METHOD FOR OPERATING A FIELD DEVICE OF PROCESS MEASURING TECHNOLOGY AND A FILLING PLANT WITH WHICH THE METHOD IS EXECUTED

Description

[0001] The invention relates to a method for operating a field device for process measurement technology, wherein the field device is in communication connection with at least one other bus participant via a fieldbus for transmitting fieldbus messages, wherein the field device is predefined with at least one state rule for switching from normal operation of the field device to diagnostic operation of the field device when the state rule is fulfilled, wherein the state rule depends on at least one state variable of the field device known to the field device. Furthermore, the invention also relates to a filling system with a first filling point and with at least one second filling point, wherein the first filling point has at least one first filling valve, at least one first flow or level sensor, and at least one first control and evaluation unit for controlling and monitoring the filling process of the first filling point, and wherein the second filling point has at least one second filling valve.comprising at least one second flow or level sensor and at least one second control and evaluation unit for controlling and monitoring the filling process of the second filling point, wherein the first filling valve, the first flow or level sensor, the second filling valve and the second flow or level sensor are in communication connection with each other via a fieldbus, wherein at least the first control and evaluation unit is given a state rule for switching from normal operation of the first flow or level sensor to diagnostic operation of the first flow or level sensor when the state rule is fulfilled, wherein the state rule depends on at least one state variable of the first flow or level sensor known to the first flow or level sensor.

[0002] Field devices used in process measurement technology are typically installed "in the field," meaning in an industrial environment such as an automation and / or process plant. These field devices, with their specific measurement tasks, can include flow, level, pressure, temperature, or pH meters. Field devices can also function as actuators, such as control valves, linear actuators based on various technologies (electric, hydraulic, pneumatic), stepper motors, and so on.

[0003] In many process plants, field devices of both categories are used together, for example, to implement process control procedures. A good example of the combined use of such field devices is filling systems, where defined quantities of a medium are filled into designated containers. The filling points of a filling system typically each have a filling valve and a flow or level sensor. The filling valve allows active control over the fill quantity, while the flow or level sensor monitors the filling process through measurement.

[0004] Many field devices used in process measurement technology, in addition to a normal operating mode (representing the "normal" operating state of the field device – in the case of a field device functioning as a measuring instrument), also have a diagnostic mode. As the name suggests, this diagnostic mode serves to diagnose the field device, checking its correct functionality. This is necessary to ensure reliable process operation, for example, in the aforementioned filling plant, in which the field device is integrated. The diagnostic procedures can vary considerably from one field device to another and from one measurement principle to another.Common diagnostic methods include zero-point monitoring, which is used in many measurement concepts, electrical conductivity monitoring, or – for example, when using magnetic-inductive flowmeters – monitoring of the electrode noise of the measuring electrodes. The specific type of diagnostic method and the precise diagnostic procedure are not important here.

[0005] Diagnostic procedures can often only be performed effectively if the field device is in a suitable condition. This applies, for example, to zero-point monitoring, which verifies whether, for instance, a flow meter is actually measuring zero flow as zero flow. Another example is conductivity monitoring in magnetic-inductive flow meters, for which at least a small flow rate must be present. Electrode noise monitoring is another example for magnetic-inductive flow meters.

[0006] To ensure that diagnostic operation is only performed when the field device is in a suitable state, a corresponding state rule is defined within the field device, and its fulfillment is verified. If the state rule is met, the field device switches from normal operation to diagnostic mode to perform the relevant diagnosis. The state rule depends on at least one state variable of the field device that is known to the field device. In the example of zero-point monitoring, the state variable of the field device is the primary measured variable itself, which in the case of a flow meter is the flow rate. However, other state variables can also play a role, such as a temperature detected by the field device—for example, the temperature of the medium flowing through a flow meter—the rate of change of a measured variable, and so on.

[0007] Using diagnostics triggered by evaluating a state rule can be relatively complex because the relevant state variables must be acquired and evaluated, especially if these variables are not actually of interest during normal operation. If a state variable can no longer be acquired correctly, the field device may no longer be able to enter diagnostic mode independently, which is obviously problematic.

[0008] US 2022 / 229423 A1 discloses a method for verifying the conformity of safety sensor devices with standards using rules, wherein the detection of an anomaly triggers a re-inspection of a safety sensor device.

[0009] The object of the present invention is therefore to further develop the previously described method for operating a field device for process measurement technology in such a way that the aforementioned disadvantages are at least partially avoided.

