Searching for possible technical causes or solutions for faults of production units

By using electronic computing units and knowledge models in production units, identifying possible causes of cycle operation forming machine failures, solving the problem that it is difficult to quickly find the cause of failure in the prior art, realizing computer-assisted fault diagnosis and solution provision, improving the The efficiency and accuracy of the production process.

CN119072664APending Publication Date: 2024-12-03ENGEL AUSTRIA
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202280095335.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-04-29
Filing Date
2022-10-14
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

The prior art cannot quickly and reliably find the most likely technical cause or solution for the failure of a particular production unit through computer-aided means, especially in production units that consider multiple objects, especially in cases involving circular operation forming machines. .

Method used

Design a device and computer-implemented method, using electronic computing units, machine interfaces and knowledge models, input observations by operators, identifying objects that may cause failure, and using algorithms to calculate the most likely technical causes or solutions, providing To the operator. The device includes at least one electronic computing unit coupled to the machine interface, accesses the knowledge model, identifying a fault object consistent with the observations, and providing a calculated technical cause or solution through the machine interface.

Benefits of technology

Achieve rapid and reliable failure cause analysis and solution provision based on operator observations, helping operators quickly find the most likely cause or best solution for production unit failure, improving the efficiency and accuracy of the production process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119072664A_ABST
    Figure CN119072664A_ABST
Patent Text Reader

Abstract

An apparatus and a method for finding a possible technical cause or solution of a failure of a particular production unit, which is one of a plurality of production units, and a computer program for configuring such an apparatus or performing such a method are described, the specific production unit is composed of a plurality of different objects and comprises at least one cyclically operated forming machine.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] In one aspect, the invention belongs to the field of finding possible technical causes or solutions for a malfunction of a specific production unit, which is one of a plurality of production units, wherein the specific production unit consists of a plurality of different objects and contains at least one forming machine operating in a cycle.

[0002] One aspect of the invention relates to a device for finding possible technical causes of a malfunction of a specific production unit, which is one of a plurality of production units, wherein the specific production unit consists of a plurality of different objects and contains at least one forming machine operating in a cycle.

[0003] Another aspect of the invention relates to a computer-implemented method for finding possible technical causes or solutions for a failure of a specific production unit, which is one of a plurality of production units, wherein the specific production unit consists of a plurality of different objects and contains at least one forming machine operating in a cycle.

[0004] Yet another aspect of the present invention relates to a computer program which, when executed by a computer, configures the computer as a device for finding possible technical causes or solutions to a failure of a specific production unit, or causes the computer to execute a method for finding possible technical causes or solutions to a failure of a specific production unit.

[0005] A further aspect of the invention relates to a data carrier signal carrying such a computer program.

[0006] US 4,658,370 A teaches an apparatus for constructing and interpreting a knowledge model having separate parts encoding control knowledge, factual knowledge, and judgment rules in the automotive field. An operator can provide values ​​on specific questions to receive assistance for retrieving knowledge.

[0007] EP 3 551 420 BL teaches a method for evaluating and / or visualizing the process state of a production system comprising at least one cyclically operated forming machine, the method comprising the following steps: determining the values ​​of a plurality of selected process variables continuously or in discrete times, and comparing the current value of each selected process variable and / or a variable derived therefrom with one or more reference values ​​by means of a calculation unit, and determining a deviation or a rate of change. Each selected process variable is assigned to at least one logical group by the calculation unit; at least two different logical groups are provided; and for each logical group, the state of the logical group is evaluated by the calculation unit based on the process variables assigned to the logical group and / or visualized by a display device.

[0008] EP 3 754 447 AL teaches a device for monitoring a production facility, wherein the device comprises a computing unit, at least one sensor, a memory unit and an output device. In the memory unit, at least three possible process states are stored in each of at least one set of process variables, and at least one algorithm is stored, by which one of the different possible process states that is actually present can be calculated, the different possible process states with respect to the corresponding set of process variables are classified according to whether a measure is required or recommended, a command of the computing unit prompts it to execute the associated algorithm and thereby calculate which of the different possible process states is actually present, and to check whether the actually present process state is classified as such a process state, so as to generate and output an electronic message according to the calculated process state.

