The equal intervals in the drum speeds increase the processing speed and the accuracy when evaluating the data, especially when only a small amount of data is available.

Because the measurement time is the same for different drum speeds, a simplified mathematical algorithm can be used.

If the cards are fed with a new raw material or the clothing is changed or a different basic setting of the card is made, the measurement to determine the neps and drive power can be carried out several times at each drum speed, with the added duration (t) of the individual measurements giving the total time (T). The number of neps in the card sliver or an energy-optimized mode of operation can be further improved by designing the card control to automatically adjust the carding gap when the card temperature varies. This means that at higher temperatures but the same card speed, the carding gap is kept constant even though the drum expands. By keeping the carding gap constant at different card temperatures, the minimum possible number of neps is always achieved regardless of the card temperature, regardless of the operating mode (energy-optimized or nep-optimized).

Depending on the configuration of the card control and the control of the spinning preparation plant, the optimization program can be started on the control of at least one card or the control of the spinning preparation plant.

Because the result of the optimization program is transmitted from the control of the spinning preparation system to the control unit of at least one card, the operator can manually confirm a selection of the quality category. The operator's experience can prevent a false start, especially if the data basis was insufficient or if the data was incorrectly transmitted between the individual control units.

Alternatively, the result of the optimization program can be determined by the control system of the spinning preparation plant after selection of the quality category and confirmation by the operator, so that an automatic start in the selected operating mode takes place for all cards of the spinning preparation plant.

Because the operator can select directly when starting the optimization program whether at least one carding machine should be operated in a nit-optimized or energy-optimized manner, a greater weighting is given to the classification or clustering of the data with regard to the selected operating mode.

The card according to the invention has an inlet side for fiber flakes, wherein the card is designed to feed the fiber flakes to a rotating drum by means of at least one licker-in, wherein the fiber flakes are broken down to the individual fiber, aligned and cleaned between fixed carding elements and rotating flat bars and the drum, and the resulting fiber web can be transferred from the drum to a doffer, which is followed by a device for converting the fiber web into a fiber band. The card has at least three sensors that are designed to detect the fiber web for neps or short fiber accumulations. The card also has a control system with an operating device by means of which an operator can select, after entering a raw material and the production quantity, whether the card should start an optimization program with which the card should be operated in a nep-optimized and/or energy-optimized mode. If, from the operator's point of view, the raw material is of high quality and a low number of neps is to be expected in the card sliver produced, the operator can directly select the energy-optimized mode of operation so that the spinning preparation control system evaluates the data with a higher weighting for energy optimization. On the other hand, if the raw material is poor or dusty, the operator can directly select the nep-optimized mode of operation so that the spinning preparation control system evaluates the data with a higher weighting in terms of quality. If the operator has little experience or is not yet able to estimate the number of neps, he can start the optimization mode to get a neutral suggestion for an energy-optimized or nep-optimized operating mode. Alternatively, this selection can also be made in the spinning preparation control system.

When the optimization mode is started, at least one card starts a series of measurements in which, if the card settings and raw material remain unchanged, only the setting values are checked. If the raw material has changed, the card starts several series of measurements in which the at least three sensors detect the fiber web, for example on the doffer. The sensors are designed to detect at least neps or short fiber accumulations in the fiber web based on a previous calibration and they transmit the data to the card control. The card control simultaneously monitors the power consumption of the drives and transmits both sets of data to the higher-level control of the spinning preparation.

The sensors can be arranged in a fleece guide element that guides the fiber web between rollers. However, the sensors can also be arranged anywhere else, provided that the flat fiber web between the drum and the web funnel is detected for neps.

Preferably, the carding machine is designed to automatically adjust the carding gap. This allows the number of neps to be further reduced depending on the drum speed. At a lower drum speed, which is associated with lower energy consumption, the carding gap can be reduced via the automatic carding gap adjustment, which can reduce the number of neps.

Because the card is designed to determine the energy consumption at different speeds, it can be operated with an energy-optimized mode of operation, a nuisance-reduced mode of operation or with a mixture of both modes of operation.

