Monitoring method for a production system and corresponding production system

Abstract

The invention relates to a monitoring method for a production system (1), in particular for a coating system for coating components (e.g. motor vehicle body parts) with a coating agent (e.g. paint), comprising the following steps: · defining a desired operating state of the production system (1), in particular by defining a desired paint and a desired paint pressure for painting the motor vehicle body parts; · operating the production system (1) according to the defined operating state; · measuring operating parameters of the production system (1) during the operation of the production system (1) in the defined operating state, in particular by measuring an applied amount of paint; and · monitoring the production system (1) by comparing the determined operating parameters with reference values for the operating parameters, wherein the reference values are defined by the following steps: o defining the reference values for the operating parameters on the basis of the respective operating states of the production system (1) for a plurality of different operating states of the production system (1); o storing the defined reference values in an assignment to the associated operating states in a memory table (5); and o reading the reference values from the memory table (5) on the basis of the currently defined operating state of the production system (1). The invention further relates to a production system (1) for carrying out the monitoring method.

Description

[0001] DESCRIPTION

[0002] Monitoring procedures for a production facility and corresponding production facility

[0003] Technical field of the invention

[0004] The invention relates to a monitoring method for a production plant (e.g., a paint shop). Furthermore, the invention relates to a corresponding production plant with a monitoring device that implements the monitoring method according to the invention.

[0005] Background of the invention

[0006] Modern paint systems for painting automotive body components can operate in various operating states, which are differentiated by various operating variables and are also referred to as categories. For example, one operating state can be defined by the supply of blue paint at a paint pressure of 4.5 bar, while another operating state can be defined by the supply of red paint at a paint pressure of 4.5 bar. The paint color (red or blue) and the paint pressure (4.5 bar) are therefore the variable operating variables that define the respective operating state.

[0007] It is also known to monitor such painting systems during operation by measuring various operating parameters of the painting system and comparing them with specified reference values ​​in order to detect possible malfunctions. For example, the paint quantity can be measured as an operating parameter, which in the aforementioned first operating state with blue paint and a paint pressure of 4.5 bar can be in a range of 110 ml - 120 ml, while the paint quantity in the aforementioned second operating state with red paint and a paint pressure of 4.5 bar can be in a range of 130 ml - 135 ml. The aforementioned limit values ​​(minimum and maximum values) for the paint quantity form reference values ​​for comparison with the respectively measured paint quantity as an operating parameter. If the specified range is exceeded, an error message can then be generated.The problem with this known monitoring method is that a multitude of operating variables can be adjusted and measured during the operation of a paint shop. It is therefore difficult to specify appropriate reference values ​​for the individual operating parameters for the various operating states.

[0008] On the one hand, this often leads to the reference values ​​being set too roughly during operation, as the effort required to precisely adjust the reference values ​​for all operating conditions would be too great. In practice, this leads to potential errors being detected only late.

[0009] On the other hand, determining suitable reference values ​​for the multitude of operating conditions is very time-consuming.

[0010] One approach to solving this problem is the use of machine learning methods, such as those known from DE 10 2019 112 099 B3. In this case, the reference values ​​can be determined automatically using machine learning. However, this approach is also associated with various problems.

[0011] The multitude of different possible operating states can hardly be fully covered in machine learning.

[0012] Furthermore, supervised learning requires the identification of so-called labels that characterize the respective operating state. However, such labels are often not available in practice.

[0013] Furthermore, the neural networks often used in machine learning are not easily readable by humans. Therefore, an operator cannot assess whether a particular operating state is already well represented by the underlying model.

[0014] Finally, with regard to the general technical background of the invention, reference should also be made to DE 10 2009 013 561 A1, DE 10 2015 112 361 A1, WO 2006 / 037137 A1 and WO 84 / 02592 A1.

[0015] Description of the invention

[0016] The invention is therefore based on the object of creating a correspondingly improved monitoring method for a production plant. Furthermore, the invention is based on the object of specifying a production plant that can implement the monitoring method according to the invention.

[0017] The monitoring method according to the invention is generally suitable for a production plant. However, the monitoring method according to the invention is preferably used for a coating plant for coating components (e.g., motor vehicle body components) with a coating agent (e.g., paint).