[0009] EP 3 774 267 A1 teaches a method for automatic process monitoring and / or process diagnostics of a part-based process, in particular a production process, in particular an injection molding process, comprising the following steps:

[0010] a) performing an automatic reference lookup to obtain a reference value from a value of at least one process variable;

[0011] b) performing anomaly detection based on the reference value found in step (a);

[0012] c) Performing automatic cause analysis and / or automatic fault diagnosis based on qualitative models of process relationships and / or based on dependencies between various process variables.

[0013] Known techniques do not enable an operator of a production unit to find quickly and reliably in a computer-aided manner possible technical causes or solutions, in particular the most likely technical causes or solutions, for a failure of a specific production unit based on observations made by the operator about the specific production unit, because they have fixed schemes that do not take into account the possible technical causes or solutions of the specific production unit.

[0014] The object of the present invention is to provide a device and a computer-implemented method for quickly and reliably finding possible technical causes, in particular the most likely technical causes, or solutions, in particular the best solutions, for a failure of a specific production unit in a computer-aided manner based on observations made by an operator about the specific production unit.

[0015] Other objects and advantages of the invention will in part be apparent and in part will be obvious from the description and drawings.

[0016] When the term "electronic computing unit" is used in the context of the present disclosure, the "electronic computing unit" describes the smallest entity of a CPU that can read and execute program instructions independently. Each electronic computing unit appears to the operating system as an independent processor that can be addressed in parallel. Each CPU known in the art provides at least one electronic computing unit, but in the context of high-performance computing, modern computer systems typically have more than one electronic computing unit. For example, a CPU may be a multi-core processor with multiple cores. A core is an independent actual processing unit within a CPU that can read and execute program instructions independently of other cores of the CPU. In addition, each core can allow multithreading, that is, one physical core appears to the operating system as multiple processing entities, sometimes referred to as "hardware threads". In other cases, each core of a CPU may be a single processing entity, or the CPU itself may be a single processing entity. In addition, it should be understood that it is assumed that the term CPU should cover GPUs.

[0017] In the present disclosure, the term "machine interface" means any object that allows providing input to a device and / or obtaining output from a device. In particular, the machine interface can be a human-machine interface (in particular for human operators, in short "HMI") or a data bus (in particular for non-human operators such as software agents).

[0018] One object of the present disclosure relates to a device for finding possible technical causes or solutions for a malfunction of a specific production unit, which is one of a plurality of production units, wherein the specific production unit is composed of a plurality of different objects and contains at least one forming machine operating in a cycle, the device comprising:

[0019] - at least one electronic computing unit;

[0020] - at least one machine interface operatively coupled to at least one electronic computing unit;

[0021] at least one knowledge model that provides information for all production units of the plurality of production units, the information specifying, for a plurality of possible failures, which objects present in one or more of the plurality of production units may cause a given failure,

[0022] At least one electronic computing unit is configured to:

[0023] - accepting input from an operator of the particular production unit, the input specifying observations observed by the operator regarding a malfunction of the particular production unit;

[0024] - based on the input, accessing at least one knowledge model and identifying, for those objects forming part of the particular production unit, at least one object that is likely to exhibit a fault consistent with observations of an operator of the particular production unit;

[0025] - using an algorithm to calculate at least the most likely technical cause or solution for a particular combination of observations provided as input by an operator;

[0026] - providing the calculated at least one technical cause or solution to an operator via at least one machine interface.

[0027] Another object of the invention relates to a computer-implemented method for finding possible technical causes or solutions for a malfunction of a specific production unit, which is one of a plurality of production units, wherein the specific production unit comprises a plurality of different objects (i.e., is composed of a plurality of different technical components and is configured to operate in different process steps) and contains at least one cyclically operating forming machine, the method using:

[0028] - at least one electronic computing unit;

[0029] - at least one machine interface operatively coupled to at least one electronic computing unit;

[0030] at least one knowledge model that provides information for all production units of the plurality of production units, the information specifying, for a plurality of possible failures, which objects present in one or more of the plurality of production units may cause a given failure,

[0031] And the method at least comprises the following steps:

[0032] - accepting input from an operator of the particular production unit, the input specifying observations observed by the operator regarding a fault of the particular production unit;

[0033] - based on the input, accessing at least one knowledge model and identifying, for those objects forming part of the particular production unit, at least one object that is likely to exhibit a fault consistent with observations of an operator of the particular production unit;

[0034] - using an algorithm to calculate at least the most likely technical cause or solution for a particular combination of observations provided by an operator as user input;

[0035] - providing the calculated at least one technical cause or solution to an operator via at least one machine interface.