The spinning preparation system according to the invention has a control system and several cards, the control system being designed to process at least the data of one card, preferably several cards, using a mathematical algorithm, in a first step the data of the at least one card being clustered or categorized to generate quality categories and in a second step being further processed using a regression model, the result of the mathematical algorithm being transmitted to each card for setting a neps and/or energy-optimized setting. The advantage of the mathematical algorithm lies in the grouping of the data into quality categories, which are subsequently assigned to an optimal drum speed using the regression model. The method delivers reliable results with a high degree of probability even with a small amount of data.

Because the first step of clustering is determined by a K-means algorithm or a fuzzy c-means algorithm or by a hierarchical cluster analysis, the data can be clearly assigned to the selected quality categories. If the operating mode is chosen beforehand based on knowledge of the quality of the raw material, this influences the weighting in the assignment of the data.

Alternatively, the first step of grouping or categorizing the data can be determined by a semi-supervised learning model called label propagation, which allows the system to train itself.

In the second step, the classified data can be further processed by a regression model that is designed as a random forest regressor or a polynomial classifier or an artificial neural network. The number of nits is assigned to a speed with optimized energy consumption. The assignment according to these mathematical models delivers a good result with a high degree of probability even with a small amount of data.

If the data from several cards is processed in the control system of the spinning preparation system and compared with each other in the first step, the control system can issue a visual or acoustic warning signal if the clustered or categorized data deviates from a specified reference. This allows incorrect settings on individual cards to be identified and subsequently corrected.

Further measures improving the invention are presented in more detail below together with the description of a preferred embodiment of the invention with reference to the figures.

Show it:

Figure 1:

a schematic side view of a spinning preparation machine in the form of a card, in which the device according to the invention is used;

Fig.2

an enlarged view of an installed fleece guide element;

Fig. 2a

a detailed view of the fleece guide element;

Fig.3

a first embodiment of a sensor arrangement in a fleece guide element;

Fig. 3a

a second embodiment of a sensor arrangement in a fleece guide element;

Fig.4

a spinning preparation plant with an arrangement of several cards with a control system for summarizing and evaluating data.

Fig.1 shows a card 100 according to the prior art, in which fiber flakes are guided via a shaft to a feed roller 1, a feed table 2, via at least one licker-in 3a, 3b, 3c, to the drum 4 or the tambour. On the drum 4, the fibers of the fiber flakes are parallelized and cleaned by means of fixed carding elements 13, suction hoods and separating knives and by means of rotating carding elements arranged on a revolving flat system 17, which are designed as flat bars 14. The resulting fiber web 16 is then conveyed via a doffer 5, a doctor roller 6 and several squeezing rollers 7, 8 to a fleece guide element 9, which forms the fiber web with a web funnel 10 into a fiber band, which is transferred via take-off rollers 11, 12 to a subsequent processing machine or a can 15. The adjustment of the flat bars 14 and the carding elements 13 to the drum 4 (carding gap) is carried out via sliding strips (not shown here), which can have wedge-shaped elements aligned against each other. The card 100 is adjusted and operated using an operating unit 18, which can be designed as a monitor with an integrated input device.

In Figure 2 and 2a the exemplary arrangement of the fleece guide element 20 between the doffer 5, the doctor roller 6 and a squeegee roller 7 is shown, in which the fiber web 16 is removed from the doffer 5 by the doctor roller 6 and guided along a concave upper side 20c of the fleece guide element 20 to the web funnel 10.

Irrespective of this embodiment, it is not necessary to integrate the sensors 30 described below for detecting neps in this particular fleece guide element. It is important that the sensors 30 are aligned with the flat fiber web 16, somewhere between the drum 4 and the web funnel 10, through which the fiber web 16 is formed into a fiber band.