[0010] The previously derived problem is solved in the method described at the beginning by capturing and storing at least some of the fieldbus messages during a learning phase. The field device is, by definition, connected to at least one other bus participant via a fieldbus. Typically, a large number of field devices are interconnected via a fieldbus, especially those operating within a coherent process. In principle, the fieldbus messages exchanged can be captured by all bus participants, even if the messages are addressed to only one specific bus participant among all those connected. In the aforementioned learning phase, at least some of the fieldbus messages are captured and stored.By capturing and storing at least some of the fieldbus messages during the learning phase, it is possible to look back at the past of fieldbus communication, and past fieldbus messages that were transmitted via the fieldbus are thus available.

[0011] In a number of acquisition steps, in each case where the state rule is fulfilled, the acquired and stored fieldbus messages are at least partially saved as a learning data record, so that a number of such learning data records with fieldbus messages are acquired, whereby the fieldbus messages preceded the fulfillment of the state rule.

[0012] In an evaluation step, a message rule for switching the field device from normal operation to diagnostic mode is derived by evaluating several learning data records. This rule depends on at least one fieldbus message. The fieldbus message does not necessarily have to be addressed to the field device. It must be a fieldbus message that was captured during the learning process and included in the learning data records. The message rule for switching from normal operation to diagnostic mode is generally derived by examining the fieldbus messages of the learning data records for possible commonalities related to triggering diagnostic mode or fulfilling the state rule.

[0013] The field device then applies the message rule, either instead of or in addition to the state rule, to switch from normal operation to diagnostic mode. When the field device evaluates the message rule instead of the state rule, it no longer needs to acquire the state variables otherwise required for evaluating the state rule. This can potentially save on device-related costs, such as the number of sensors needed to acquire state variables or the processing power of the computing unit typically integrated into every field device. This method also prevents the diagnostic mode from being reliably accessed due to incorrectly acquired state variables. Simultaneously executing the state rule and the message rule allows for a plausibility or consistency check of the results from both methods.For example, conclusions can be drawn about incorrect recording of state variables.

[0014] In a preferred embodiment of the method, during the learning phase, the fieldbus messages are acquired by the field device, and the acquisition and evaluation steps are performed on the field device. This variant is designed so that each field device individually replaces the state rule with a corresponding message rule, thus allowing for consideration of the individual characteristics of the field device or its installation situation. In particular, no additional hardware is required. This approach is advantageous for field devices that have powerful hardware and sufficient computing power to execute the method.

[0015] In an alternative embodiment of the method, the fieldbus messages are acquired by a service bus participant during the learning phase, and the acquisition and evaluation steps are performed on the service bus participant. The field device evaluates the state rule and notifies the service bus participant that the rule has been fulfilled. Preferably, this notification is sent via a fieldbus message. This allows the learning phase for acquiring the message rule to be performed on a suitable, sufficiently powerful bus participant with a computing capacity significantly exceeding that of a typical process measurement field device. Preferably, the service bus participant transmits the derived message rule to the corresponding field device, which then applies the message rule.In a preferred embodiment of the method, the message rule is also transmitted to other field devices that are in an identical application situation with an identical state rule.

[0016] Another preferred embodiment of the method involves deriving the message rule from the training datasets based on a regression analysis, with the various bus messages as independent variables and the fulfillment of the state rule as the dependent variable. Regression analysis is suitable for establishing relationships between the dependent and independent variables. The method identifies correlations and models them using the resulting regression relationship. In a preferred embodiment of this method, the bus messages with the smallest partial correlation to the fulfillment of the state rule are successively, i.e., sequentially, excluded until only a predetermined number of independent variables remain in the message rule. This significantly simplifies the message rule.It should be taken into account that correlations do not necessarily reflect causality.

[0017] In a preferred embodiment of the method, random training data sets are stored that are unrelated to the fulfillment of the state rule. These random training data sets are therefore not temporally linked to the fulfillment of the state rule. A message rule is then derived from these random training data sets using a regression analysis. If the message rule derived from the random training data sets is identical to the message rule obtained from the regular training data sets, or if the message rule derived from the random training data sets depends on the same fieldbus messages as independent variables to a predetermined percentage, the message rule derived using the regular training data sets is discarded.This prevents fieldbus messages unrelated to fulfilling the state rule from being incorrectly considered. This can apply, for example, to bus messages that are transmitted periodically, especially at high frequency, on the fieldbus.

