METHOD FOR MONITORING A MACHINE

Description

[0001] The invention relates to a method for monitoring a machine, wherein a transfer step of a first pre-product from a first machine part to a second machine part and / or a rolling process of the first pre-product is carried out. Furthermore, the invention relates to a computer program product.

[0002] Discontinuously operating machines in production, such as tire building machines, are monitored for their geometric properties either manually or with optical sensors that monitor the geometry of individual components. This requires a high level of personnel input or a large number of optical sensors that record the geometries of the individual components. Retooling the machines requires resetting the optical sensors.

[0003] It is known that continuously running machines, such as ball bearings, can measure accelerations, thereby detecting a change in the vibration profile. Patent document DE 10 2019 217712 A1 discloses a method for measuring an open or overlapping joint of a material layer during the production of a green tire in a tire building machine.

[0004] Against this background, the object of the invention is to implement a method for monitoring a machine in such a way that a machine can carry out a transfer step of a first preliminary product from a first machine part to a second machine part and / or during a rolling process of the first preliminary product in a cost-efficient, high-frequency and reliable manner.

[0005] This object is achieved by a method according to the features of patent claim 1 and a computer program product according to the independent claim. The subclaims relate to particularly useful developments of the invention.

[0006] According to the invention, a method for monitoring a machine is provided, wherein a transfer step of a first preliminary product from a first machine part to a second machine part and / or a rolling process of the first preliminary product is carried out, wherein the acceleration measurement values ​​or characteristic figures derived therefrom of a component of the rubber processing machine are determined, wherein the acceleration measurement values ​​are compared with a reference value, wherein a signal is output when a difference between the acceleration measurement values ​​or the derived characteristic figures and the at least one reference value is exceeded.

[0007] A rolling process is a process step in which a pre-product is pressed onto a component or another pre-product by a rotationally symmetrical tool rolling on the pre-product.

[0008] According to the invention, the method allows a manufacturing step leading to a defective product to be detected in a discontinuous process based on acceleration values ​​of a component of a machine. Upon detection of such a process step, a signal is output so that measures can be taken to adapt the process step causing the defect. It has been recognized that a change in the vibration profile indicates a defective component not only in continuously running processes that have a continuous vibration profile, but that a discontinuous vibration profile of a discontinuous process can also be used to monitor a machine.

[0009] A preferred embodiment provides that the acceleration measurement value contains information about the frequency and / or the amplitude of individual accelerations, as well as preferably about the temporal behavior, i.e., the temporal progression, of individual accelerations. The frequencies, or in the case of an oscillation, the duration of the oscillation, as well as the amplitude or magnitude of the acceleration, have proven particularly suitable for determining a faulty process step. It has been shown that, in particular, the temporal relationship between individual accelerations or individual acceleration measurement values ​​is particularly suitable for detecting process steps that lead to defective products.

[0010] A further preferred embodiment provides that the method has a start time, wherein the start time can be set by a process step of the monitored process and / or by exceeding a threshold acceleration value, wherein the process cycle is preferably terminated after less than 5 seconds, more preferably after less than 1 second and most preferably after less than 0.2 seconds. It has been shown that a process cycle that is as short as possible leads to more reliable detection of a faulty process step. In this case, it has proven advantageous to set the start time by a specific event, since this allows the measured values ​​recorded at a point in time in the process, i.e. a specific sub-step of the process, to be easily compared with one another. A particularly reliable detection of a faulty process step is achieved.

[0011] A further preferred embodiment provides that the difference between acceleration measurement values ​​and / or characteristic numbers derived therefrom and at least one reference value with regard to the frequency and / or the amplitude and / or the time period that has elapsed since the starting time is formed.

[0012] The difference between the acceleration measurement value and the reference value in terms of its frequency or duration as well as its amplitude or size and the time of its occurrence during the process cycle have proven to be particularly suitable for detecting a faulty process step.

[0013] A further preferred embodiment provides that the reference value is determined by reference processes. It has been shown that reference values ​​generated by reference processes particularly reliably identify faulty process steps, while fault-free process steps are particularly reliably identified as such. Reference processes are understood to be process executions that generate reference values ​​generated by fault-free and deliberately faulty processes and are manually classified as faulty or fault-free.

[0014] A further preferred embodiment provides that the reference value is generated from a plurality of manufacturing processes, wherein the generation of the reference value is preferably carried out using artificial intelligence, machine learning, and / or neural networks. Generating reference values ​​from a plurality of manufacturing processes is advantageous because particularly low personnel expenditure is required to train the method, since the method evaluates accelerations of a large number of manufacturing processes and independently generates thresholds for faulty and error-free processes. The use of artificial intelligence, machine learning, and / or neural networks has proven particularly advantageous.