[0036] Another object of the present invention relates to a computer program, which, when executed by a computer, causes the computer to be configured as an apparatus according to one of the embodiments described in the present disclosure or to be configured to execute a method according to one of the embodiments described in the present disclosure.

[0037] A further object of the invention relates to a data signal carrying such a computer program.

[0038] Preferred embodiments of the invention are described in the dependent claims. In order not to repeat text paragraphs, whenever an embodiment of the inventive device is described, this should be understood to refer to the inventive method and vice versa. It should also be understood that the phrase "technical reason or solution" should be interpreted to mean "only technical reason is provided" or "only solution is provided" or "both technical reason and solution are provided".

[0039] In some cases, the "operator" referred to in this disclosure is a person, such as an operator of one or more production units or a service technician sent by the manufacturer of one or more objects of the production unit, in particular a forming machine operating in a cycle.

[0040] In some cases, the term "operator" does not refer to a person, but to a non-human entity, such as a software agent or an observation system, for example as described in EP 3 551 420 B1 or EP 3 754 447 A1.

[0041] A production unit may contain exactly one or more than one forming machines operating in a cycle.

[0042] In addition to the at least one cyclically operating forming machine, the production cell may contain peripheral devices such as robots, thermal control devices for the at least one cyclically operating forming machine, etc.

[0043] There are different possibilities for an algorithm for calculating the at least one technical cause or solution, for example a shortest path algorithm or a maximum flow algorithm.

[0044] Preferred examples of cyclically operated forming machines are injection molding machines or injection presses, in particular for producing plastic parts.

[0045] Observations observed by an operator regarding a malfunction of a specific production unit may refer to observations made by a human operator based solely on human senses without the assistance of a device (e.g., a hose such as oil or noise of unknown origin), and / or it may refer to observations made by a human operator based on readings from at least one sensor or observation system or auxiliary system of the production unit (e.g., sensor values ​​such as regarding temperature values, pressure values, error messages, warnings from a monitoring system, frequent requests for spare parts, etc.), and / or it may refer to observations made by a non-human operator based on signals from at least one sensor or from an observation system.

[0046] Observations made by an operator regarding a malfunction of a particular production unit may refer to the physical condition of an object (e.g., an oil hose) or a product (e.g., a deformed product) being produced by a forming machine operating in a cycle, and / or it may refer to observations made by an operator regarding a production process (e.g., a cycle time too high) or other processes (e.g., too long a time being taken to heat an object).

[0047] The machine interface may include at least one display as an output device. The display may be in the form of a computer monitor, a tablet computer, a cell phone, a device to be worn on the operator's body such as a smart watch or smart glasses, etc.

[0048] The machine interface may include at least one keyboard as input device. In addition or alternatively, the machine interface may include a voice input device and / or a camera device as input device.

[0049] The knowledge model includes:

[0050] - Different objects, such as documents, machine parts, process steps, parameters, solutions, problems, ...

[0051] - Different types of relationships and defined relationships between different objects;

[0052] - A collection of structured data (e.g. mechanical parts, solutions, ...) and / or unstructured data (e.g. PDF-like files).

[0053] In particular, the knowledge model may be at least one database.

[0054] The knowledge model may contain the relationships between observations and different objects (e.g. different machine parts) of a forming machine of multiple production units, production processes and / or cyclic operation. The knowledge model may also contain technical reasons and / or solutions. It may be constructed by the manufacturer based on the knowledge of human experts and / or constructed and / or updated by operators.

[0055] In some embodiments, in addition to at least one knowledge model, a statistical database is used. The statistical database contains the frequency or probability of occurrence of each observation (e.g., global observations for a specific machine type of injection molding machine, local observations for a specific machine type in a production area or for a specific injection molding machine), so that the probability of each possible observation can be calculated. The observations in the knowledge model can have further labels or tags, such as application type, and the machine operator can limit the observations provided by the device by inserting additional labels or tags. The statistical database can be constructed by the manufacturer based on the knowledge of human experts and / or constructed and / or updated by the operator.