[0018] In an alternative refinement of the method, the derivation of the message rule from the training data sets is based on training an artificial neural network, with at least some of the bus messages from the training data sets as input variables and the fulfillment of the state rule as the output variable. It is particularly preferred that the relative time of occurrence of the respective bus message is also used, at least partially, as an input variable. Preferably, the relative time is defined as the interval between the occurrence of the bus message and the time of fulfillment of the state rule. This allows for the consideration of a temporal sequence of specific bus messages.

[0019] A further development of the procedure is characterized by the fact that, during the acquisition steps, fieldbus messages not addressed to the field device itself are stored in a training data set. In a bottling plant, for example, these could be messages sent by filling valves – even those of neighboring filling points. This ensures that the message rule is based on a networked information base that is independent of the specific field device.

[0020] The problem described above is also solved in the filling system with a first filling point and at least one second filling point by, according to one variant, designing the control and evaluation unit of at least the first filling point in such a way that it performs the previously described procedure during operation. According to another variant, a service bus participant is connected via the fieldbus to the first filling valve and the first flow or level sensor of the first filling point, and to the second filling valve and the second flow or level sensor of the second filling point, and the filling system then performs a procedure as described above in connection with the service bus participant.

[0021] The bottling plant is a highly interconnected system with numerous identical subunits, namely the filling stations. This makes the process particularly efficient.

[0022] In detail, there are numerous possibilities for designing and further developing the claimed method and the claimed filling plant. Reference is made, on the one hand, to the claims subordinate to the first claim, and on the other hand, to the following description of exemplary embodiments in conjunction with the drawing. The drawing shows Fig. 1 shows a prior art method for operating a field device for process measurement technology, in which the diagnostic operation is triggered by checking a state rule, and a corresponding filling plant on which the method is operated. Fig. 2 shows a schematic of the inventive method for deriving and applying a message rule, the checking of which triggers the diagnostic operation. Fig. 3 shows a field device arrangement similar to the one described above. Fig. 1 , in which the method for deriving a message rule is executed on a field device, Fig. 4 schematically shows an embodiment in which a service bus participant is additionally provided on which the method for deriving a message rule is executed, Fig. 5 several training data sets with fieldbus messages and the derivation of a message rule and Fig. 6 the training of a neural network with training data sets to obtain the message rule.

[0023] Fig. 1 Figure 1 shows a method 1 known from the prior art for operating a field device F1 for process measurement technology, wherein the field device F1 is in communication connection with at least one other bus participant F2, F3, F4 via a fieldbus 2 for transmitting fieldbus messages FMi. The fieldbus 2 establishes a serial network between the field device F1 and the other bus participants F2, F3, F4, which in the illustrated embodiments are also field devices. With regard to the fieldbus messages FMi, it is initially not essential to distinguish which bus participant F sent the fieldbus message FMi or which bus participant F receives the bus message FMi; in this respect, the designator "FMi" refers to a non-specific fieldbus message or even to the entirety of the fieldbus messages FMi transmitted via the fieldbus 2.

[0024] In this case, the field device F1 is a flow meter. During normal operation fn of the field device F1, it performs its intended task, namely flow measurements in measuring mode.

[0025] In the field device F1, a state rule rx is specified for switching from the normal operation fn of the field device F1 to a diagnostic operation fd of the field device F1, whereby the switch from the normal operation fn to the diagnostic operation fd is made when the state rule rx is fulfilled (rx(x)=! true).

[0026] The diagnostic operation fd is generally used to monitor various functional aspects of the field device F1. The state rule rx depends on at least one state variable x known to the field device F1. This state variable x can be the device's own measured value, but also other parameters relevant to the operation of the field device F1 for diagnostic purposes, such as the device temperature, pressure in the device compartment, rate of change of the measured value, operating time, etc. To verify the state rule rx, i.e., to check whether rx(x) = ? true, one or more state variables x must be acquired, which is computationally intensive. Furthermore, there is a risk that the evaluation of the state rule rx will yield erroneous results if the state variable x is acquired incorrectly.

[0027] In the Figs. 2 to 6Various aspects of the inventive method 1 are now presented, by which a message rule fn based on the evaluation of fieldbus messages FMi is derived, by whose evaluation - depending on the result of the evaluation - the field device can then be switched from normal operation fn to diagnostic operation fd.