In this exemplary embodiment, the fleece guide element 20 is essentially formed by four sides 20a, 20c, 20d, 20e, which enclose a cavity 20f. The front side 20a has at least partially a translucent element 20b, which is designed to enable the detection area or the field of view of a sensor 30 located in the cavity 20f to the fiber web 16 located in the clothing 5a of the doffer 5. The translucent element 20b can be arranged only in the area of the viewing angle of a sensor 30, or extend as a continuous translucent element 20b at least partially or completely over the working width of the card 100. The front side 20a of the fleece guide element 20 is thus aligned at a small distance from the surface of the doffer 5. The concave upper side 20c of the fleece guide element 20 guides the fiber web 16 from the doctor roller 6 to the squeeze rollers 7, 8. One difference from the prior art is that the sensor 30 detects the fiber web 16 that is still in the teeth of the set 5a of the pickup 5. The fiber web 16 thus brushes against the almost vertically arranged front side 20a of the fleece guide element 20 and thus also continuously over the translucent element 20b, which thus becomes significantly less dirty than the horizontally arranged concave upper side 20c. According to the prior art, the translucent element was arranged in the horizontally arranged concave upper side 20c, which was not fully touched by the fiber web 16, which allowed dirt to settle. This results in a longer service life before the fleece guide element 20 or the translucent element 20b needs to be cleaned. A single sensor 30 can be arranged so that it can move within the fleece guide element 20, or at least one sensor 30 is arranged in a fixed position. If several fixed sensors 30 are arranged within the fleece guide element, these can be arranged at a regular distance from one another. The transparent element 20b can be designed as a glass or plastic pane, behind which a polarization filter 31 is arranged. The polarization filter 31 is therefore arranged between the transparent element 20b and the sensor 30. The polarization filter 31 is designed for circular polarization, whereby the reflected rays of the sensor 30, which are reflected by a shiny surface, for example the metallic fitting 5a, are blocked out. The reflection of the rays on the matt fibers and Interference particles, however, remain visible to the sensor 30. The sensor 30 thus only detects the fibers and the interference particles contained therein of the fiber pile 16 and can detect thick spots or fiber knots, neps, shell neps or even foreign parts via image analysis. For this purpose, white light is generated, which in combination with the polarization filter can be used to mask out the set 5a. Within the fleece guide element 20, the polarization filter 31 is framed by a reference film 32, with which a white balance can be carried out. The representation of the sensor 30 in the Figure 2a is only schematic.

Each sensor 30 is equipped with its own computer 35, which can immediately evaluate the recorded data. The design of each sensor 30 with its own computer 35 results in parallel processing of the determined data, so that the determined values are available more quickly. The sensor 30 can, for example, be designed as a CCD or CMOS sensor, with which individual images can be recorded. A simple binary assignment of the data takes place in the computer 35. As a boundary condition, for example, a certain size of a shell nit, or a certain size of a fiber nit or fiber knot can be defined, which is done by evaluating the image of the sensor data, either in the individual computer 35 of the sensors 30, or together in the control of the card 100, or together in the control 43 of the spinning preparation system. In the classic sense of measurement technology, this calibrates the sensitivity of the measurement data. The evaluation of the data in the respective computers 35 of the sensors 30 enables faster data processing due to the parallel processing. The evaluated data are transmitted from the computers 35 via the carding machine control system to the control system 43 of the spinning preparation system for categorization or clustering and further processing.

In Figure 3 For example, five sensors 30 are distributed within the fleece guide element 20, preferably evenly at a distance a across the working width A of the card. With a width of the drum 4 of, for example, 1280 mm, this results in a working width A of approximately 1180 mm, which is detected by five sensors, each with a detection width of 20 to 30 mm. In the evaluation of the card control, this results in a separate track for each sensor, which is detected. With fixed sensors 30, with a width of the drum 4 of 1000 mm, a minimum number of three sensors 30 within the fleece guide profile 20 has proven advantageous, with which the number of neps can be determined with sufficient accuracy. With regard to the cost of the sensors 30 and the width of the drum 4, the arrangement of five sensors 30 has proven to be optimal, with which the number of neps can be determined with very high accuracy. The five sensors 30 are preferably arranged at the same distance a across the working width A of the card within the fleece guide element 20. To detect the neps, around 10,000 images must be taken via the sensors per A measured value is determined and evaluated. For a card production of, for example, 80 kg/h, a quantity of 100 meters of fiber web 16 passing through the doffer 5 is formed as a measured value, of which around 10,000 images are taken. In the case of uneven width distribution, a band nib mean value with an error of 3% can be determined using the number of tracks, in this case with five sensors 30. If only three sensors 30 are used, the error of the band nib mean value increases to 12%. If nine sensors are used ( Figure 3a ) 30, which are arranged at a distance b from one another, the error of the band nit mean value is reduced to 1%. The use of a large number of sensors 30 not only increases the accuracy in determining the nits, but also shortens the time required to determine the measured value, since more images are recorded at the same time and the computers 35 process them in parallel. For example, the measuring time for three sensors 30 is approximately 40 seconds, for five sensors 30 approximately 30 seconds, and for nine sensors 30 approximately 15 seconds. Due to the limited installation space within the fleece guide element 20 with a drum width of 1000 mm to 1500 mm, the arrangement of three to nine fixed sensors for determining the nits has proven to be optimal. On the one hand, there is sufficient installation space available over the working width A for the distribution of the sensors 30, on the other hand, sufficient accuracy in combination with a measuring speed is guaranteed, and ultimately the costs are kept within affordable limits.

