METHOD FOR OPERATING A DEVICE, COMPUTER PROGRAM PRODUCT AND DEVICE FOR PRODUCING A PRODUCT

Abstract

A method for operating a device to produce a product includes capturing at least one quality data record that includes measured values of one or more quality parameters that each correspond to one property of the product. The method captures at least one associated machine data record including actual values of several adjustable machine parameters of the device, chronologically assigns the quality data record to the machine data record, and generates a first data record comprising chronologically-correlated measured values and actual values. The preceding steps are repeated at least once to generate at least one second data record. A correlation is determined between the quality parameter(s) and the machine parameter(s). The method provides a corresponding target value for at least one of the adjustable machine parameters on the basis of the control model, proceeding from a target value of the quality parameter(s).

Claims

1. A method for operating a device for producing a product, the method comprising: capturing at least one quality data record comprising measured values of one or more quality parameters that each correspond to one property of the product; capturing at least one associated machine data record comprising actual values of one or more adjustable machine parameters of the device; chronologically assigning the quality data record to the machine data record and generating a first data record comprising chronologically-correlated measured values and actual values; repeating the preceding steps at least once to generate at least one second data record; determining a correlation between the one or more quality parameters and the one or more adjustable machine parameters by comparing the captured first and second data records and creating a control model for the device; providing a corresponding target value for at least one of the one or more adjustable machine parameters on the basis of the control model, and proceeding from the corresponding target value of the one or more quality parameters.

2. The method according to claim 1, further comprising: capturing at least one environmental data record comprising measured values of one or more environmental parameters; and assigning the one or more environmental parameters chronologically to the machine data record in order to therefore form a part of the respective captured first and second data records.

3. The method according to claim 1, characterized in that the data records are determined at least once on the basis of test results of the device for producing a product and are provided for creating the control model for the device.

4. The method according to claim 1, wherein the first and second data records are determined by collecting a plurality of chronologically-correlated measured values and actual values from production plants and creating the control model for the device using the collected chronologically-correlated measured values and actual values.

5. The method according to claim 2, wherein actual value of the one or more adjustable machine parameters and at least one of the one or more environmental parameters are captured, and, proceeding from the actual value, the measured values of the one or more quality parameters corresponding to the actual value are output using the control model, and this value or these values are compared with the target value(s) of the quality parameter(s) and a deviation is determined, and wherein, proceeding from this deviation, the device is controlled based at least in part on the actual value of the at least one environmental parameter.

6. The method according to claim 5, further comprising: comparing actual values of the one or more adjustable machine parameters with the respective corresponding target value; and controlling the device based on a deviation of the actual value of the one or more adjustable machine parameters from the respective corresponding target value.

7. The method according to claim 5, wherein the actual values of the one or more adjustable machine parameters and of the at least one environmental parameters are continuously captured while the device is operating.

8. The method according to claim 5, wherein the actual values of the one or more adjustable machine parameters and of the at least one environmental parameters are captured cyclically while the device is operating.

9. The method according to claim 1, wherein the control model is determined according to the correlation between the one or more quality parameters and the one or more adjustable machine parameters, taking into account the at least one of the one or more quality parameters by means of machine learning.

10. The method according to claim 1, wherein the machine data record comprises a plurality of adjustable machine parameters, and a weighting of an influence of each one of the plurality of adjustable machine parameters on each of the one or more quality parameters is assigned to each of the one or more adjustable machine parameters according to the correlation with the one or more quality parameters of the quality data record.

11. The method according to claim 10, wherein achieving the target value of the one or more of the quality parameters, at least one of the one or more of the respective adjustable machine parameters is regulated with the respective weighting with respect to the one or more quality parameter(s) according to the control model.

12. The method according to claim 5, further comprising determining whether one or more of the actual values of the one or more adjustable machine parameters or the at least one environmental parameters captured while controlling lie within a value range of the one or more adjustable machine parameters captured in the creation of the control model and/or the captured at least one environmental parameters.

13. The method according to claim 12, wherein, in the event that the actual values are outside the value range, determining whether the deviation of the actual values repeatedly occurs, and, if so, outputting an error signal.

14. The method according to claim 12, wherein, in the event that the actual values are outside the value range, determining whether the deviation from the value range is significant for the control model, and if so, outputting an error signal.

15. The method according to claim 14, wherein the deviation is is significant if either the deviating actual value relates to one or more adjustable machine parameter which is regulated with a high weighting to the target value of one or more quality parameters, or if the deviation has exceeded a predetermined threshold value.

16. The method according to claim 15, wherein the control model switches off the device, or requests manual control interventions, if the deviation is significant or repeatedly occurs.

17. A computer program product comprising commands which, when run on a computer, cause the device to execute the steps of the method according to claim 1.

18. A device for molding a hollow body, comprising a computer program product according to claim 17.

19. The device according to claim 18, wherein the device is a component of a device configured for one or more of (i) injection molding or for blow molding, (ii) a dryer for starting material for the production of the product, (iii) an extrusion blowing device, or (iv) a stretching blowing device.

20. The device according to claim 19, wherein the device is an extruder.

Description

[0080] The invention will be explained in more detail below with reference to schematic figures, which show only exemplary embodiments. The following are shown:

[0081] FIG. 1: a schematic representation of a device for producing a product

[0082] FIG. 2: a flowchart of a method according to the invention

[0083] FIG. 1 shows a simplified schematic representation of a device for producing a product which, in the present case, is designed as, for example, a blow-molding machine 100. The blow-molding machine 100 has an infeed 110, a blow-molding region 120, and a discharge 130. Preforms are fed to the blow-molding machine in a known manner by means of the infeed 120. In the blow-molding region 120, these are inflated with compressed air and are thereby stretched with the aid of a rod. In the discharge 130, the completely inflated containers are collected and/or removed. This production process as such is known and will therefore not be explained in more detail here.