[0028] Fig. 2 The procedure is shown in its entirety in Figure 1. In a learning mode, at least some of the fieldbus messages FMi are captured and stored. It is not necessary to capture and store all fieldbus messages FMi; a selection of them can be captured and stored. This ensures that past fieldbus messages FMi can be accessed at any time. For example, a certain number of fieldbus messages FMi could be stored using a ring buffer principle, so that this number of past but most recent fieldbus messages FMi is always known.

[0029] In multiple acquisition steps (5), if the state rule rx is fulfilled, the four fieldbus messages FMi that were acquired and stored are at least partially saved as a learning data record S_FMi, resulting in multiple learning data records S_FMi_j, S_FMi_1, S_FMi_2. This provides various collections of fieldbus messages that describe the message state on fieldbus 2 before the state rule rx was fulfilled.

[0030] In a subsequent evaluation step 6, a message rule rm, dependent on at least one fieldbus message FMi, is derived by evaluating several learning data records S_FMi_j. This rule governs the switch from the normal operation fn of field device F1 to the diagnostic operation fd of field device F1. In general terms, the derivation of the message rule rm is based on examining the fieldbus messages FMi of the learning data records F_FMi_j for commonalities related to triggering the diagnostic operation or fulfilling the state rule rx.

[0031] With the derived message rule rm, it is now possible for the field device F1 to apply the message rule rm instead of, or in addition to, the state rule rx to switch from the normal operation fn of the field device F1 to the diagnostic operation fd of the field device F1. This is in Fig. 2 Shown below. The field device F1 has the following features: Fig. 1The previously used state rule rx has been replaced by the message rule rm, which naturally no longer depends on one or more state variables x, but on fieldbus messages FMi. The fieldbus messages FMi can be all fieldbus messages FMi or only a specific selection of fieldbus messages FMi.

[0032] In Fig. 3 It is shown that in learning mode 3, the fieldbus messages FMi are acquired by the field device F1, and the acquisition steps 5 and the evaluation step 6 are executed on the field device F1. This has the advantage that no additional technical equipment is required to carry out procedure 1.

[0033] The exemplary embodiment according to Fig. 4This shows an alternative approach. In learning mode 3, the fieldbus messages FMi are acquired by a service bus participant FS, and acquisition steps 5 and evaluation step 6 are executed on the service bus participant FS. The service bus participant FS is a more powerful computer than the field device F1, thus providing greater computing capacity for learning mode 3. The field device F1 evaluates the state rule rx and informs the service bus participant FS that the state rule rx is fulfilled (rx = ! true), in this case by means of a fieldbus message. Therefore, the service bus participant FS has all the information necessary to derive the message rule rm. The service bus participant FS then transmits the derived message rule rm to the field device F1, where the message rule rm is applied, either as an alternative or in addition to the state rule rx. The in Fig. 4The state rule rx displayed in the field device F1 is replaced in a subsequent step by the transmitted message rule rm.

[0034] Fig. 5 Figure 5 shows an example of multiple acquisition steps 5 and the evaluation step 6. A total of four learning data records, S_FMi_1, S_FMi_2, S_FMi_3, and S_FMi_4, are displayed. The learning data records S_FMi_j were always created after the state rule rx was fulfilled (rx(x) = ! true). The last four fieldbus messages FMi stored during learning mode 3 were included in the respective learning data records as a rule. The learning data records S_FMi_j can now be examined for commonalities regarding the occurrence of fieldbus messages FMi. The examination for commonalities in the exemplary embodiment according to... Fig. 5This led to the occurrence of fieldbus messages FM2 and FM3 within the last four recorded fieldbus messages FMi being recognized as characteristic for triggering the diagnostic mode fd (measured by the behavior of the state rule rx). The dependency falls under the exemplary embodiment according to Fig. 5 This can be done without further ado. Algorithmically, evaluation step 6, and thus the derivation of the message rule rm from the training data sets S_FMi_j, can be implemented based on a regression analysis, with the various bus messages FMi as independent variables and the fulfillment of the state rule rx as the dependent variable. In this context, it has proven advantageous to sequentially exclude the bus messages FMi with the smallest partial correlation to the fulfillment of the state rule rx until only a predefined number of independent variables remain as part of the message rule rm.