Even if only a uniform distance a, b between the sensors 30 is shown in this embodiment, this can also be uneven. In this case, the data evaluation algorithm may need to be adapted. It can therefore be advantageous to arrange the sensors 30 at a greater or smaller distance from the center of the fiber web 16, since the detection of certain neps, thick spots or foreign parts can occur more frequently in the edge area (due to side flight) or in the center of the fiber web 16 (due to carding gap differences across the drum width) due to the specific card construction.

If the sensors 30 are also to be used to detect foreign parts, all sensors 30 together must capture around 25,000,000 images. When using five sensors 30 in a fleece guide element 20, each sensor must therefore generate 5,000,000 images before a reliable statement can be made about the presence of foreign parts. A corresponding amount of fiber web must be detected, which means that the measurement process and evaluation takes around 18 hours. When using nine sensors 30, the measurement and evaluation time is still 10 hours, which means that the card production is at least processing a roving and must be destroyed if a serious error occurs.

In order to shorten this process, the invention proposes to combine the data from the sensors 30 of at least one card 100, preferably at least two cards, in a control system 43 of the spinning preparation and to evaluate them together. In contrast to EN 10 2019 115 138 A1 The aim of the invention is to detect an exact number of neps (= shell neps, short fiber accumulations and fiber knots) and thus to optimally adjust all cards in the carding plant. Alternatively or additionally, the cards can be operated in an energy-optimized manner with a maximum number of neps. This does not exclude the possibility that the sensors can also detect foreign parts and other undesirable components and that the upstream blowroom machines can be adjusted by the control system 43, as is the case in the EN 10 2019 115 138 A1 described.

Figure 4 shows a carding machine with a large number of cards 100, the machine control of which is connected to a higher-level control 43 of the spinning preparation. The higher-level control 43 sums up the individual data from the sensors 30 from the individual cards 100 and evaluates them according to different criteria. The evaluation can relate to the proportion of short fibers, or to foreign parts, or to the number of neps or knots, or to characteristics such as cloudiness, thin/thick spots or fiber orientation. A display is designed to show different types of fiber, as well as foreign parts, neps or dirt particles. Via an input module 44, which can be designed as a mobile input with a keyboard such as a laptop, tablet or smartphone, not only a predetermined waste quantity or separation rate can be entered, so that an optimization of the data can already take place within the control 43. The optimization program can be started via this, after which the carding machine can be operated in a neps-optimized and/or energy-optimized manner.

The control system 43 transmits the new specifications with the optimized data to the control system of the individual cards, which can then be adjusted by the operator or automatically depending on the quality requirements and energy consumption. The aim is to optimize the setting so that the required quality of the starting material (fiber ribbon formed from the fiber web) of the card is achieved and maintained.

The combination of the data from the sensors 30 from several cards 100 in the control system 43 shortens the time for capturing the required number of images and reduces the time for processing the data. This means that any faulty card production can be stopped early on before further processing begins in the subsequent spinning mill.