[0018] In accordance with the prior art described at the outset, the monitoring method according to the invention also initially provides for a desired operating state of the production plant (e.g., painting plant) to be defined. For example, the desired paint (e.g., red paint) and the desired paint pressure (e.g., 4.5 bar) for painting motor vehicle body components can be defined. The desired operating state is then defined by these two operating variables. However, the invention is not limited to these two operating variables (paint color and paint pressure) with regard to the definition of the operating state. Rather, the respective operating state of the production plant can also be determined by other operating variables and, in particular, by a plurality of operating variables, as will be explained in detail later.

[0019] Furthermore, the monitoring method according to the invention, in accordance with the prior art described at the outset, also provides that the production plant is then operated according to the previously defined desired operating state.

[0020] During operation of the production facility according to the previously defined desired operating state, operating parameters of the production facility are then measured. For example, the amount of paint applied at any given time can be measured, as already explained in the description of the state of the art. This is the amount of paint used to paint a motor vehicle body.

[0021] Furthermore, the monitoring method according to the invention, in accordance with the prior art described above, also provides for monitoring the production plant by comparing the determined operating parameters (e.g., applied paint quantity) with reference values ​​(e.g., maximum and minimum values) for the respective operating parameters. Depending on the result of the comparison, an error message can then be generated, for example, if the determined value of the operating parameter exceeds a specified permissible range.

[0022] The invention is distinguished from the prior art described at the outset by the determination of reference values ​​for comparison with the determined operating parameters.

[0023] Thus, the reference values ​​(e.g. maximum and minimum values ​​of the paint quantity) for the operating parameters (e.g. paint quantity) are initially defined for several different operating states of the production plant depending on the respective operating states (e.g. defined by paint color and paint pressure) of the production plant.

[0024] The reference values ​​thus defined are then stored in a memory table (database) in a map to the corresponding operating states of the memory table. The memory table can therefore store the appropriate reference values ​​for monitoring for a variety of possible operating states of the production plant.

[0025] During operation of the production plant, the reference values ​​can then be easily read from the memory table depending on the current operating status of the production plant.

[0026] This allows the appropriate definition of reference values ​​for a wide range of operating conditions of the production plant, thus avoiding the problems mentioned above.

[0027] In a variant of the invention, the reference values ​​for the respective operating state are determined by supervised or unsupervised machine learning from historical measurement data of the operating parameters.

[0028] Alternatively, in a simple embodiment, it is possible for the individual reference values ​​for the respective operating state to be calculated by averaging the respective operating parameter during operation of the production plant in the respective operating state.

[0029] For example, the aforementioned averaging can be performed over a specific operating period or over a specific number of production events, for example, over a specific number of painted vehicle bodies. It should also be noted that the averaging is preferably performed on a rolling basis, in particular over an immediately preceding operating period or over a specific number of immediately preceding events (e.g., a specific number of painted vehicle bodies).

[0030] Furthermore, it is possible to determine the reference values ​​for the respective operating states of the production facility as follows. First, the production facility is operated in the respective operating state. A fault-free operating period or a single fault-free event (e.g., a faultlessly painted vehicle body) is then determined during the operation of the production facility in the respective operating state. The operating parameters (e.g., paint quantity) are then measured during the fault-free operating state or during the fault-free event. The reference values ​​can then be determined depending on the operating parameters measured during the fault-free operating period or during the fault-free event.

[0031] Alternatively, the reference values ​​for the operating parameters can be set manually by an operator.

[0032] It was already mentioned above that the respective operating state of the production plant can be determined, for example, by the paint color and the paint pressure. However, within the scope of the monitoring method according to the invention, the operating states of the production plant can also be determined by various other operating variables of the production plant. Examples include the following operating variables.

[0033] • Type of paint to be applied,

[0034] • Pressure of the paint to be applied,

[0035] • Flow rate of the paint to be applied,

[0036] • Speed ​​of a rotary atomizer for applying the paint,

[0037] • Charging voltage and / or charging current of an electrostatic paint charge,

[0038] • Directing air flow to form a spray jet of the rotary atomizer,

[0039] • Movement path of a robot, in particular painting path of a paint impact point of a path-controlled painting robot,

[0040] • Temperature of the coating agent,

[0041] • Viscosity of the coating agent,

[0042] • Switching times of a coating agent valve. The respective operating state of the production plant can be determined by several operating variables of the production plant, whereby any combination of the aforementioned operating variables is possible to define the respective operating state. The invention is therefore not limited to the example mentioned above, in which the operating state is defined by exactly two operating variables, namely the paint color and the paint pressure. Thus, the operating state of the production plant can be defined by more than two, three, four, five, or six different operating variables.