[0015] A further preferred embodiment provides that a comparison of the acceleration measurement values ​​or derived key figures with the at least one reference value can lead to more than two different signal outputs, in particular to a signal that the acceleration measurement values ​​and / or derived key figures are within the target value and / or that the acceleration measurement values ​​and / or derived key figures are within the target value and are close to a non-target range with regard to their sizes and / or that the acceleration measurement values ​​and / or derived key figures are in a specific one of several non-target ranges. It has proven advantageous to enable more than two different signal outputs since this can, for example, already provide an indication of a possible cause of an error through the signal.

[0016] A further preferred embodiment provides that the machine is designed for the production of rubber precursors or rubber products, in particular for tire building. Machines for the production of rubber precursors or rubber products, in particular tire building machines, involve a large number of transfer steps and rolling processes, making the method particularly advantageous for these machines.

[0017] According to the invention, a computer program product is provided for determining an acceleration measurement value of a machine, which comprises instructions which, when the program is executed by at least one processor unit, cause the processor unit to carry out the method according to the invention.

[0018] The invention allows for numerous embodiments. To further clarify its basic principle, measurement curves associated with the method are shown in the drawing and are described below.Fig. 1 Curves of acceleration over time of machine parts.

[0019] Figure 1shows curves of accelerations over time of machine parts, where the dotted line 1 represents an acceleration curve of a transfer step of a preliminary product. The curve begins with small positive acceleration values, which are plotted upwards on the y-axis, and then small negative acceleration values. Following this, with increasing values ​​on the x-axis, medium positive and larger negative acceleration values ​​are shown, which alternate in rapid succession and form an initial section with larger acceleration values. The medium and larger acceleration values ​​are only reached over a very short period of time. The accelerations form peaks. In the further course, sections with low accelerations are shown, which alternate with sections with larger acceleration values. In the second half of the curve, only smaller accelerations are shown.The acceleration curve of the dotted line 1 was determined when transferring a pre-product during the production of a good part.

[0020] Dashed line 2 and dot-dash line 3 represent two further acceleration curves of machine parts. The accelerations were determined during the transfer of a pre-product during the production of scrap. The acceleration curves begin with low values ​​and then reach large positive values ​​and medium negative values, which persist over relatively short periods of time. Further on, only smaller values ​​were determined, which exhibit a periodic sequence in some sections.

Claims

1. Method for monitoring a machine, wherein a step of transferring a first preliminary product from a first machine part to a second machine part and / or an operation of rolling on the first preliminary product is performed, characterized in that measured acceleration values of a component of the rubber-processing machine are determined, wherein the measured acceleration values or characteristic values derived on the basis of these are compared with at least one reference value, with a signal being output if a difference between the measured acceleration values or the derived characteristic values and the at least one reference value is exceeded.

2. Method according to Claim 1, characterized in that the measured acceleration value comprises information about the frequency and / or the amplitude of individual accelerations and preferably about the temporal behaviour of individual accelerations.

3. Method according to either of Claims 1 and 2, characterized in that the method has a starting point in time, wherein the starting point in time can be set by a process step of the monitored process and / or by the exceeding of a threshold acceleration value, with the method cycle preferably being ended after less than 5 seconds, more preferably after less than 1 second and most preferably after less than 0.2 seconds.

4. Method according to one of the preceding claims, characterized in that the difference between measured acceleration values or characteristic values derived therefrom and at least one reference value with regard to the frequency and / or the amplitude and / or the period of time that has elapsed since the starting point in time is formed.

5. Method according to one of the preceding claims, characterized in that the reference value is ascertained by reference processes.

6. Method according to one of the preceding claims, characterized in that the reference value is generated from a plurality of production processes, wherein the generation of the reference value is preferably generated by using artificial intelligence, machine learning and / or neural networks.

7. Method according to one of the preceding claims, characterized in that a comparison of the measured acceleration values or derived characteristic values with the at least one reference value can lead to more than two different signal outputs, in particular in relation to one signal, in that the measured acceleration values and / or the characteristic values are on target and / or in that the measured acceleration values and / or the characteristic values are on target and in terms of their magnitudes are in the vicinity of a non-target range and / or in that the measured acceleration values and / or the characteristic values are in a specific one of a number of non-target ranges.

8. Method according to one of the preceding claims, characterized in that the machine is designed for producing preliminary rubber products or rubber products, in particular for building tyres.

9. Computer program product for ascertaining a measured acceleration value of a machine, comprising commands which, when the program is executed by at least one processor unit, causes it to perform the method according to one of Claims 1 to 8.