The combination of the data from the sensors 30 of at least two cards in the control system 43 results in another advantage. If the control system 43 determines that the measured values of the sensors 30 change simultaneously or in the same way in all cards, it can be assumed that this has a common cause, such as a change in the raw material, a change in one or more machines or blow room lines, or a common change in the processing conditions, such as the temperature or humidity in the spinning preparation. If, on the other hand, the change is only observed in one card, the cause is more likely to be found in the card itself. Comparing the measured values of at least two cards in the control system 43 can therefore be used to determine more precisely which machine needs to be intervened in order to achieve the desired quality.

When starting the card, the operator enters the raw material and the desired production quantity into the control of each individual card 100 using an operating unit 18. The production quantity in kg/h is also linked to the reference drum speed n _R or drum speed, so that these alternatives can be regarded as equivalent inputs. However, these differ from card type to card type, since the drum diameter and working width can vary depending on the card type. However, according to the current state of the art, this information is stored in the card control, so that only the input varies. Then, for the same production quantity, the carding gap is entered or set either by the operator or by the control of the card 100. A larger carding gap must be set when the machine is cold if a higher drum speed is desired, since heat development then increases and the carding gap therefore changes more. However, the larger carding gap is counterproductive to reducing neps. With a smaller carding gap, the number of neps can be reduced, but fiber damage increases. With automatic carding gap adjustment, the carding gap can be determined automatically by entering the raw material. Although the operator also enters the production quantity into the control unit 18, the carding gap remains constant because it is adjusted to be narrower or wider via the temperature level on the drum 4.

The operator can start the optimization program according to the invention by means of the operating unit 18 on the card 100 or via the control 43 or the input module 44 of the spinning preparation system by operating the card 100 for a predetermined time window with a graduated deviation of the referenced drum speed n _R up or down, starting from the referenced drum speed n _R. The aim is to determine the detected neps and the energy consumption at each drum speed, so that the operator can choose between a neps-optimized and/or energy-optimized operating mode. For example, it starts with the referenced drum speed n _R = 500 rpm for the duration T ₁ , then with n _R -10 rpm for the duration T ₂ , then with n _R -20 rpm for the duration T ₃ , then with n _R -30 rpm for the duration T ₄ . The process is then operated with the referenced drum speed n _R +10 rpm for the duration T ₅ , then with n _R +20 rpm for the duration T ₆ , then n _R +30 rpm for the duration T ₇ . The duration T ₁ ... T _x for the respectively graduated operating mode of the card above and below the referenced drum speed depends on the size of the data determined by the sensors 30, but is the same for all measuring processes. The speed differences to the referenced drum speed n _R can occur at equal intervals, for example in steps of 10 or 20, or at different intervals and rows.

If, from the operator's point of view, the raw material is of high quality and a low number of neps is to be expected in the card sliver produced, the operator can directly select the energy-optimized operating mode so that the spinning preparation control evaluates the data with a higher weighting for energy optimization. However, if the raw material is poor or dusty, the operator can directly select the nep-optimized operating mode so that the spinning preparation control evaluates the data with a higher weighting in terms of quality. If the operator has little experience or is not yet able to estimate the number of neps, he can start the optimization mode to receive a neutral suggestion for an energy-optimized or nep-optimized operating mode. Alternatively, this selection can also be made in the spinning preparation control.

