Order-picking system and method for eliminating malfunctions in installations for transporting goods

Abstract

The invention relates to an order-picking system for eliminating malfunctions in installations for transporting goods. The device comprises a computing device comprising a machine learning model, operating data about at least one installation for transporting goods, and a plurality of malfunction correction measures for at least one installation for transporting goods, each malfunction correction measure being assigned to one cause of a malfunction. The computing device receives a message about a malfunction in at least one installation for transporting goods and uses the machine learning model to determine the most probable cause of the malfunction from the message and the operating data, and at least one malfunction correction measure is then selected and output in the form of instructions. The invention also relates to a method and to a computer program product.

Description

[0001] COMMISSIONIERS SYSTEM AND PROCEDURES FOR TROUBLE SHOOTING IN GOODS TRANSPORT SYSTEMS

[0002] The invention relates to a picking system and a method for troubleshooting in goods transport systems.

[0003] As with most systems with mechanical components, malfunctions can occur in goods handling systems, which may include conveyor belts, conveyor rollers, lifting devices, transport vehicles, and other transport equipment for goods in various packaging or even without packaging. Such malfunctions can have various causes. If a malfunction occurs, operating or technical personnel must find and eliminate the cause, which usually requires at least a partial shutdown of the system. A downtime negatively impacts availability and throughput.

[0004] In more complex systems, however, the connection between the cause of the malfunction and the malfunction itself is not always obvious, requiring personnel to be trained for the respective system and, if modifications are made, retrained. A malfunction can also be referred to as an error. A message about a malfunction could then be referred to as an error message.

[0005] Furthermore, the level of knowledge of personnel may vary, resulting in errors in troubleshooting and thus reduced system availability. Since personnel typically have printed manuals or electronic manuals available on a portable device, they have to work on two devices—the manual and the system—while searching for the cause of the fault, which is additionally time-consuming.

[0006] The present invention is therefore based on the object of overcoming the disadvantages of the known prior art solutions and improving the elimination of malfunctions in goods transport systems. A picking system for storing goods and picking goods into orders, comprising a goods transport system and a device for eliminating malfunctions in the goods transport system, as well as a corresponding method, is provided.

[0007] The object of the invention is achieved by the subject matter of the independent claims. Advantageous further developments are found in the dependent claims.

[0008] This task is solved, in particular, by a machine learning model that uses a report about a malfunction in a goods transport system and operating data from the goods transport system to determine the most likely cause of the malfunction. It then selects at least one suitable troubleshooting measure from the available troubleshooting measures, each of which is assigned a cause of the malfunction. The selected measure is displayed in the form of instructions, and feedback information is used to record whether the malfunction has been rectified or not. The machine learning model can, in particular, relate to a model from the field of artificial intelligence (machine learning).

[0009] A goods transport system can have at least one

[0010] - stationary conveyor technology, for example roller conveyors, belt conveyors, vertical conveyors, lifting devices, overhead conveyors, and / or

[0011] - mobile conveyor technology, for example driverless transport vehicles, in particular single-level storage and retrieval machines (shuttles), multi-level storage and retrieval machines (stacker cranes), Automated Guided Vehicles (AGVs), and / or Autonomous Mobile Robots (AMRs), and / or

[0012] - Sorting device for sorting the goods into a sequence, and / or

[0013] - handling device, for example robots for reloading goods, in particular order-picking robots (palletizers) or robots for separating goods from a stack of goods (depalletizers), and / or

[0014] - Storage rack, for example stationary storage rack, mobile storage rack, and / or

[0015] - Loading aids for the transport of goods, for example trays, containers, boxes, hanging bags for the hanging transport of goods on an overhead conveyor, and / or

[0016] - Workstation for processing orders, in particular automatic and / or manual workstation, for example picking station, packing station, and / or order management system for electronically recording orders, and / or data processing device and / or data storage device, for example Warehouse Management System (WMS), Warehouse Control System (WCS).

[0017] In particular, a picking system for storing goods and picking goods to orders may comprise such a goods transport system.

[0018] A stationary conveyor system of the goods transport system can be designed according to an embodiment disclosed in WO 2015 / 051390 A1, which in particular discloses a roller conveyor.

[0019] A stationary conveyor system of the goods transport system can be designed according to an embodiment disclosed in WO 2012 / 106745 A1, which in particular discloses a lifting device for a transport vehicle (single-level storage and retrieval machine).

[0020] A mobile conveyor system of the goods transport system can be designed according to an embodiment disclosed in WO 2016 / 168878 A1, which in particular discloses a single-level storage and retrieval machine, a stationary storage rack and a vertical conveyor for transporting goods.

[0021] A stationary conveyor system of the goods transport system can be designed according to an embodiment disclosed in WO 2020 / 160585 A2, which in particular discloses a suspended conveyor device and a suspended pocket.