[0056] The apparatus may be configured as one physical component comprising at least one electronic computing unit, at least one machine interface and at least one database.

[0057] Alternatively, all or at least some of the above components may be located in different physical devices and may communicate with each other via at least one data channel and / or the cloud.

[0058] In some embodiments, at least one electronic computing unit is configured to accept user input to identify a particular production unit and / or a forming machine for a particular cycle operation. Alternatively, at least one electronic computing unit may be configured to automatically retrieve this information, for example, if a machine interface is mounted to an object of a particular production unit, then at least one electronic computing unit may infer the identity of the particular production unit via the identity of the machine interface used by the operator to contact the device.

[0059] In some embodiments, at least one electronic computing unit is configured to provide a list of possible observations to an operator via at least one machine interface and accept user input that selects an observation from the list of possible observations that best matches an observation observed by the operator regarding a failure of a particular production unit. In this way, the operator can find relevant observations in an assisted manner. Alternatively, the operator can enter the observations in free text form or verbally, and the at least one electronic computing unit can use text or voice recognition to match the entered observations with the list of stored observations.

[0060] In some embodiments, at least one electronic computing unit may be configured to provide a list of objects present in a particular production unit (in this way, the operator can find relevant objects in an assisted manner) and to provide a list of possible observations based on the selection of a particular object made by the operator. Preferably, the list of possible observations is sorted according to frequency of occurrence.

[0061] In some embodiments, at least one electronic computing unit is configured to be operably connected to at least one statistical database, the at least one statistical database containing the frequency or probability of occurrence of each observation result, so as to calculate at least the most likely technical cause or solution of the observation result provided as input by the operator, and provide the calculated at least one technical cause or solution to the operator via at least one machine interface. In some embodiments, only the most likely technical cause or solution can be provided to the operator, however, preferably, the at least one electronic computing unit is also configured to calculate a limited number of possible technical causes or solutions (e.g., two, three or more technical causes or solutions), and provide the calculated possible technical causes or solutions to the operator via at least one machine interface.

[0062] The at least one statistical database can be generated, for example, in different ways:

[0063] - Each time an operator identifies a solution or technical cause of a problem that occurs during the operation of the machine, this information is entered into the database. In addition, in the case where a possible solution or technical cause turns out not to be a valid solution or technical cause at all, this information is also entered into the database. After sufficient information has been accumulated, a statistical analysis is performed and the correlation between the problem and the solution or technical cause is determined and the probability is calculated. Of course, this process must be carried out for all the different objects and their relationships.

[0064] -The probabilities between the problem or technical cause and the solution are defined by experts.

[0065] - ad hoc probabilities (e.g., based on equal distributions) are given and modified by machine learning.

[0066] - A mixture of the above three in equal or weighted fashion.

[0067] In some embodiments, at least one electronic computing unit is configured to calculate the probabilities of possible technical causes or solutions by using the probabilities of each technical cause or solution from a statistical database, and preferably to calculate the total probability of the selected combination of observations and objects by using a shortest path algorithm or a maximum flow algorithm.

[0068] In some embodiments, the at least one electronic computing unit is configured to provide at least one suggestion of a solution based on the calculated at least one technical cause to an operator via the at least one machine interface.

[0069] In these embodiments, at least one electronic calculation unit may be configured to calculate the combined probabilities of different solutions based on the calculated technical causes.For the calculation, statistical data from the statistical database may be combined with an operator's possible up ranking of the selected technical causes.

[0070] In some embodiments, at least one electronic computing unit is configured to be operably connected to a knowledge model that provides possible solutions or technical causes responsible for different observations and / or failures, and the at least one electronic computing unit is also configured to provide an operator via at least one machine interface at least one possible cause or solution of an observation provided by the operator as user input and an object selected by the operator.

[0071] In some embodiments, at least one electronic computing unit is configured to accept operator input specifying whether the proposed solution has worked, and is configured to use the input to update the statistical database.

[0072] In some embodiments, at least one electronic computing unit is configured to accept operator input to:

[0073] -Build and / or update user-specific knowledge models, and / or

[0074] - Build and / or update user specific statistics database.