Two data sets are used to determine the detected nits and the energy consumption. One relates to the production-relevant data, which is determined by the sensors 30, among other things. The other data set relates to the machine status data, in which the energy consumption is determined at a certain drum speed. For each measurement run to determine the number of nits, the current power of the drive is also determined. Both data sets are combined. Several measurement runs with an individual time or duration t ₁ ... t _x can be carried out for each duration T ₁ to T _x in order to determine the result with a high degree of certainty. If the raw material has been used unchanged for several hours, a single measurement run is sufficient for a check. If a new raw material is used or another machine setting has changed, it is advisable to carry out multiple measurement runs with the same drum speed. For example, with a drum speed of n _R = 500 rpm for the duration T ₁ = 30 min, five measurement runs can be started, each lasting t = 6 minutes. Of course, this requires a calibration of the sensitivity of the sensors 30, from which size of a short fiber accumulation or nep a measured value is counted. A control of the sensitivity can be carried out by manually evaluating a fiber band taken in a textile technology laboratory, so that due to this feedback the data from all sensors 30 are evaluated in the same way. Since at least three fixed sensors 30 are used according to the invention, at least three data sets with a number of neps are obtained for each sensor 30 and measuring process, which are shown as an example over the working width in the following table. This means that for each measuring process 4.1 to 4.5, corresponding to the number of sensors 30 - here three sensors 30 - there are measured values for a partial width or track of the working width A, which are extrapolated to the entire working width A and thus to production. Table 1 No. Duration t [min] Drum speed [n] Delivery speed [m/min] Power kW] Number of nits [1/g] 4.1 6 500 350 11.5 93 4.2 6 500 350 11.2 92 4.3 6 500 350 11.4 98 4.4 6 500 350 11.5 95 4.5 6 500 350 11.8 97

These multiple measurement runs according to Table 1 are carried out in equal numbers for each drum speed. An evaluation of all measurement runs for the respective drum speed is carried out, for example, after clustering or categorizing the measured values or data, whereby the measured values 4.1 to 4.5 were summarized in Table 2 below under measured value no. 4. The duration T and t of all individual measurements is the same. The duration or time T used here for each speed range corresponds to the sum of the individual measurement duration t. Table 2 No. Duration T [min] Drum speed [n] Delivery speed [m/min] Power kW] Number of nits [1/g] 1 30 470 350 10.5 156 2 30 480 350 10.8 130 3 30 490 350 11.2 112 4 30 500 350 11.5 95 5 30 510 350 11.8 75 6 30 520 350 12.1 79 n 30 530 350 12.5 87

The table shows that as the drum speed increases, energy consumption increases, but up to a certain point the number of detected neps also decreases, even though the carding gap remains unchanged. As the drum speed increases, which is given here as 520 rpm, for example, the number of detected neps can increase again.

The evaluation of the measured values is carried out in the control system 43 of the spinning preparation by a mathematical algorithm which runs as follows:
The measured values are evaluated by clustering the data. Each cluster is assigned a quality category of high, medium or low. The data of each cluster can be processed in a K-Means algorithm, whereby the drum speed is predicted by a regression model for each quality category. The K-Means algorithm is used for cluster analysis of data, whereby a previously known number of k groups is formed from a set of similar objects. The value for k corresponds to the number of quality categories, which can be between 1 and X, and can therefore also include 5 or 10 categories. In this example, k=3 was chosen for the three quality categories of high, medium and low. The value k can be determined using the elbow method or the silhouette coefficient. With the elbow method, a value for each k is calculated using the sum of the squared distances of the data points to their nearest cluster center, while the silhouette coefficient is calculated using the similarity between a data point and its own cluster in comparison to other clusters. As an alternative to the K-Means algorithm, other methods can be used to categorize or cluster the data, such as fuzzy c-means algorithm or hierarchical cluster analysis.

Alternatively, the measured values can be summarized into quality categories using the semi-supervised learning model label propagation. To do this, the user must set boundary conditions, so-called labels. These are predefined target variables that correspond to a designation of a class. As a boundary condition, a small amount of data is labeled by the operator from experience by predefining or entering a small range of the number of nits for the respective quality category. These n labels are then propagated and in the end exactly n classes are created.

The data from both concepts can be further processed in the spinning preparation control system 43 using a regression model that determines or predicts the optimal drum speed for each quality category. This can be a random forest regressor that delivers reliable results with a high degree of probability even with a small amount of data. This is a tree algorithm that can be easily adapted to the number of decision nodes and the depth using its parameters. Alternatively, a polynomial classifier or an artificial neural network could be used.