[0022] A mobile conveyor system of the goods transport system can be designed according to an embodiment disclosed in WO 2016 / 141395 A1, which in particular discloses a multi-level storage and retrieval machine and a stationary storage rack.

[0023] A sorting device of the goods transport system can be configured according to an embodiment disclosed in WO 2024 / 000003 A1, which in particular discloses a sorting device for sorting goods in a picking system. A handling device of the goods transport system can be configured according to an embodiment disclosed in WO 2018 / 132855 A1, which in particular discloses a picking robot and a picking station for automatically processing orders.

[0024] A workstation for processing orders of the goods transport system can be designed according to an embodiment disclosed in WO 2018 / 006112 A1, which in particular discloses a picking station for manually processing orders.

[0025] The advantage of this approach is the automated detection of the cause of a fault and the guided, thus more efficient, troubleshooting. Since the corrective measures are displayed, even an operator without the necessary prior knowledge or training can resolve the fault.

[0026] Another advantage is that, after the measures to eliminate the fault have been implemented, the automated detection of the proper functioning of the goods transport system can be achieved by evaluating the return flow information provided by the sensors of the goods transport system. The automated detection of the proper functioning of the goods transport system based on the return flow information provided by the sensors can include informing the computing device about the completion of a previously proposed troubleshooting measure. If the return flow information provided by the sensors subsequently returns to a normal range outside of which it was prior to initiating the troubleshooting measure, the computing device can automatically conclude that the proposed troubleshooting measure was successful.The corresponding feedback information can thus be given the status “validated” and used to retrain the machine learning model.

[0027] The machine learning model can be trained using data on faults, causes of faults, and troubleshooting measures as training data. The data on faults, causes of faults, and troubleshooting measures can be linked based on recorded faults and their resolution. The machine learning model can be configured to group multiple faults based on messages received by the computing device, based on the temporal, spatial, or otherwise identified as related occurrence of the faults, and to classify them accordingly into predefined fault groups. The computing device can be configured to consider the classification when determining the most likely cause.

[0028] The machine Lem model can be retrained using the collected feedback information.

[0029] The detection device may be configured to detect the return flow information via sensors and / or receive it as input from an operator.

[0030] The computing device may further be configured to validate the return flow information acquired via sensors and received as input from an operator.

[0031] The computing device can be configured to automatically detect the proper functioning of the goods transport system by evaluating the return flow information provided by the sensors.

[0032] The output device may comprise a display device configured to display the selected troubleshooting measure and / or a sound output device configured to acoustically output the selected troubleshooting measure.

[0033] The device and / or the at least one goods transport system can further comprise at least one receiving device that receives at least one goods transport system or sections thereof. The output device can be configured to display the recording, preferably in combination with the selected troubleshooting measure. The output device can be configured to output the selected troubleshooting measure in the form of instructions, separately displaying individual steps of an instruction to the operator.

[0034] The output device for outputting the selected troubleshooting measure in the form of instructions must be designed to display combinations of graphic representations, for example of the affected section of the goods transport system, with text elements, for example instructions for the operator.

[0035] The output device for outputting the selected troubleshooting measure in the form of an instruction can be configured to provide the operator with feedback options by means of which the operator can select the successful completion of a step of the instruction, the feedback that a suggested step is not possible and / or the selection of an alternative cause of the error.

[0036] Another embodiment is a computer program product with a program comprising software code sections for executing a method for troubleshooting in goods transport systems when the program is executed on a computer. The computer program product can comprise a computer-readable medium on which the software code sections are stored, wherein the program is directly loadable into an internal memory of the computer.

[0037] The method may further comprise training the machine learning model using data about faults, causes of faults, and troubleshooting actions as training data. The data about faults, causes of faults, and troubleshooting actions based on recorded faults and their resolution may be linked.

[0038] The method may further comprise grouping, by the machine learning model, multiple faults based on the temporal, spatial, or otherwise identified as related occurrence of the faults and based on the received messages, and classifying them according to predefined fault groups. The classification may be taken into account when determining the most likely cause. The method may further comprise retraining the machine learning model using the collected feedback information continuously, at specified intervals, or at specified times.

[0039] Collecting the return flow information may include collecting it via sensors and / or receiving it from an operator as input.

[0040] The method may further comprise validating, by the computing device, return flow information acquired via sensors and received as input from an operator.

[0041] The method may also include automated detection of the proper functioning of the goods transport system by evaluating the return flow information acquired by the sensors.

[0042] The selected troubleshooting measure can be output visually via a display device and / or acoustically via a sound output device.

[0043] The method may further comprise recording, by at least one recording device, the at least one goods transport system or sections thereof, wherein the output of the selected troubleshooting measure may comprise outputting the recordings, preferably in combination with the selected troubleshooting measure.

