Method for Parameterising at Least One Device

Abstract

In a method for parameterising at least one device (30), at least one environmental value (31) of the device is determined via at least one sensor and/or an automation component. It is checked whether parameters are allocated to the at least one environmental value in a parameter database (32). If parameters are allocated to the at least one environmental value (31) in the parameter database (32), the device (30) is parameterised with parameters from the parameter database (32). If no parameters are allocated to the at least one environmental value (31) in the parameter database (32), the device (30) is parameterised with new parameters. The new parameters are then allocated (54) to the at least one environmental value (31) in the parameter database (32).

Claims

1. Method for parameterising at least one device (30) comprising the following steps: determination (13) of at least one environmental value (31) of the at least one device by means of at least one sensor and/or an automation component, which supplies multiple pieces of secondary data, checking whether parameters are allocated to the at least one environmental value in a parameter database (32), parameterisation (17, 51) of the at least one device (30) with parameters from the parameter database (32), if parameters are allocated to the at least one environmental value (31) in the parameter database (32), parameterisation (17, 51) of the at least one device (30) with new parameters, if no parameters are allocated to the at least one environmental value (31) in the parameter database (32), wherein new parameters are then allocated (54) to the at least one environmental value (31) in the parameter database (32).

2. The method according to claim 1, wherein the at least one device (30) is a production device for a food industry.

3. The method according to claim 1, wherein the at least one device (30) and the at least one sensor and/or the automation components are recognised via identification data, which can be read from a network.

4. The method according to claim 1, wherein the at least one environmental value (31) is selected from a group consisting of temperature, humidity, pressure, flow rate, chemical composition of a gas, brightness, vibration, incline and electromagnetic field strength.

5. The method according to claim 1, wherein several of the at least one environmental values (31) are determined, and it is checked whether parameters are allocated to a combination of all environmental values (31) in the parameter database (32).

6. The method according to claim 1, wherein at least one mathematical function is stored in the parameter database (32), in which function environmental values are used as arguments and parameters are used as function values, wherein the function has degrees of freedom.

7. The method according to claim 1, wherein the parameter database (32) is set up as a one-dimensional look-up table in which parameters are allocated to an environmental value, or is set up as a multi-dimensional look-up table, in which parameters are allocated to a combination of environmental values.

8. The method according to claim 6, wherein, when allocating the parameters, dependencies between environmental values (31) are analysed via Principal Component Analysis, and the parameters are applied to as few environmental values (31) as are determined via the at least one senor and/or the at least one automation component.

9. The method according to claim 1, wherein the allocation of parameters takes place via a learning model (15), which is implemented by a neural network and in which all environmental values (31) are used as variables of an input layer (61) and the parameters arise from an output layer (63).

10. The method according to claim 1, wherein a digital image (16) of the at least one device (30) is created and that the parameterisation (51) of the at least one device (30) with new parameters is carried out, in which process the new parameters are chosen in such a way that an actual state of the at least one device (30) is aligned with a target state of the at least one device (30), as stored in the digital image.

11. The method according to claim 1, wherein the parameter database is filled with parameters during the operation of the at least one device (30) and is then made available to another at least one device (30).

12. The method according to claim 1 including a computer program configured to perform each step of the method.

13. The method according to claim 12 including a machine-readable storage medium, on which the computer program is saved.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Exemplary embodiments of the invention are depicted in the drawings and are more precisely explained in the following description.

[0029] FIG. 1 shows a flow chart of a method according to the prior art.

[0030] FIG. 2 shows a flow chart of a method according to an exemplary embodiment of the invention.

[0031] FIG. 3 schematically shows the course of an exemplary embodiment of the method according to the invention, in a training phase of a device.

[0032] FIG. 4 schematically shows the course of an exemplary embodiment of the method according to the invention, in an operational phase of a device.

[0033] FIG. 5 schematically shows a learning model, which is used in a method according to an exemplary embodiment of the invention.

EXEMPLARY EMBODIMENTS OF THE INVENTION

[0034] FIG. 1 shows a common reaction of a machine to changing environmental conditions, according to the prior art. A change 10 of the environmental conditions leads to parameters of the machine 11 having to be adapted. Subsequently, two reactions are possible:

[0035] On the one hand, a manual change 21 of the parameters of the machine can be carried out by a user.

[0036] On the other hand, it is possible that a new set of parameters for the changed environmental conditions already exists. This is manually downloaded from a parameter database by a user 22 and supplied to the machine.

[0037] Both options lead to a temporary halt of the machine.

[0038] Various exemplary embodiments of the method according to the invention react to the change 10 of the environmental conditions in the way represented in FIG. 2. Here it is also necessary for parameters to be adapted 11. At the same time, it is possible that a model and/or a database of the machine are no longer up to date 12. Using sensors of the device, which are configured as automation components that supply multiple pieces of secondary data, a determination 13 of several environmental values of the device now takes place. A parameter database of the device is now searched through for a parameter set, which is allocated to the determined environmental values. If such a parameter set is not found, then the necessary parameters are determined via a dynamic parameter database 14, a learning model 15, or by dynamic adaptation of a digital image 16 in real time. Then a parameterisation 17 of the device is carried out, in which the new parameter set is made available to the device. Following this, the device continues operating 18 with the new parameters. This is possible without halting the device.