As a result, the control 43 of the spinning preparation is designed to transmit the measurement data of at least one card evaluated by means of an algorithm to the card 100, so that after selecting the desired quality level, it can be operated either with the lowest possible number of neps or in an energy-optimized manner. The control of the card can automatically set the associated drum speed and operate in the desired operating mode, or display the suggested values to the operator on the control unit 18 and leave the selection to him with a confirmation button or by setting the card. The data determined via the evaluation model for the following quality levels do not have to match the actual measured values, but are a correlation of the determined data: Table 3 Quality level Drum speed [n] Delivery speed [m/min] Power kW] Number of nits [1/g] High 510 350 11.8 75 Medium 490 350 11.2 108 Low 470 350 10.5 135 Table 3 shows a recommendation if the operator has not selected a nitrate or energy optimized mode of operation when starting the optimization mode.

For example, if the operator has selected a nit-optimized setting, the weighting factor in the K-Means method is changed and the recommended setting may look like Table 4: Table 4 Quality level Drum speed [n] Delivery speed [m/min] Power kW] Number of nits [1/g] High 520 350 12.0 68 Medium 500 350 11.5 92 Low 480 350 10.8 115

With an energy-optimized setting or selection by the operator, the weighting factor in the K-Means method is changed and the recommended setting can look like Table 5: Table 5 Quality level Drum speed [n] Delivery speed [m/min] Power kW] Number of nits [1/g] High 480 350 10.8 115 Medium 470 350 10.5 135 Low 450 350 10.3 263

Without having to select the desired operating mode beforehand, after starting the optimization program and running through the measurement intervals at different speeds, the operator can select whether he wants to operate the card with the lowest possible number of neps or with an acceptable number of neps within specified limits at which the card's energy consumption is minimal. The operator receives a suggestion from the control 43 of the spinning preparation system, which is displayed on the operating unit 18 and which only needs to be confirmed. Alternatively, the control 43 of the spinning preparation system can be designed to intervene in the control of at least one card 100 and to automatically start the desired operating mode according to the operator's specifications. The operator then makes the selection at the level of the control 43 of the spinning preparation system.

As a suggestion, Table 3 above can be displayed or represented graphically. Alternatively, the values in Table 3 can also be displayed on the control 43 of the spinning preparation and from there at least one card 100 can be started automatically or manually. The same applies if the operator has directly selected a nep-optimized or energy-optimized operating mode and then either the data in Table 4 or Table 5 is displayed or represented graphically.

If it is ensured that the cards have the same settings, for example setting of licker-in, adjustable knives, fixed carding elements, distance between doffers, etc., the test run on one card is sufficient to set all the other cards using the determined values via the control 43 of the spinning preparation. The process is time-optimized, however, and runs faster with at least two cards, as more data can be processed in parallel.

With the automatic adjustment of the carding gap, the carding gap is kept at a constant value, regardless of the state of wear of the clothing, which only depends on the raw material and the temperature of the card. In this respect, worn clothing can increase the number of neps, but the optimized setting of the card recommended in Tables 3 to 5 also applies here.

A laboratory test of the card sliver showed a deviation of 1% to 3% for different quality levels compared to the data determined by the K-Means algorithm with Random Forest Regression.

Reference symbols

100

Cardoon

1

Feed roller

2

Dining table

3a, 3b, 3c

Lingerer

4

drum

5

customer

5a

set

6

Scraper roller

7.8

Squeeze roller

9

Fleece guide element

10

Flort funnel

11, 12

Take-off rollers

13

Carding element

14

Cover bar

15

Jug

16

Fiber pile

17

Revolving lid system

18

Control unit

20

Fleece guide element

20a

front

20b

element

20c

Top

20d

back

20e

bottom

20f

cavity

30

sensor

31

Polarizing filter

32

Reference slide

35

calculator

43

steering

44

Input module

A

Working width

a

Distance

b

Distance

No

Reference drum speed

t1 - tx

Time

T1 - Tx

Time

Claims (20)