[0035] Fig. 6Figure 1 shows another embodiment for deriving the message rule rm from the acquired training data sets S_FMi_j, which here is based on training an artificial neural network 7. At least some of the bus messages FMi from the training data sets S_FMi_j are used as input variables, and the fulfillment of the state rule rx is used as the output variable. Typically, a portion of the training data sets S_FMi_j is used to test the trained artificial neural network 7. It has proven advantageous to use, at least partially, the relative times FM2_t, FM3_t of the occurrence of the respective bus message FM2, FM3 as input variables, thus also taking into account the temporal relationship between the occurrence of the fieldbus messages FMi. The trained neural network 7 is then transferred to the field device F1 and calculated there in addition to, or instead of, the state rule rx.

[0036] The described method 1 is particularly advantageously applicable in connection with filling plants 8. In the Fig. 1 , 3 and 4The use of the field device F1 and the other bus participants F2, F3, F4 in such a filling plant 8 is therefore schematically illustrated. The field device F1 is here a first flow or level sensor F1, the other bus participant F2 is a first filling valve F2, wherein the first flow or level sensor F1 and the filling valve F2 together with a first control and evaluation unit, which is not explicitly shown, form a first filling point 9, the first control and evaluation unit serving to control and monitor the filling process of the first filling point 9.The bus participant F3 is a second flow or level sensor F3, the further bus participant F4 is a second filling valve F4, wherein the second flow or level sensor F3 and the second filling valve F4 together with a second control and evaluation unit, which is also not explicitly shown, form a second filling point 10, the second control and evaluation unit being used to control and monitor the filling process of the second filling point 10. In the illustrated embodiments, the first control and evaluation unit is encompassed by the first flow or level sensor F1 and the second control and evaluation unit is encompassed by the second flow or level sensor F3.The figures indicate that the first flow or level sensor F1, the second flow or level sensor F3, the first filling valve F2 and the second filling valve F4 are in contact with a process P, in this example with the filling process.

[0037] The first filling valve F2, the first flow or level sensor F1, the second filling valve F4, and the second flow or level sensor F3 are in communication with each other via a fieldbus 2. In the Fig. 1 and 4It can be seen that the first control and evaluation unit is given a state rule rx for switching from normal operation fn of the first flow or level sensor F1 to diagnostic operation fd of the first flow or level sensor F1 when the state rule rx is fulfilled. According to the general doctrine, the state rule rx depends on at least one state variable x of the first flow or level sensor F1 that is known to the first flow or level sensor F1.

[0038] The bottling plants 8 in the Fig. 1 and 3 are characterized by the fact that the control and evaluation unit of the first filling point 9 is designed in such a way that the control and evaluation unit carries out the previously generally described procedure 1 in operation, i.e., in the learning operation 3, it performs a plurality of acquisition steps 5 and the evaluation step 6, and applies the message rule rm derived therefrom.

[0039] The bottling plant 8 in Fig. 4 In contrast, it is characterized by the fact that a service bus participant FS is connected to the first filling valve F2 and the first flow or level sensor F1 of the first filling point 9 and to the second filling valve F4 and the second flow or level sensor F3 of the second filling point 10 via the fieldbus 2 and the filling system 8 carries out the procedure 1 as previously described in general terms in connection with the use of the service bus participant FS. Reference sign

[0040] 1. Procedure 2. Fieldbus 3. Learning mode 4. Acquisition and storage of fieldbus messages 5. Acquisition step 6. Evaluation step 7. Artificial neural network 8. Filling plant 9. First filling station 10. Second filling station F1 Field device, 1. Flow or level sensor F2 Other bus participant, 1. Filling valve F3 Other bus participant, 2. Flow or level sensor F4 Other bus participant, 2. Filling valve FM, FMi Bus messages P Process fn Normal operation fd Diagnostic operation rx State rule rm Message rule x State variable S_FMi Learn-in data record S_FMi_j Learn-in data records FS Service bus participant