[0084] FIG. 1 also shows a controller 200. Via connections 102 indicated here by dashed lines, the controller 200 is connected to sensors arranged on the blow-molding machine 100. The controller 200 may be designed as a separate unit, but is generally an integral part of the device. The sensors may be temperature sensors, clock generators, position sensors, and the like. Actual values of machine parameters can be captured with the sensors. The sensors in this case are only schematically indicated. A temperature sensor 101 is shown as a placeholder for a plurality of sensors.

[0085] In the shown exemplary embodiment, the connections 102 between the controller 200 and the blow-molding machine are wired via cables. However, these connections can also be implemented wirelessly or via fiber-optic cables.

[0086] Furthermore, FIG. 1 shows a human-machine interface (HMI unit), i.e., an operating unit 103. Via this operating unit 103, a machine operator can monitor the device and, for example, specify a target value of a quality parameter.

[0087] FIG. 2 shows a schematic process of a method for operating a device for producing a product, e.g., a method for operating the blow-molding machine 100 from FIG. 1.

[0088] In a first step, a quality data record 10 is captured. The quality data record 10 comprises, for example, several measured values of wall thicknesses of a container.

[0089] In a second step, a machine data record 20 is captured. The machine data record 20 comprises actual values of the temperature sensor 101 (see FIG. 1).

[0090] The quality data record 10 and the machine data record 20 have been captured simultaneously. Accordingly, the quality data record 10 can be chronologically assigned to the machine data record 20, and a data record 30 can be formed. Within the data record 30, the measured values and the actual values are correlated chronologically.

[0091] This temporal correlation will be explained in more detail with reference to the following simplified example. At the point in time t1, for example, the wall thickness D1 of a first container B1 in the discharge 130 (see FIG. 1) is measured. The wall thickness D1 corresponds to a quality parameter. At the same time, the temperature K1 of the cavity in which a second container B2 is inflated at the same point in time t1 is measured in the blow-molding region 120 (see FIG. 1). The temperature K1 corresponds to a machine parameter. The temperature V1 of a preform of a third container B3 in the infeed 110 (see FIG. 1) is again measured simultaneously. This temperature V1 can be treated as a machine parameter or as a quality parameter. In the present example, the temperature V1 is treated as a machine parameter.

[0092] In a next step, all containers B1, B2, and B3 are moved one station further. That is, the preform of the third container B3 is moved further from the infeed 110 into the blow-molding region 120, and the inflated second container B2 is moved further from the blow-molding region 120 into the discharge region 130, while the first container B1 is removed from the discharge region 130. A new preform of a fourth container B4 is provided in the supply line 110.

[0093] At the point in time t2, the wall thickness D2 of the container B2 is measured. The preform of the container B3 is located in the cavity for inflating the container B3, wherein the temperature K2 is measured. At the same time, the temperature V2 of the new preform of a container B4 is re-measured. These processes are now repeated for the other containers B2, B3 through Bn.

[0094] That is, the temperature K1 of the container B2 is measured at time t1, but the effect of this temperature K1 on the wall thickness D2 of the container B2 is measured only at time t2. In other words, for a measured value of a quality parameter—in this example, D2—there are a plurality of actual values that lie at different intervals in time before the point in time at which the measured value of the quality parameter is captured. These temporal correlations are created in the data record 30.

[0095] Following the creation of the data record 30, a second data record 30′ is created. This comprises capturing a second machine data record 20′, capturing a second quality data record 10′, and accordingly generating a data record 30′ as described above.

[0096] The quality data records 10 and 10′ may comprise further quality parameters, which are preferably all captured simultaneously. Therefore, a further quality parameter in addition to the wall thickness can, for example, be the opacity of a wall of the container, or the concentricity of a closure relative to a container bottom.

[0097] The machine data records 20 and 20′ can also have measured values of further machine parameters.

[0098] In the next step, the data records 30 and 30′ are merged, and a correlation 40 between the measured values and the actual values is determined.

[0099] On the basis of these correlations, a target value 50 of the captured machine parameter, or, when there are several machine parameters in the respective machine data records 20, 20′, target values 50 for each captured machine parameter, can be determined for each target value of a quality parameter. As already stated, there may be several combinations of target values 50 of the machine parameters which lead to the same target value of the quality parameter. In the event that, for example, a machine parameter is invariable (or, if applicable, an invariable environmental parameter is included in the method), the target values must be determined on the basis of two fixed values (target value of the quality parameter and invariable value of the device/environment), which reduces the possible combinations. These relationships are stored in a control model for the device, or a control model for the device is created on the basis of these relationships.

[0100] Since a plurality of data records 30, 30′, represented as data records 30″, is, usefully, generated, machine learning is used to determine the correlations 40. This makes it possible to compare a plurality of values with one another and to determine similarities or patterns and the like, and link them to one another, even if the relationships are no longer obvious.

[0101] In a subsequent step of the method, a target value of the quality parameter is selected, and corresponding target values 50 of machine parameters are transmitted to the controller 200 (see FIG. 1) in order to correspondingly operate the blow-molding machine 100 (see FIG. 1). In doing so, the control model, or only the corresponding values of the machine parameters that were determined using the control model, can be transmitted to the control unit. As soon as the measured actual values of the blow-molding machine 100 (see FIG. 1) deviate from the target values 50, the control system can readjust them. Should, for example, values change on which the control system cannot exert any influence (for example, the outside temperature), the target values 50 can be adapted in accordance with a specification for the corresponding non-influenceable value. For this purpose, the target values 50 are anchored in a data matrix—in particular, in a control model—which has been transmitted to the control 200 as part of the target value 50 of the quality parameter (see FIG. 1).