Method for operating at least one card in a spinning preparation plant, in which fiber flocks are broken down to the individual fiber between a garnished rotating drum (4) and fixed and rotating carding elements, aligned and cleaned and the resulting fiber web is transferred from the drum (4) to a doffer (5) and subsequently formed into a fiber band, having a card control with an operating unit (18), wherein after input of a raw material and a production quantity by an operator, the card control determines the reference drum speed (n _R ), characterized in that the operator starts an optimization program that allows the operator to choose between a nep and/or energy-optimized operating mode, whereby the card control carries out an automatic series of measurements in which several sensors (30) detect the number of neps in the fiber web at different drum speeds over a respective predetermined period of time (T) and at the same time the drive power of the card is determined, wherein the data of the sensors (30) and the drive power determined thereby are transmitted to a higher-level control (43) of the spinning preparation department, which, on the basis of at least these data and by means of a mathematical algorithm, proposes to the operator an operating mode of the card in different quality categories. Method according to claim 1, characterized in that the different drum speeds are above and below the reference drum speed (n _R ). Method according to claim 2, characterized in that the different drum speeds are at equal drum speed intervals above and below the reference drum speed (n _R ). Method according to claims 1 to 3, characterized in that the time period (T) is the same for the different drum speeds. Method according to claims 1 to 4, characterized in that the measurement for determining the neps and drive power is carried out several times at each drum speed, whereby an added duration (t) of the individual measurements gives the total time duration (T). Method according to one of the preceding claims, characterized in that the control of the carding machine is designed to automatically adjust the carding gap at different speeds of the drum (5). Method according to one of the preceding claims, characterized in that the optimization program is started at the control of the at least one card (100) or the control (43) of the spinning preparation plant. Method according to one of the preceding claims, characterized in that the result of the optimization program is transmitted from the control (43) of the spinning preparation plant to the operating unit (18) of the at least one card (100) for manual selection of the quality category and confirmation by the operator. Method according to one of the preceding claims 1 to 7, characterized in that the result of the optimization program after selection of the quality category and confirmation by the operator through the control (43) of the spinning preparation plant, the automatic operation of all cards (100) of the spinning preparation plant is started. Method according to claim 1, characterized in that the operator can directly select, when starting the optimization program, whether the at least one carding machine is to be operated in a nib-optimized or energy-optimized manner. Card with an inlet side for fiber flakes, the card being designed to feed the fiber flakes to a rotating drum (4) by means of at least one licker-in (3a, 3b, 3c), the fiber flakes being broken down to the individual fiber, aligned and cleaned between fixed carding elements and rotating flat bars (17) and the drum (4), and the resulting fiber web being able to be transferred from the drum (4) to a collector (5) which is followed by a device for converting the fiber web into a fiber band, having at least three sensors (3) which are designed to detect the fiber web for neps or short fiber accumulations, characterized in that the card has a control with an operating device (18) by means of which an operator can start an optimization program after entering a raw material and the production quantity, with which the card is to be operated in a nep-optimized and/or energy-optimized mode of operation. Card according to claim 11, characterized in that the at least three sensors (30) detect the fiber web on the doffer (5). Card according to one of claims 11 to 12, characterized in that the sensors (30) are arranged in a fleece guide element (20). Card according to claim 11, characterized in that the card is designed to adjust the carding gap automatically. Card according to claim 11, characterized in that the card is designed to determine the energy consumption at different speeds. Spinning preparation plant with a control system (43) having a plurality of cards (100) which are operated according to a method of claims 1 to 9, wherein the control system (43) is designed to process at least the data of one card (100), preferably of a plurality of cards, with a mathematical algorithm, wherein in a first step the data of the at least one card (100) are clustered or categorized to generate quality categories and in a second step are further processed with a regression model, wherein the result of the mathematical algorithm is transmitted to each card (100) for setting a neps and/or energy-optimized setting. Spinning preparation plant according to claim 16, characterized in that the first step of clustering is determined by a K-means algorithm or a fuzzy c-means algorithm or by a hierarchical cluster analysis. Spinning preparation plant according to claim 16, characterized in that the first step of categorization is determined by a partially supervised learning model label propagation. Spinning preparation plant according to claim 16, characterized in that a random forest regressor or a polynomial classifier or an artificial neural network is used as the regression model. Spinning preparation system according to claim 16, characterized in that the data of several cards are processed in the control system (43) of the spinning preparation system and compared with each other in the first step, so that if the clustered or categorized data deviate from a predetermined reference, the control system outputs an optical or acoustic warning signal.

Images