[0044] Outputting the selected troubleshooting measure in the form of instructions may include separately displaying individual steps of an instruction to the operator.

[0045] Outputting the selected troubleshooting measure in the form of instructions can include the combined display of graphical representations, for example, of the affected section of the goods transport system, and text elements, for example, instructions for the operator. Outputting the selected troubleshooting measure in the form of instructions can include providing feedback options to the operator, allowing the operator to confirm the successful completion of a step of the instructions, report that a suggested step is not possible, and / or select an alternative cause of the error.

[0046] The embodiments show possible variants. The invention is not limited to the specifically illustrated variants. Rather, combinations of the individual variants are also possible. In particular, the features of the system can also be implemented in a method, and vice versa.

[0047] For a better understanding of the invention, it is explained in more detail with reference to the following figures. They show, in highly simplified, schematic representations:

[0048] Fig. 1 shows schematically a device according to the invention for troubleshooting in goods transport systems; and

[0049] Fig. 2 shows a schematic flow diagram of a method for troubleshooting in goods transport systems;

[0050] Fig. 3 schematically shows the output of a troubleshooting measure in the form of instructions by an output device.

[0051] By way of introduction, it should be noted that in the variously described embodiments, identical parts are provided with identical reference symbols or component designations. The disclosures contained throughout the description can be applied analogously to identical parts with identical reference symbols or component designations. Furthermore, the position information chosen in the description, such as top, bottom, side, etc., refers to the directly described and illustrated figure, and these position information must be applied analogously to the new position in the event of a change in position.

[0052] Fig. 1 schematically shows at least parts of a device 1 for troubleshooting in goods transport systems 2, wherein Fig. 1 also shows a goods transport system 2 as well as data flow elements and data elements. Fig. 2 schematically shows a flowchart of a method 100 for troubleshooting in goods transport systems 2. The following description explains the invention in both figures. In Fig. 1, a fault 8 can be seen at a location on the goods transport system 2. This can be, for example, damaged goods. The goods can be, for example, a box or other object.

[0053] The device 1 can be integrated into the already existing operating device of the goods transport system 2, or a separate device, such as a desktop computer, a laptop computer or a portable computing system, such as a tablet or a smartphone.

[0054] It should also be noted that a device 1 can be set up and used dedicated to a specific goods transport system 2, but it can also be set up and used for a plurality of similar goods transport systems 2 and also set up and used for a plurality of different goods transport systems 2. If a device 1 is used for multiple goods transport systems 2, the output device 10 must identify, in addition to the information explained below, the specific goods transport system 2 in which the malfunction 8 occurs.

[0055] It is also conceivable for multiple devices 1 to be used simultaneously. For example, a second or additional device 1 can be provided as a replacement if the first device 1 fails. Likewise, for example, when used for multiple goods transport systems 2 at different locations, each fault 8 of each goods transport system 2 can be output at each of these locations. Finally, for example, in the case of portable devices 1 carried by different people, faults 8 can be output on each device 1 and thus to each of these people.

[0056] The device 1 comprises a computing device 3, which comprises, for example, processors such as CPUs and / or GPUs. The method 100 according to the invention can be carried out using the computing device 3.

[0057] The computing device 3 comprises a machine learning model 4, which is trained, for example, using expert input. This can be achieved by classifying errors or error clusters. Troubleshooting measures 6 and possible error causes 7 are linked to the respective error clusters. In this state of the data model, there is then both a basic set of errors or malfunctions 8 or error clusters and a basic set of troubleshooting measures 6, which are linked to one another based on the expert feedback.

[0058] The machine Eem model 4 can also be trained by using data on malfunctions 8, causes 7 for malfunctions 8 and troubleshooting measures 6 as training data, wherein the data on malfunctions 8, causes 7 for malfunctions 8 and troubleshooting measures 6 are linked to one another on the basis of recorded malfunctions 8, for example of identical or similar goods transport systems 2, and their rectification.

[0059] The computing device 3 further comprises operating data 5 of the goods transport system 2. This operating data 5 includes, for example, structural and logical arrangements and dependencies of the individual components within the goods transport system, various status information of components and network addresses of the system components of the goods transport system.

[0060] The following describes specific examples which, however, do not imply exclusive applicability of the subject matter. For example, the operating data 5 can include the structure of the system, the sequence in which the respective components are arranged within the goods transport system 2, and which sensors are to be assigned to the respective components. This makes it possible to trace the possible routes of goods, and this information on the possible sequences, including identification numbers or module numbers of the individual components, enables the assignment of one or more error or malfunction messages 9, which carry information on the module numbers of the components in which the malfunction 8 occurs, to the one or more affected components at which one or more malfunctions 8 are present.