[0043] As already mentioned above, the operating parameter to be monitored in a painting system can be, for example, the paint quantity. However, within the scope of the monitoring method according to the invention, other operating parameters of the production system can also be measured and monitored. Examples of these operating parameters include the following.

[0044] • Paint flow,

[0045] • Coating agent pressure of the coating agent,

[0046] • Charging voltage and / or charging current of an electrostatic paint charge,

[0047] • Steering air pressure and / or steering air flow to form a spray jet of the paint,

[0048] • Speed ​​of a rotary atomizer for applying the paint,

[0049] • Air pressure of drive air to drive a turbine of a rotary atomizer,

[0050] • Torque and / or electrical currents from drive motors of drive axes of a robot.

[0051] The invention has been described above only with regard to the monitoring of a single operating parameter, namely the amount of paint applied in a painting system. However, the monitoring method according to the invention can also be applied to the monitoring of a variety of operating parameters of the production system.

[0052] It was already mentioned above that the memory table contains the appropriate reference values ​​for the operating parameter to be monitored (e.g. paint quantity) for various operating states (e.g. paint color and paint pressure). The assignment of the operating states to the appropriate operating parameters is defined within the scope of the invention by so-called rules. In the simplest case, there is a target value as a reference value for each operating parameter to be monitored and a permissible tolerance; if exceeded, the situation is classified as faulty. The rules can, however, also be more complex. For defining the aforementioned rules, various options exist within the scope of the invention, which are briefly explained below.

[0053] • Unsupervised learning: This involves taking a rolling average from historical data and using it as a reference value. However, it can also involve a more complex calculation.

[0054] • Supervised Learning: The user defines OK operating states, i.e., operating states that are error-free and in order (e.g., time periods or individual workpieces). Only these are then used to calculate the reference value.

[0055] • Manual input by users / experts: These can be exceptional situations in particular, which can hardly be learned automatically in operation, but which have been determined from experience or in laboratory tests.

[0056] Furthermore, it should be noted that the term "operating state" used within the scope of the invention can also define a respective path section ("brush") of a painting path that is traversed by a paint impingement point of a painting robot. Various operating variables (e.g., paint flow, shaping air flow, rotational speed of a rotary atomizer, charging voltage of an electrostatic paint charge, etc.) can then be specified for each path section, with these operating variables characterizing the operating state in the respective path section on the painting path. The operating variables of the painting system can therefore be dynamically changed from path section to path section as a painting path is traversed. Reference values ​​can then be specifically specified for each path section of the painting path and the associated operating state.

[0057] In addition to the monitoring method described above, the invention also claims protection for a production facility designed to carry out the monitoring method according to the invention. For example, the production facility may be a coating facility for coating components (e.g., motor vehicle body components) with a coating agent (e.g., paint). However, the invention is not limited to a coating facility with regard to the type of production facility.

[0058] The production plant according to the invention initially comprises a plurality of sensors for measuring operating variables (e.g., paint color and paint pressure) and / or operating parameters (e.g., paint quantity) of the production plant during operation. For example, the applied paint quantity can be measured as an operating parameter in a painting plant. Furthermore, the production plant according to the invention, in accordance with conventional production plants, comprises a control system for setting the desired operating state of the production plant and for querying the measured operating parameters (e.g., paint quantity) and / or operating variables (e.g., paint color and paint pressure) from the sensors.

[0059] Furthermore, the production plant according to the invention, in accordance with known production plants, also comprises a monitoring device for monitoring the production plant by comparing the operating parameters of the production plant measured by the sensors with predetermined reference values ​​for the operating parameters. Depending on the comparison, an error message or other reaction can then be triggered.

[0060] The production plant according to the invention is distinguished from the prior art in that the monitoring device has a memory table (database) for storing the reference values ​​for the operating parameters in an assignment to the associated operating states.

[0061] Furthermore, the monitoring device preferably has a control generator for determining the reference values ​​for the operating parameters based on the measured operating parameters. The relationships between the operating variables (e.g., paint color and paint pressure) of the production system for determining the operating status, on the one hand, and the corresponding reference values ​​for the monitored operating parameters (e.g., paint quantity), on the other hand, thus form rules generated by the control generator.

[0062] The control generator can determine the reference values ​​for the respective operating state, for example, through unsupervised or unsupervised machine learning from historical measurement data of the operating parameters.