[0039] In an exemplary embodiment of the invention, the device is a production device for frozen fish. If the temperature changes as an environmental value, this affects the sliding behaviour of the frozen fish on a conveyor belt of the device. When introducing the frozen fish into a packing tube, the frozen fish can come to a standstill more quickly than usual. Consequently, a sealing tool does not hit the gap between two frozen fish, but rather hits a frozen fish in the packaging. This leads to the machine halting, waste, and possibly to the sealing tool being damaged. Conventionally, the cause of the error is investigated by the machine operator. The force of a pusher, which moves the frozen fish onto the conveyor belt, is in this case increased as a parameter. If this incident occurs in the method according to the invention, the problematic environmental value of the temperature is linked with the changed parameter of the force of the pusher in the parameter database. In the event of a reoccurrence of such a temperature, this parameter is automatically applied, so that future manual interventions can be avoided.

[0040] In another exemplary embodiment of the method according to the invention, the device is a moveable device in the form of a refuse truck. An environmental value determined in the method is also the temperature in this exemplary embodiment. Heating by solar radiation or cooling by weather conditions can lead to changes of the viscosity of a hydraulic oil in a hydraulic system of the refuse truck, as well as to elongation of mechanical components such as axles or rods. This leads to certain areas of the refuse truck not being reached by the rubbish compactor as desired. Instead, movements are too short or too long. In addition, there is deviation from the chronological sequences of the start-up profile. Based on the sensor readings of the temperature, and with the help of the further steps of the method according to this invention, the system pressure and volume flow can, as parameters of the hydraulic drive, be changed, so that endpoints are again approached with the desired accuracy.

[0041] In a further exemplary embodiment of the method, the device is a system for inserting carbon dioxide into food packaging. The volume of the gas depends on the temperature and the pressure of the gas. In pipe networks or industrial processes, these two variables are, however, not constant. The method according to the invention makes it possible to continuously adapt control parameters of the device on the basis of these environmental values, in order to enable an interference-free operation of the system and introduce the necessary amount of gas.

[0042] If a learning process 15 is used in the method according to the invention, a training phase of the device can first be provided, before this is employed in an operational phase. The process of the training phase is represented in FIG. 3. The device 30 is influenced by external environmental values 31. If there are no parameters yet known for the operation of the device 30 under the existing environmental values 31, then a user 40 carries out an adaptation 50 of the parameters in a parameter database 32. Using these, a parameterisation 51 of the device 30 is then carried out. A read-out 52 of these parameters takes place, in order to make them available to a probabilistic model 33. The environmental values 31 are also made available to said model, as sensor readings 53 of the device 30.

[0043] After completion of the training phase an operational phase is carried out, in accordance with FIG. 4. The device 30 continuously determines environmental values 31 via sensors. If there are already parameters stored in the parameter database 32 which are allocated to these environmental values 31 or which lie in the tolerance range around these environmental values 31, then the device 30 is parameterised 51 via these parameters. Otherwise, the environmental values 31 are transferred into the probabilistic model 33 as sensor readings 53, said model adapting the parameters for the existing environmental values 31 and making them available 54 to the parameter database 32. Then, a parameterisation 51 of the device 30 with these altered parameters takes place.

[0044] The probabilistic model 33 is a learning model that is implemented by a neural network, and whose structure is outlined in FIG. 5. It has an input layer 61, hidden layers 62 and an output layer 63. Each of the layers 61, 62, 63 consists of nodes 60. The environmental values 31, which are provided as sensor readings 53, are provided to the input layer 61, wherein each node 60 of the input layer 61 corresponds to an environmental value 31. This is shown for three environmental values 31. In each hidden layer 62, in all nodes of a layer an association of values with all values of the input layer 61 or with all values of the preceding hidden layer 62 takes place. Through a weighted summation of the environmental values 31, parameters for the device 30 are produced in the output layer 63. Alternatively to the weighting, non-linear functions can be used. This is shown with three nodes in the output layer 63, for three parameters.

[0045] In another exemplary embodiment of the method according to the invention, a deterministic model 14 is used to parameterise 51 the device 30, if there are no parameters yet associated with the current environmental values 31 stored in the parameter database 32. The method sequence here follows the depiction in FIG. 4, wherein the model 33 is not a probabilistic model, but rather the deterministic model.

[0046] In yet another exemplary embodiment of the method according to the invention, a digital image 16 of the device 30 is dynamically adapted in real time, in order to generate parameters for the device 30, should these not be present in the parameter database 32 for the current environmental values 31. In this digital image, the determined environmental values 31, the parameters of the device 30 and all automation values of the device 30, which suggest how the current parameters under the current environmental values 31 affect the operation of the device 30, are collected as fine-grained data. Further data for the description of the device 30 can be collected as coarse-grained data, in order to filter out important data for the generation of new parameters from the dataset available via the device 30.

[0047] A parameter database 32 filled with data and optionally the model 33 linked with it can, in all exemplary embodiments of the method of a device 30, be transferred to another device 30. To this end, a software agent recognises the device 30 and its sensors by means of identification data, with which it is also recognised what environmental values 31 can be determined during the operation of the device 30. The identification data can here be designed by the software agent, through a standardised protocol for the exchange of data via a network connection. In one exemplary embodiment of the method, this protocol can be OPC UA (Open Platform Communications Unified Architecture). The new device 30 to be put into operation is then provided with a pre-created parameter database 32 with an appropriate combination of environmental values 31 and parameters as well as, optionally, a model 33.