[0061] An example of a specific sequence is the sequence of a roller conveyor to a lifting device, which connects several roller conveyors, each of which, in turn, supplies goods to one or more single-level storage and retrieval machines, also called shuttles. If one or more error messages 9 occur during the transfer of the lifting device, it is clearly expected, based on the sequence of components, that this is not causally related to another error message 9 from the first roller conveyor.

[0062] At the same time, these operating data also indicate that one or more error messages 9 for the single-level stacker crane in the lowest vertical position is highly unlikely to be related to one or more error messages 9 for the single-level stacker crane in the highest vertical position, provided the single-level stacker cranes in the intermediate positions are not affected or there is no direct connection of any kind between them. It can therefore be assumed with a higher probability that the error messages 9 received are due to different faults 8.

[0063] Based on these structural and logical arrangements, critical parts of the goods transport system 2 can also be identified where faults 8 have a particularly large influence on the availability of the system 2. Furthermore, within the framework of the operating data 5, the status information enables a further restriction in the determination of the cause of the fault 7, such that, based on the existing vertical position of a lifting device, a causal connection between an error message 9 for a lifting device in the lowest vertical position is not related to an error message 9 for a roller conveyor in the highest vertical position.

[0064] Together with the temporal information of the occurrence of the error messages 9, time units can also be defined together with the structure of the components and their operating states as part of the operating data 5. Within such time units, a causal relationship between the occurrence of the error messages 9 on specific components can then be recognized as possible or ruled out. Thus, an error message 9 on a lifting device will probably be related to another error message 9 on a lifting device that occurs 24 hours later, provided the lifting device has been idle in the meantime, whereas a relationship can probably be ruled out in the case of continuous operation within these 24 hours. The operating data 5, e.g.Dependencies of states of components of the goods transport system 2 or dependencies of the structural arrangement of components of the goods transport system thus support the machine learning model 4 in the assignment of the faults 8 and specifically in that different faults 8 can also be identified as such so that suitable causes 7 can then be assigned to them and subsequently suitable fault rectification measures 6 can be selected.

[0065] In addition, the computing device 3 comprises a plurality of troubleshooting measures 6 for the goods transport system 2. When used for multiple goods transport systems 2, the computing device 3 can also comprise troubleshooting measures 6 for all goods transport systems 2. In the troubleshooting measures 6 comprised in the computing device 3, each troubleshooting measure 6 is assigned to at least one cause 7 of a fault 8.

[0066] However, various combinations are also possible: One troubleshooting measure 6 may be suitable for eliminating various faults 8 or their causes 7. One fault 8 or its cause 7 can also be eliminated by various troubleshooting measures 6. One fault 8 can have various causes 7. One cause 7 can cause various faults 8. And finally, the occurrence of a combination of several causes 7 can also cause a fault 8 that is not caused by the individual causes 7. Likewise, a combination of troubleshooting measures 6 can be suitable for eliminating a specific fault 8 or its cause.

[0067] An example of a fault 8 is a sensor that is triggered incorrectly or differently, for example, which detects objects on the transport route so that they can be directed along different paths depending on their size.

[0068] A possible cause 7 that could be responsible for this malfunction could be, for example, a defective object during transport that, due to a change in shape, does not trigger a sensor or triggers it differently than an undamaged object. Other possible causes 7 could be a simply incorrectly positioned object, a defective sensor, or even restricted visibility in the case of an optical sensor, for example, due to smoke or steam.

[0069] Appropriate fault-correcting measures 6 would then include removing the defective object, repositioning the object, replacing the sensor and ventilating the system.

[0070] A goods transport system 2 delivers a message 9 when a fault 8 occurs. In the prior art, this is simply output, for example, as an error code on a display of the goods transport system 2, and the goods transport system 2 is stopped.

[0071] According to the invention, this message 9 is now received by the computing device 3 in step 110 of the method 100. The machine learning model 4 then determines the most probable cause 7 for the malfunction 8 in step 120 based on the information contained in the message 9 and the operating data 5 of the goods transport system 2 available to the model.

[0072] The most probable cause 7 can be determined by the machine learning model 4, for example, through grouping 170 and classification 180. The machine learning model 4 groups several faults 8 based on received messages 9 based on the temporal, spatial, or otherwise identified as related occurrence of the faults 8, thus combining them into error clusters and classifying them according to predefined fault groups. This classification can then be taken into account when determining the most probable cause 7.