[0063] Alternatively, the control generator can calculate the reference values ​​as mean values ​​of the measured operating parameters or use the operating parameters during a fault-free operating period or during a fault-free event as reference values.

[0064] Furthermore, the monitoring device can have a rule editor for manual input of the reference values ​​by an operator.

[0065] Other advantageous developments of the invention are characterized in the subclaims or are explained in more detail below together with the description of the preferred embodiment of the invention with reference to the figures.

[0066] Brief description of the drawings

[0067] Figure 1 shows a flow chart to illustrate the monitoring method according to the invention in a painting system for painting motor vehicle body components.

[0068] Figure 2 shows a flow chart to illustrate the determination of the reference values ​​by forming a moving average of the measured operating parameters.

[0069] Figure 3 shows a modification of Figure 1, where the reference values ​​are derived from the measured operating parameters during a fault-free operating period or during a fault-free event.

[0070] Figure 4 shows a schematic representation of a painting system according to the invention.

[0071] Figure 5 shows a schematic representation of a memory table according to the invention for the assignment of the reference values ​​to the various operating states of the painting system.

[0072] Figure 6 shows a schematic representation of a painting track which is divided into track sections, with an operating state being specified and an operating parameter being monitored for each track section.

[0073] Detailed description of the drawings

[0074] In the following, the flow chart shown in Figure 1 is first explained, which illustrates the monitoring method according to the invention using the example of a painting system.

[0075] In a first step S1, a desired operating state of the painting system is selected. In this example, the operating state of the painting system is determined by the fact that red paint is to be applied at a paint pressure of 4.5 bar. However, this is merely a simplified example to facilitate understanding of the invention. In practice, the operating state of the painting system can be determined by a variety of different operating variables. In a second step S2, reference values ​​for operating parameters of the painting system are then read out from a memory table, specifically according to the operating state previously selected in step S1. In this example, the read-out reference value relates to the applied amount of paint, which is read out from the memory table depending on the previously selected operating state.For example, the reference value can represent a maximum value and a minimum value for the amount of paint applied.

[0076] In the next step S3, the painting system is then operated according to the previously selected operating state, ie in this example with red paint with a paint pressure of 4.5 bar.

[0077] During operation of the paint shop in the selected operating mode, operating parameters of the paint shop are then measured in step S4. In this simplified example, the applied paint quantity is measured as the operating parameter.

[0078] In the next step (S5), the paint shop is monitored by comparing the measured operating parameter (paint quantity) with the reference values ​​for this operating parameter. For example, when applying blue paint at a paint pressure of 4.5 bar, the applied paint quantity should be at a reference value of 120 ml, with a tolerance of 5 ml possible, as shown in Fig. 5.

[0079] If the measured operating parameter (e.g., paint quantity) lies outside the permissible range specified by the reference value (e.g., 120 ml with a tolerance of 5 ml), an error signal can be output in step S6. Furthermore, the control system can then initiate appropriate countermeasures.

[0080] The following describes the flow chart shown in Figure 2, which explains the determination of the reference values.

[0081] In a step S1, a desired operating state of the painting system is first selected, for example for painting red paint with a paint pressure of 4.5 bar.

[0082] In the next step (S2), the painting system is then operated in the previously selected operating mode. During operation of the painting system, operating parameters (e.g., the amount of paint applied) are measured in step (S3).

[0083] In a step S4, a moving average of the measured operating parameters (e.g. applied paint quantity) is then determined during operation in the selected operating state.

[0084] In the next step S5, the reference values ​​for the operating parameters are then determined depending on the previously determined moving average of the measured operating parameter (e.g. applied paint quantity).

[0085] The flow chart according to Figure 3 is now described below, which also explains the determination of the reference values ​​according to an alternative to the embodiment according to Figure 2. To avoid repetition, reference is therefore largely made to the above description of Figure 2.

[0086] A special feature here is that a fault-free operating period or a fault-free event (e.g., a fault-free painted vehicle body) is determined during the paint shop's operation in the desired operating state. The reference values ​​for monitoring are then determined based on the operating parameters measured during the fault-free operating period or during the fault-free event.

[0087] Figure 4 shows a schematic representation of a production plant 1 according to the invention, which can be, for example, a painting plant for painting motor vehicle body components, as already mentioned above.