Claims

1. Method (1) for operating a field device (F1) for process measurement technology, wherein the field device (F1) is in communication with at least one other bus subscriber (F2, F3, F4) via a field bus (2) for transmitting field bus messages (FMi), wherein at least one status rule (rx) is specified to the field device (F1) for changing from a normal mode (fn) of the field device (F1) to a diagnostic mode (fd) of the field device (F1) when the status rule (rx) is fulfilled, wherein the status rule (rx) depends on at least one status variable (x) of the field device (F1) known to the field device (F1), characterized in that at least some of the fieldbus messages (FMi) are captured and stored (4) in a learning mode (3), that in a plurality of capture steps (5), the captured and stored fieldbus messages (FMi) are at least partially stored as a learning data set (S_FMi_1, S_FMi_2) in each case if the status rule (rx) is fulfilled, so that a plurality of learning data sets (S_FMi_j) are captured, that in an evaluation step (6), a message rule (rm) dependent on at least one fieldbus message (FMi) for changing from normal mode (fn) of the field device (F1) to diagnostic mode (fd) of the field device (F1) is derived by evaluating a plurality of learning data sets (S_FMi_j), and that the field device (F1) applies the message rule (rm) for changing from the normal mode (fn) of the field device (F1) to the diagnostic mode (fd) of the field device (F1) instead of or in addition to the status rule (rx).

2. Method (1) according to claim 1, characterized in that in the learning mode (3) the fieldbus messages (FMi) are captured by the field device (F1) and the capture steps (5) and the evaluation step (6) are performed on the field device (F1).

3. Method (1) according to claim 1, characterized in that in learning mode (3) the field bus messages (FMi) are captured by a service bus subscriber (FS) and the capture steps (5) and the evaluation step (6) are performed on the service bus subscriber (FS), wherein the field device (F1) evaluates the status rule (rx) and communicates the fulfilment (rx=!true) of the status rule (rx) to the service bus subscriber (FS), in particular by means of a field bus message.

4. Method (1) according to claim 3, characterized in that the service bus subscriber (FS) transmits the derived message rule (rm) to the field device (F1) and the message rule (rm) is applied there.

5. Method (1) according to any one of claims 1 to 4, characterized in that the message rule (rm) is derived in the evaluation step (6) from the learning data sets (S_FMi_j) on the basis of a regression analysis, with the various bus messages (FMi) as independent variables and the fulfilment of the status rule (rx) as the dependent variable.

6. Method (1) according to claim 5, characterized in that the bus messages (FMi) with the smallest partial correlation in the fulfilment of the status rule (rx) are sequentially excluded until only a predetermined number of independent variables are still part of the message rule (rm).

7. Method (1) according to any one of claims 1 to 4, characterized in that, in the evaluation step (6), the message rule (rm) is derived from the learning data sets (S_FMi_j) on the basis of training an artificial neural network (7), with at least some of the bus messages (FMi) of the learning data sets (S_FMi_j) as input variables and the fulfillment of the status rule (rx) as output variable, in particular wherein the relative time (FM2_t, FM3_t) of the occurrence of the respective bus message (FM2, FM3) is also used at least partially as input variable.

8. Method (1) according to any one of claims 1 to 7, characterized in that, in the capture steps (5), such field bus messages (FMi) are stored in a learning data set (S_FMi_j) which were not addressed to the field device (F1) itself.

9. Filling system (8) with a first filling point (9) and with at least one second filling point (10), wherein the first filling point (9) has at least one first filling valve (F2), at least one first flow or fill level sensor (F1) and at least one first control and evaluation unit for controlling and monitoring the filling process of the first filling point (9), and wherein the second filling point (10) has at least one second filling valve (F4), at least one second flow or fill level sensor (F3) and at least one second control and evaluation unit for controlling and monitoring the filling process of the second filling point (10), wherein the first filling valve (F2), the first flow or fill level sensor (F1), the second filling valve (F4) and the second flow or fill level sensor (F3) are in communication with each other via a field bus (2), wherein at least the first control and evaluation unit is provided a status rule (rx) for changing from a normal mode (fn) of the first flow or fill level sensor (F1) to a diagnostic mode of the first flow or fill level sensor (F1) when the status rule (rx) is fulfilled, wherein the status rule (rx) depends on at least one status variable (x) of the first flow or fill level sensor (F1) known to the first flow or fill level sensor (F1), characterized in that the control and evaluation unit of at least the first filling point (9) is designed in such a way that the control and evaluation unit performs the method (1) according to any one of claims 1 to 2 or 5 to 8, in this respect referring back to claim 1 or 2, or that a service bus subscriber (FS) is connected to the first filling valve (F2) and the first flow or fill level sensor (F1) of the first filling point (9) and to the second filling valve (F4) and the second flow or fill level sensor (F3) of the second filling point (10) via the field bus (2) and the filling system (8) performs a method according to claims 3 or 4 or according to claims 5 to 8, in this respect referring back to claim 3 or 4.