[0073] If the machine learning model 4 has not yet been trained, this can be carried out in a step 160. In this case, error messages 9 from system 2 are grouped based, for example, on the time interval to incidents or error cases, and known faults 8 are initially classified by experts based on their expert experience. These groupings and classifications can be used as training data or as a reference for the machine learning model 4. In Fig. 2, step 160 is listed before step 110, but it can also be carried out at a different time, which should simply be before step 120. In steps 170 and 180, i.e. the inference, the incidents or error cases can then be classified into different error groups according to the machine learning model 4, i.e. the error cases can be assigned to the then known error groups using the machine learning model 4.Probable error causes 7 can then be identified from the error group, and troubleshooting measures 6 can be suggested for them. The classification, which is made possible by appropriate training in step 160, increases the quality of the error cause determination in step 120. In Fig. 2, steps 170 and 180 are listed directly before step 120, but it can also be executed at a different time, which should only be before step 120 and after step 160. Classification 180 without prior grouping is also conceivable; in this case, the faults 8 or their messages 9 are divided into classes without prior grouping in step 170.

[0074] Independently of this, the machine learning model 4 can be trained in step 160. If necessary, step 160 can be repeated to contribute to an increase in quality using the example of the error cases from step 170. A machine learning model 4 trained on the basis of the operating data 5 of a goods transport system 2 can also be used in another goods transport system 2, whereby the systems can be structured similarly or differently, or a machine learning model 4 can be trained using operating data 5 from multiple goods transport systems 2.

[0075] Temporal relationships can indicate a causal relationship if, for example, if a transport on a conveyor system fails immediately after an expected position sensor is not triggered or if, due to the failure of a sensor within an expected process, further messages are sent from the subsequent steps in a short time period.

[0076] Spatial relationships may indicate that multiple disruptions 8 occur in the same part of a goods transport system 2 and thus also indicate a relationship between the disruptions 8. Other relationships may consist of identical types of goods leading to clusters of disruptions 8 within the goods transport unit 2 or light irradiation at certain times of day leading to glare of sensors at certain times of the year.

[0077] On the basis of the determined most probable cause 7 of the fault 8 and the operating data 5, a suitable fault rectification measure 6 for eliminating the fault 8 is then selected in step 130 by means of the existing assignments of the fault rectification measure 6, the causes 7 and the faults 8 that are present in the computing device 3.

[0078] Device 1 further comprises an output device 10, on which the selected, suitable troubleshooting measure 6 can be output in step 140. The output device 10 can comprise a display device 15, such as a monitor or a display with various symbols, on which the selected troubleshooting measure 6 can be displayed. The output of the troubleshooting measure 6 can be configured in the form of instructions, optionally also with moving images. In Fig. 1, the damaged object of the fault 8 can be seen as an example on the display device 15.

[0079] The display 15 can of course include further information, such as an error log or other further information on the current and / or previous faults 8. With a monitor and corresponding input devices, an operator 14 also has the option of carrying out further configurations, operations and / or interactions with the goods transport system 2.

[0080] The device 1 and / or the goods transport system 2 can also further comprise one or more recording devices 17. In Fig. 1, only one recording device 17 is shown in the goods transport system 2 by way of example. Such recording devices 17 can record the goods transport system 2 or sections thereof in step 210, for example as a video or as a series of images. The output device 10 can then display this video or images as a live stream to enable visual observation. This can occur independently of the fault messages 9. The recording devices 17 can also be used to document the faults 8 and to support fault elimination. The video signal or the images can be displayed in combination with the selected fault rectification measure 6 on the output device 10 or the display device 15 in step 140. In Fig.As an example, in Fig. 1, a symbol for a video overlay can be seen on display device 15 to the right of the damaged object. In Fig. 2, step 210 is listed directly before step 140, but it can also be executed at a different time—even continuously—that should simply be before step 140.

[0081] The output device 10 can, alternatively or in addition to the display device 15, also comprise an audio output device 16, through which the selected troubleshooting measure 6 can be acoustically output. This can be done, for example, as an acoustic instruction, preferably step by step if the troubleshooting measure 6 is more complex.

[0082] The device 1 further comprises a detection device 11, in which return flow information 12 is detected in step 150. The return flow information 12 indicates, for example, that the fault 8 has been remedied.

[0083] The return flow information 12 can be detected via sensors 13 and / or entered by an operator 14. An example of sensor detection here would be that sensors 13, which may also have initially detected fault 8, resulting in message 9, now determine that the fault no longer exists. This is not shown in Fig. 1.

[0084] Sensors 13 that can detect return flow information 12 include, for example, light barriers, time-of-flight sensors, cameras, various integrated sensors for position, speed and acceleration, e.g. in drives, and general position sensors.

[0085] An example of the input of return flow information 12 by an operator 14 can be realized by the operator entering on the recording device 11, which may comprise, for example, a keyboard or similar input device, that the fault 8 has been eliminated and / or that the currently required step of the fault rectification measure 6 has been executed. The path of the return flow information 12 entered by an operator 14 to the computing device 3 is schematically shown in Fig. 1.