[0075] In some embodiments, at least one electronic computing unit is configured to:

[0076] - Identify at least two different objects that may exhibit a fault consistent with observations of an operator of a particular production unit;

[0077] - providing the identified at least two objects to an operator via at least one machine interface;

[0078] - accepting operator input selecting one of the provided at least two identified objects;

[0079] - Calculate at least one technical reason or solution for a specific combination of the following:

[0080] Observations provided by the operator as input, and

[0081] Objects selected by the operator.

[0082] Although in the following preferred embodiments of the present invention it is assumed that the cyclically operating forming machine is an injection molding machine, it should be understood that the following description also relates to a cyclically operating forming machine in general.

[0083] First example of the present invention:

[0084] A machine operator of a production unit comprising an injection molding machine detects a loss of oil from the injection molding machine. The machine operator inspects the injection molding machine and detects the hose for the oil. The machine operator does not know why the injection molding machine has lost oil and starts the apparatus of the present invention to investigate the problem in detail. First, using a machine interface of the apparatus, the operator selects a specific production unit and a specific injection molding machine, for example by inserting a product ID of the production unit or by selecting the injection molding machine in a representation of a local machine park at the production site. In a first response, the apparatus may give, via the machine interface, an overview of the currently most likely possible observations for the selected type of injection molding machine, for example, a global observation on all machines of this type or a local observation on machines of this type at the production site.

[0085] The machine operator selects the observation from the list of possible observations that best matches the observation observed by the machine operator regarding the failure of a particular production unit. In this specific example, the observation is "oil" with respect to the component "hose". The machine operator can insert the observation using a text input device such as a keyboard of a machine interface. The device identifies all objects (e.g., hoses, hydraulic blocks, and oil pumps) of a particular production unit that may be the cause of the observation "oil hose" based on information about a particular production unit and / or a particular injection molding machine such as a machine type, and provides them to the machine operator via a machine interface (e.g., as a list). In a specific embodiment of the present invention, the device classifies the provided objects (e.g., 1. hose, 2. oil pump, 3. hydraulic block) according to the frequency (probability) with which each object typically causes the selected observation. In the next step, the machine operator selects a specific object - such as a hose - that may lead to the observation "oil hose".

[0086] Alternatively, it can be provided that the device provides a list of all objects of a production unit or injection molding machine, and the machine operator selects an object of the production unit and obtains all possible observations that may be caused by the selected object, preferably sorted by frequency (probability) of occurrence. In this specific example, the machine operator then selects the observation "Hose for oil".

[0087] For a specific combination of a selected observation and an object (in this example the combination "oil" and "hose"), the ECU of the device calculates a possible predefined number of most likely technical causes that could have led to the observation of the selected object and provides them to the machine operator via the machine interface, for example, as a ranked list according to the probability of each technical cause. In this specific example, the two most likely technical causes are "torn hose" and "leaky pump". Based on this information, the machine operator can check the problem at the injection molding machine. The machine operator checks the hose and concludes that the hose is OK. Therefore, the machine operator selects the second most likely cause "leaky pump".

[0088] In the next interaction step, the device outputs the most likely possible solutions and the technical reasons that may have caused the pump to leak via the machine interface. Since it is in the interest of the machine operator to start production as quickly as possible, the machine operator selects the offered solution "Install new pump". In order to prevent problems with the pump in the future, the machine operator also wants to analyze the reasons why the pump may have failed. The machine operator can perform this investigation independently or can request support from the machine manufacturer. The device therefore offers a further input possibility via the machine interface: "Find root cause" or "Hand over root cause analysis to service personnel". In this specific example, the machine operator selects "Hand over root cause analysis to service personnel".

[0089] The service employee can enter the system at the current input point and continue the root cause analysis. In the specific example, the device outputs the following root causes via the machine interface: "System pressure limit exceeded" and "Pump seal incorrectly assembled". In the next step, the service employee selects a possible root cause and obtains further information about which system parameters to analyze in order to confirm or exclude the root cause. Based on this information, the service employee analyzes the parameters of the hydraulic system and detects a defective pressure relief valve, which may lead to excessive system pressure and thus to pump failure (leakage). In a final step, the service operator replaces the defective valve.

[0090] Second example of the present invention:

[0091] The purpose of the following embodiments is to present the most likely potential technical problems and solutions to the observations made by a machine operator at a production cell.

[0092] The probability may be calculated based on the configuration and type of the production unit and / or in particular the injection molding machine and on the specific history of the production unit and / or in particular the injection molding machine.