[0088] Production system 1 is initially connected to a controller 2 via an interface. On the one hand, controller 2 can control production system 1 via the interface and set the desired operating state by specifying operating parameters (e.g., paint color and paint pressure). On the other hand, controller 2 can also query operating parameters (e.g., paint quantity) of production system 1 via the interface, which are measured by sensors in production system 1.

[0089] The controller 2 can then initially store the measured operating parameters (e.g., paint quantity) in a database 3 as raw data, together with the specified operating variables (e.g., paint color and paint pressure). The raw data stored in database 3 can then be processed and stored in another database 4 in processed form.

[0090] In addition, a memory table 5 is provided in which reference values ​​(e.g. maximum value and minimum value) are stored for the operating parameters (e.g. paint quantity) for different operating states (e.g. different paint colors and paint pressures) of the production plant 1.

[0091] The reference values ​​(e.g., maximum and minimum paint quantity) for monitoring production plant 1 are stored in memory table 5, each assigned to the respective operating state. Memory table 5 thus contains appropriate reference values ​​for a variety of different operating states.

[0092] The reference values ​​can be generated by an automatic rule generator 6, which contains the processed raw data from the database 4 and then stores the reference values ​​in the memory table 5 in an assignment to the respective operating state.

[0093] Alternatively, the reference values ​​can also be generated by a manual rule editor 7 and stored in the memory table 5, whereby the manual rule editor 7 can communicate with an operator 9 via a human-machine interface 8.

[0094] In addition, a monitoring module 10 is provided, which applies the previously generated rules and can influence the controller 2.

[0095] Furthermore, Figure 5 shows a simplified table for four different operating states, each characterized by different paint colors and different paint pressures, and represented in a separate row. Each row of the table thus corresponds to a respective operating state. Furthermore, the memory table shows a reference value for the monitored operating parameter, namely the paint quantity, for each operating state (i.e., for each row).

[0096] Finally, Figure 6 shows a schematic representation of a painting path 11, which is traversed by a paint impact point of an application device (e.g. rotary atomizer) which is guided by a multi-axis painting robot over the component surface of the component to be painted. The painting path 11 is defined by several path points P1-P8, which are determined according to the geometry of the component to be coated, a process also referred to as teaching. The painting path 11 consists of several consecutive path sections BA1-BA7, whereby the operating state of the painting system can be dynamically adjusted in the individual path sections BA1-BA7 when traversing the painting path 11. In each of the path sections BA1-BA7, various operating variables (e.g. path speed, paint current, charging voltage of an electrostatic paint charge, etc.) can therefore be dynamically adjusted.An operating state is set for each of the track sections BA1-BA7. A reference value for the operating parameter to be monitored can then be read from the memory table for each of the track sections BA1-BA7 and the associated operating state.

[0097] The invention is not limited to the preferred embodiments described above. Rather, a multitude of variants and modifications are possible, which also utilize the inventive concept and fall within the scope of protection. In particular, the invention also claims protection for the subject matter and features of the subclaims, independently of the respective claims referred to, and in particular even without the features of the main claim. The invention thus encompasses various aspects of the invention that enjoy independent protection.

[0098] List of reference symbols

[0099] 1 production facility

[0100] 2 Control with interface

[0101] 3 Database with measured raw data

[0102] 4 Database with processed raw data

[0103] 5 Memory table with operating states and associated reference values

[0104] 6 Automatic Rule Generator

[0105] 7 Manual Rule Editor

[0106] 8 Human-machine interface

[0107] 9 Operator

[0108] 10 Monitoring module

[0109] 11 Painting track

[0110] BA1-BA7 Track sections of the painting track

[0111] P1-P8 path points of the painting path

Claims

CLAIMS 1. Monitoring method for a production plant (1), in particular for a coating plant for coating components with a coating agent, in particular for a painting plant for painting motor vehicle body components with a paint, comprising the following steps: a) determining a desired operating state of the production plant (1), in particular by determining a desired paint and a desired paint pressure for painting the motor vehicle body components, b) operating the production plant (1) according to the specified operating state, c) measuring operating parameters of the production plant (1) during operation of the production plant (1) in the specified operating state, in particular by measuring an applied amount of paint, and d) monitoring the production plant (1) by comparing the determined operating parameters with reference values for the operating parameters,characterized by the following steps for determining the reference values: e) determining the reference values for the operating parameters as a function of the respective operating states of the production plant (1) for several different operating states of the production plant (1), f) storing the determined reference values in an assignment to the associated operating states in a memory table (5), and g) reading the reference values from the memory table (5) as a function of the currently determined operating state of the production plant (1).