[0086] The detection device 11 can also detect both types of return flow information 12, and the computing device 3 can compare them with each other in a step 200, i.e., validate the detected return flow information 12. The return flow information 12 detected by the detection device 11 is preferably forwarded to the computing device 3 for validation. The detection device 11 is therefore preferably connected to the computing device 3 via data technology. A confirmation could then only be available when both types of return flow information 12 indicate a confirmation. In Fig. 2, step 200 is listed after step 150, but it can also be executed at a different time, which should only be after step 150.

[0087] The feedback information 12 may also contain a negative confirmation, ie, the fault 8 was not eliminated by the proposed fault-remediation measure 6, or that the proposed fault-remediation measure 6 is not suitable for eliminating the fault 8. In this case, the computing device 3 can select and output a further fault-remediation measure 6.

[0088] A further development provides for the machine learning model 4 to be retrained using the feedback information 12 thus acquired in step 190. In Fig. 2, step 190 is listed after step 200, but it can also be performed at a different time, which should only be after step 150. Retraining 190 is not linked to validation 200.

[0089] In summary, the solution proposed here enables ongoing fault cases to be reported, the most probable cause 7 to be determined using a machine learning model 4, and a suitable troubleshooting measure 6 to be proposed. Based on the feedback from the activities of plant personnel or sensors 13 contained in the feedback information 12, confirmation or refutation is provided as to whether the troubleshooting measure 6 was successful. Success is clearly identifiable by the fact that the corresponding fault case is no longer actively displayed after the fault has been rectified. For cases in which the troubleshooting measure 6 has not sustainably remedied the cause 7, remedial measures, such as threshold values ​​for the occurrence of a specific fault 8 within a defined period of time, are integrated in order to detect and prevent any incorrect influence on the machine learning model 4.For example, a particular fault 8 can realistically occur 5 times within a period of one hour, and five occurrences within a period of 10 minutes indicate that the cause 7 has not been successfully remedied.

[0090] The machine learning model 4 determines, locally or centrally, the most probable cause 7 of a fault 8 and selects a suitable measure 6 to eliminate the fault 8. The selected measure 6 is interactively suggested to the operator 14 via the output device 10. The operator 14 can acknowledge the elimination via the recording device 11 or make further entries. Additionally or alternatively, the elimination can be recorded using sensor data acquired from sensors 13. Validation of the acknowledgment entered by the operator 14 with the sensor data is also possible. The feedback information 12 thus obtained at the recording device 11 can in turn be made available to the machine learning model 4 as additional training data, thereby further improving the quality of the machine learning model 4 with regard to determining the cause 7 and selecting the measure 6.

[0091] For example, if a specific fault 7 no longer occurs after a specific measure 6 has been carried out, this leads to a better allocation of fault 8, cause 7 and measure 6. The quality of fault rectification in the goods transport system 2 can thus be improved overall.

[0092] Fig. 3 shows, by way of example, the display of a troubleshooting measure 6 in the form of instructions on a display device 15 of an output device 10, which, on the one hand, makes printed instructions or electronic instructions on a separate portable device superfluous, and, on the other hand, also enables unskilled personnel to eliminate the fault themselves. The output of the troubleshooting measure 6 in the form of instructions comprises a combination of graphic elements 18 and text elements 19, wherein the measure in the example shown comprises three steps 21, of which the first step is shown here. The title 20 of the selected troubleshooting measure 6 is also displayed. Furthermore, control elements are displayed on the display device 15, by means of which a system operator 14 can interact with the computing device 3 or navigate through the displayed troubleshooting measure 6.In the example shown here, the displayed troubleshooting measure 6 is titled "Switch - Damaged Load Carriers." This means that damaged load carriers in the switch area were identified as the cause of the current malfunction. As mentioned above, the suggested troubleshooting measure 6 comprises three steps that the system operator 14 must follow sequentially. In the first step shown, the graphic element 18 indicates the affected area of ​​the goods transport system—a switch—while the text element 19 displayed below tells the system operator 14 which specific step they must perform first, namely removing all damaged load carriers in the switch area. Using the control element 23.3 - "Confirm step" - the system operator 14 can, after successfully completing the first step, provide feedback to the computer 3 that the first step has been completed and, via the control panel 22.2, which is represented by an arrow pointing to the right, proceed to the second step. Using control panel 22.1 - an arrow pointing to the left - the system operator 14 goes back one step in the menu navigation. If removing the load carriers in the switch area is not possible or was not successful, the system operator 14 can select control panel 23.1 - "Solution not working" - so that the computing device can output an alternative step if necessary. If this does not lead to the desired result, or if the system operator 14 recognizes that a different cause of the error exists than the one underlying the proposed troubleshooting measure 6, the system operator 14 can also manually select a different cause of the error via control panel 25 - "Change cause of error" - and thus override the cause of the error determined by the computing device 3.Furthermore, the possibility is provided to access the automatically maintained log of all or certain previous actions of processes of the computing device 3 via the control panel 24 - “Data logs”.