[0093] Based on the machine operator inputting the serial number of the production unit via the machine interface, the device requests all relevant possible observations of the production unit from the database (knowledge model). These observations are all possible observations tagged or marked with the specific type of production unit in the database.

[0094] From the set of possible observations, the device requests the probability of each possible observation from a second database (statistical database) containing the frequency of occurrence of each observation, globally for a specific machine type of injection molding machine, locally for a specific machine type in a production area or for a specific injection molding machine. The observations in the knowledge model can have further labels or tags, such as application type, and the machine operator can limit the observations provided by the device by inserting additional labels or tags.

[0095] The knowledge model also contains the relationships between the observations and different objects from the production unit or the production process, such as different machine parts.

[0096] Based on the selected observation, the device requests from the production unit all objects related to the observation from the knowledge model. The device requests from the statistical database the probability that the related objects cause a specific observation and outputs the information to the machine operator via the machine interface (preferably sorted according to their probability).

[0097] The machine operator selects an object, for example a specific object of an injection molding machine. For the selected observation and the selected object, the device calculates the most likely technical cause that may have led to the observation at the selected object.

[0098] To calculate the probabilities of possible technical causes, the device can use the probability of each technical cause from the statistical database and calculate the total probability of the combination of the selected observations and the selected objects. Possible methods for calculating the total probability are the shortest path algorithm or the maximum flow algorithm. The device can further reduce the proposed technical causes, for example, by checking whether some exclusion criteria related to the technical causes (in the knowledge model) are met by comparing the criteria with data from the production process. In addition, the machine operator can impose one or more of the reasons shown.

[0099] In the next step, the device requests possible solutions for the proposed technical cause from the knowledge model and calculates the probability of each solution. To this end, the device uses an algorithm and calculates the combined probability of the solutions based on the technical causes (without the imposed cause and its specific solution). For the calculation, statistical data from the statistical database are combined with the possible upward ranking of the selected technical causes by the machine operator. After the calculation, the device outputs the solutions to the machine operator according to their probabilities. It may also be advantageous to rank the solutions according to other criteria, such as the time or effort to implement the solutions.

[0100] The machine operator can select a solution from the list, obtain detailed information about the solution, and then confirm whether the solution has solved the problem that forms the basis of the observation. The device can then use this information to update a statistical database that is used for future solution searches. To this end, the device can rank the confirmed paths (e.g., observation→cause→solution) up through the knowledge model and downrank the imposed paths. In the next step, the device can scale the probability of each path (observation→cause→solution) in the statistical database so that the sum of outgoing relations of a particular type (e.g., with a cause) from the node is equal to 1.

[0101] By confirming the solution, the device confirms the associated cause and gives the machine operator the possibility to start a root cause analysis for the specific cause. In this case, the device again requests possible causes for this cause (root cause) from the knowledge model and the associated probabilities from the statistical database. In certain cases, the root cause can be found only by multiple (sequential) searches for causes (causes of causes) of a specific confirmed problem (cause of observation).

[0102] In another embodiment, an operator may use predefined relationships between different objects from a production unit or a production process to build a user-defined knowledge model.

[0103] In a simple implementation, the operator can copy the predefined relationships into the user-defined knowledge model. In this case, another user-defined statistical database is initially empty. The operator can then insert other objects of the production unit or production process into the user-defined knowledge model and assign them to other objects available in the database. Additionally, the operator can insert observations, problems and solutions into the user-defined knowledge model and associate them with each other and with objects from the production unit or production process. After the user-defined database is established, the operator can set it as the production database.

[0104] For example, each search for an observation or object then requests results from the manufacturer-provided knowledge model and a new user-defined knowledge model. For each path in the user-defined knowledge model, the device can rank the confirmed paths (e.g., observation→cause→solution) upwards through the knowledge model and can rank the imposed paths in the local statistical database downwards.

[0105] Embodiments of the present invention are discussed using the accompanying drawings, in which:

[0106] Figure 1 An embodiment of the device according to the invention is shown.

[0107] FIG. 2A to FIG. 2E An example of the method of the invention based on an observation ("Hose of Oil") is shown.

[0108] Figure 3 A flow chart of the example of FIG. 2 is shown.

[0109] Figure 4A A possible logical structure of a portion of a knowledge model is shown.