2. Monitoring method according to claim 1, characterized in that the individual reference values for the respective operating state are determined by supervised or unsupervised machine learning from historical measurement data of the operating parameters.

3. Monitoring method according to claim 1, characterized in that the individual reference values for the respective operating state are calculated by averaging the respective operating parameter during operation of the production plant (1) in the respective operating state.

4. Monitoring method according to claim 3, characterized in that a) the averaging is carried out over a specific operating period or over a specific number of production events, in particular over a specific number of painted motor vehicle bodies, and / or b) that the averaging is carried out in a sliding manner, in particular over an immediately preceding operating period or over a specific number of immediately preceding events.

5. Monitoring method according to one of the preceding claims, characterized by the following steps for determining the reference values for the respective operating states of the production plant (1): a) operation of the production plant (1) in the respective operating state, b) determination of a fault-free operating period or a fault-free event during the operation of the production plant (1) in the respective operating state, in particular during a fault-free painting of a motor vehicle body, c) measuring the operating parameters during the fault-free operating period or during the fault-free event, d) setting the reference values depending on the measured operating parameters during the fault-free operating period or during the fault-free event.

6. Monitoring method according to one of the preceding claims, characterized in that the reference values for the operating parameters are determined by an operator.

7. Monitoring method according to one of the preceding claims, characterized in that a) the operating states of the production plant (1) are determined by at least one of the following variables: a1) type of paint to be applied, a2) pressure of the paint to be applied, a3) mass flow of the paint to be applied, a4) speed of a rotary atomizer for applying the paint, a5) charging voltage and / or charging current of an electrostatic paint charge, a6) shaping air flow for forming a spray jet of the rotary atomizer, a7) movement path of a robot, in particular painting path of a paint impact point of a path-controlled painting robot, a8) temperature of the coating agent, a9) viscosity of the coating agent, a10) switching times of a coating agent valve, and / or b) that the measured operating parameters of the production plant (1) comprise at least the following variables: b1) amount of paint, b2) coating agent pressure of the coating agent, b3) charging voltage and / or charging current of an electrostatic paint charge, b4) shaping air pressure and / or shaping air flow for forming a spray jet of the paint, b5) speed of a rotary atomizer for applying the paint, b6) air pressure of drive air for driving a turbine of a rotary atomizer, b7) torque and / or electrical currents of drive motors of drive axes of a robot.

8. Monitoring method according to one of the preceding claims, characterized in that a) the production plant (1) is a coating plant with a coating robot which moves an application device along a predetermined coating path (11) over a component surface of a component to be coated, b) the coating path (11) is divided into several successive path sections (BA1-BA7), c) for each of the path sections (BA1-BA7) of the coating path (11) an operating state of the coating plant is defined, wherein the operating state is determined by operating variables of the coating plant, d) for each of the path sections (BA1-BA7) of the coating path (11) a reference value for monitoring an operating parameter of the coating plant is determined, and e) in each of the path sections (BA1-BA7) of the coating path (11) the operating parameter is monitored using the reference value.

9. Production plant (1), in particular a coating plant for coating components, in particular a painting plant for painting motor vehicle body components, comprising a) several sensors for measuring operating parameters of the production plant (1), b) a controller (2) for setting a desired operating state of the production plant (1) and for querying the measured operating parameters from the sensors, and c) a monitoring device (5-10) for monitoring the production plant (1) by comparing the operating parameters of the production plant (1) measured by the sensors with reference values for the operating parameters, characterized in that d) that the monitoring device (5-10) has a memory table (5) for storing the reference values for the operating parameters in an assignment to the associated operating states.

10. Production plant (1) according to claim 9, characterized in that the monitoring device has a control generator (6, 7) for setting the reference values for the operating parameters as a function of the measured operating parameters.

11. Production plant (1) according to claim 10, characterized in that the control generator (6) the individual reference values for the respective operating state are determined by supervised or unsupervised machine learning from historical measurement data of the operating parameters.

12. Production plant (1) according to claim 10, characterized in that a) the control generator (6) calculates the reference values as the mean value of the measured operating parameters, or b) the control generator (6) adopts the measured operating parameters during a fault-free operating period or during a fault-free event as reference values.

13. Production plant (1) according to one of claims 9 to 12, characterized in that the monitoring device has a rule editor (7) for input of the reference values by an operator.

Images