[0093] The exemplary embodiments show possible variants. It should be noted at this point that the invention is not limited to the specifically illustrated variants; rather, various combinations of the individual variants are also possible. The scope of protection is determined by the claims. However, the description and the drawings are to be used to interpret the claims. Individual features or combinations of features from the various exemplary embodiments shown and described may represent independent inventive solutions in themselves. The problem underlying the independent inventive solutions can be derived from the description.

[0094] All information on value ranges in this description is to be understood as including any and all sub-ranges thereof, e.g. the information 1 to 10 is to be understood as including all sub-ranges starting from the lower limit of 1 and the upper limit of 10, ie all sub-ranges begin with a lower limit of 1 or greater and end with an upper limit of 10 or less, e.g. 1 to 1.7, or 3.2 to 8.1, or 5.5 to 10.

[0095] For the sake of clarity, it should finally be pointed out that, in order to better understand the structure, some elements have been shown out of scale and / or enlarged and / or reduced in size.

[0096] Reference symbol list

[0097] Device for troubleshooting

[0098] Goods transport system

[0099] computing device

[0100] Machine learning model

[0101] Operating data

[0102] Troubleshooting measures

[0103] Causes

[0104] Disturbances

[0105] Report

[0106] Output device

[0107] Recording device

[0108] Return information

[0109] Sensors

[0110] operator

[0111] Display device

[0112] Sound output device

[0113] Recording device graphic element

[0114] Text element

[0115] Title of the troubleshooting measure

[0116] Troubleshooting steps

[0117] Navigation controls

[0118] Controls for reporting the success of a step

[0119] Control element data logs

[0120] Control element selection of an alternative error cause

[0121] Troubleshooting procedures

[0122] Receiving a message

[0123] Determine the cause

[0124] Selecting a troubleshooting action

[0125] Issue the troubleshooting action

[0126] Collecting return information Training

[0127] Group

[0128] Classify

[0129] Retraining

[0130] Validate

[0131] To record

Claims

Patent claims 1. Order picking system for storing goods and picking goods for orders with a goods transport system (2) and a device (1) for troubleshooting in the goods transport system (2), comprising a computing device (3) with a machine learning model (4); Operating data (5) about at least one goods transport system (2); and a plurality of troubleshooting measures (6) for the at least one goods transport system (2), wherein each of the troubleshooting measures (6) is assigned to at least one cause (7) of a fault (8); wherein the computing device (3) is configured to to receive messages (9) about faults (8) in the at least one goods transport system (2); to determine the most probable cause (7) of the faults (8) from the messages (9) and the operating data (5) using the machine learning model (4); to select, on the basis of the operating data (5) and the determined cause (7), from the fault rectification measures (6), at least one fault rectification measure (6) that is suitable for rectifying the fault (8); an output device (10) that is configured to output the at least one selected fault rectification measure (6) in the form of instructions; and a recording device (11) that is configured to record feedback information (12) relating to a rectification of the fault (8).

2. Order picking system according to claim 1, wherein the machine learning model (4) is trained by using data on malfunctions (8), causes (7) for malfunctions (8) and troubleshooting measures (6) as training data, wherein the data on malfunctions (8), causes (7) for malfunctions (8) and troubleshooting measures (6) are linked to one another on the basis of recorded malfunctions (8) and their resolution.

3. Order picking system according to claim 1 or 2, wherein the machine learning model (4) is configured to group a plurality of faults (8) based on messages (9) received by the computing device (3) on the basis of the temporal, spatial or otherwise identified as related occurrence of the faults (8), and to classify them according to predefined fault groups, and wherein the computing device (3) is configured to take the classification into account when determining the most probable cause (7).

4. Order picking system according to one of claims 1 to 3, wherein the machine learning model (4) is retrained by the acquired return flow information (12).

5. Order picking system according to one of claims 1 to 4, wherein the detection device (11) is configured to detect the return flow information (12) via sensors (13) and / or to receive it as input from an operator (14).

6. Order picking system according to claim 4, wherein the computing device (3) is further configured to validate the return flow information (12) detected via sensors (13) and received as input from an operator (14) with each other.

7. Order-picking system according to one of claims 4 to 6, wherein the computing device (3) is configured to automatically detect the proper functioning of the goods transport system (2) by evaluating the return flow information (12) supplied via the sensors (13).

8. Order picking system according to one of claims 1 to 7, wherein the output device (10) comprises: a display device (15) configured to display the selected troubleshooting measure (6), and / or a sound output device (16) configured to acoustically output the selected troubleshooting measure (6).

9. Order picking system according to one of claims 1 to 8, wherein the device (1) and / or the at least one goods transport system (2) further comprises at least one receiving device (17) which receives at least one goods transport system (2) or sections thereof, and the output device (10) is configured to display the reception, preferably in combination with the selected fault rectification measure (6).

10. Order-picking system according to one of the preceding claims, wherein the output device (10) for outputting the selected troubleshooting measure (6) in the form of instructions is configured to separately display individual steps of an instruction to the operator (14).

11. Order-picking system according to one of the preceding claims, wherein the output device (10) for outputting the selected troubleshooting measure (6) in the form of instructions is configured to display combinations of graphic representations, for example of the affected section of the goods transport system (2), with text elements, for example instructions for action to the operator (14).

12. Order-picking system according to one of the preceding claims, wherein the output device (10) for outputting the selected troubleshooting measure (6) in the form of an instruction is configured to provide the operator (14) with feedback options by means of which the operator (14) can select the successful completion of a step (23.2) of the instruction, the feedback that a suggested step is not possible (23.1) and / or the selection of an alternative error cause (25).

13. Computer-implemented method (100) for troubleshooting in a goods transport system (2) of a picking system for storing goods and picking goods to orders, comprising Receiving (110) messages (9) about faults (8) in at least one goods transport system (2); Determining (120), using a machine learning model (4), from the messages (9) and operating data (5) of the at least one goods transport system (2) the most probable cause (7) for the faults (8); Selection (130), on the basis of the operating data (5) and the determined cause (7), of at least one troubleshooting measure (6) suitable for eliminating the fault (8) from a plurality of troubleshooting measures (6) for the at least one goods transport system (2), wherein each of the troubleshooting measures (6) is assigned at least one cause (7) of a fault (8); Outputting (140), by an output device (10), the selected troubleshooting measure (6) in the form of instructions; and Detecting (150), via a detection device (11), return flow information (12) relating to a rectification of the fault (8).

14. The method (100) according to claim 13, further comprising training (160) the machine learning model (4) by using data about faults (8), causes (7) for faults (8) and fault rectification measures (6) as training data, wherein the data about faults (8), causes (7) for faults (8) and fault rectification measures (6) are linked to one another on the basis of recorded faults (8) and their rectification.

15. The method (100) according to claim 13 or 14, further comprising grouping (170), by the machine learning model (4), a plurality of faults (8) based on the temporal, spatial or otherwise identified as related occurrence of the faults (8) and based on the received messages (9), and classifying (180) according to predefined fault groups, and wherein the classification is taken into account when determining (120) the most probable cause (7).

16. The method (100) according to any one of claims 13 to 15, further comprising retraining (190) the machine learning model (4) using the acquired feedback information (12) in a continuous manner, at specified intervals, or at specified times.

17. The method (100) according to any one of claims 13 to 16, wherein detecting (150) the return flow information comprises detecting via sensors and / or receiving from an operator as input.

18. The method (100) according to any one of claims 13 to 17, further comprising validating (200), by the computing device (3), return flow information (12) detected via sensors (13) and received as input from an operator (14).

19. The method (100) according to one of claims 17 or 18, further comprising the automated detection of the proper functioning of the goods transport system (2) by evaluating the return flow information detected by the sensors.

20. Method (100) according to one of claims 13 to 19, wherein the output (140) of the selected troubleshooting measure (6) is carried out optically via a display device (15) and / or acoustically via a sound output device (16).

21. Method (100) according to one of claims 13 to 20, further comprising recording (210), by at least one recording device (17), of the at least one goods transport system (2) or sections thereof, wherein the outputting (140) of the selected fault rectification measure (6) comprises outputting the recordings, preferably in combination with the selected fault rectification measure (6).

22. Method according to one of the preceding claims 13 to 21, wherein the output of the selected troubleshooting measure (6) in the form of an instruction comprises the separate display of individual steps of an instruction to the operator (14).

23. Method according to one of the preceding claims 13 to 22, wherein the output of the selected troubleshooting measure (6) in the form of instructions comprises the combined display of graphic representations, for example of the affected section of the goods transport system (2), and text elements, for example instructions for action to the operator (14).

24. Method according to one of the preceding claims 13 to 23, wherein the output of the selected troubleshooting measure (6) in the form of an instruction comprises providing feedback options to the operator (14) by means of which the operator (14) can select the successful completion of a step (23.2) of the instruction, the feedback that a suggested step is not possible (23.1) and / or the selection of an alternative error cause (25).

25. A computer program product comprising instructions which, when executed by a computer, cause the computer to carry out the steps of the method (100) according to any one of claims 13 to 25.

26. A computer-readable data carrier on which the computer program product according to claim 25 is stored.

Images