[0110] Figure 4B Shows Figure 4A The knowledge model with added probability in .

[0111] Figure 5 Another example of the method of the present invention is shown.

[0112] Figure 6 Yet another example of the method of the present invention is shown.

[0113] Figure 7 Yet another example of the method of the present invention is shown.

[0114] Figure 1 An arrangement is shown comprising a machine interface and an electronic computing unit ECU (arranged in a single physical part) and a database DB, which in this example comprises a knowledge model and a statistical database and is connected to the electronic computing unit ECU via a data network. The arrangement is connected to a specific production unit PC via a cloud.

[0115] exist FIG. 2A to FIG. 2E The sequence of the inventive method is shown based on the observation "Oil Hose":

[0116] Figure 2A: A machine operator of a production unit comprising an injection molding machine detects a loss of oil of the injection molding machine IMM. Using the machine interface of the device, the operator selects a specific production unit and a specific injection molding machine IMM. The machine operator selects as INPUT the observation from a list of possible observations that best matches the observation observed by the machine operator regarding a malfunction of the specific production unit. In this specific case, the observation is "Hose of oil".

[0117] Figure 2B : Using the database DB, the device identifies as a list all objects COMP 1...COMP N of the specified production unit PC that may be the cause of the observation result "oil hose" (such as hose, hydraulic block and oil pump) based on information about the specific production unit PC and / or the specific injection molding machine IM.

[0118] Figure 2C : The device sorts the provided objects according to frequency (probability), each object COMP 1...COMP I...COMP N usually leads to the selected observation result (for example, 1. hose, 2. oil pump, 3. hydraulic block).

[0119] Figure 2D : The machine operator selects a specific object COMP I - for example a hose - which may result in the observation "Oil hose".

[0120] Figure 2E : The ECU of the device calculates, for a specific combination of a selected observation and object COMP I, the most probable technical causes CAUSE XYZ that may have led to the observation INPUT of the selected object COMP I and provides them to the machine operator via the machine interface.

[0121] Figure 3 A flow chart of the example of Figure 2 is shown. After the "Login" step, the operator selects "Machine 2" as the specific production unit from three possible production units. In this example, the production unit consists of only a single machine.

[0122] Based on the observation of the hose of oil, the operator selects the observation of "oil" from among several possible observations, and selects the object "hose" for which the observation of "oil" is an option from among several objects.

[0123] The device calculates the most likely technical cause and provides two possible technical causes "torn hose" (most likely) and "pump leakage" (second most likely).

[0124] The operator selects "Pump is leaking" and the device displays the most likely solution "Install new pump".

[0125] The device then offers a choice between "Find Root Cause" and "Hand Over to Service." The operator selects "Hand Over to Service," and the device calculates possible root causes and provides two possible root causes to the service employee. After the service employee selects "System Pressure Limit Exceeded," the device shows the parameters and settings to check, allowing the service employee to identify incorrect settings.

[0126] In a final step, the device updates the statistics database.

[0127] Figure 4A shows a possible logical structure of a knowledge model for all production units in a plurality of possible production units (in Figure 4A ) provides information which specifies, for a plurality of possible faults, which objects present in one or more of a plurality of production units may cause a given fault.

[0128] exist Figure 4B In the above example, the probability has been added to Figure 4A , and the algorithm follows the path with the highest probability.

[0129] In another example of the invention, the probability is calculated as a combination of the initially provided expert knowledge and the history of the validation path through the knowledge database. The knowledge database may be composed of Figure 4A The figure shown is used to represent it.

[0130] The history of the validation path through the knowledge database is given by Figure 5 The records shown are given. These can be stored in a statistical database.

[0131] Based on this history, in the first step, the statistical probability P of the historical record is calculated according to the following formula stat

[0132]

[0133] Where N rel is the number of records of the relationship between two specific objects and N path is the number of records of the same path leading to the first object (=body).

[0134] This gives:

[0135] P hose / oily =N hose / oily / N hydraulicSystem / hose =6 / 6=1

[0136] P oily / hoselsTorn =N oily / hoselsTorn / N hydraulicSystem / hose / oily =1 / 6=0,17

[0137] P hoselsTorn / replaceHose =N hoselsTorn / replaceHose / N hydraulicSystem / hose / oily / hoselsTorn =1 / 1=1

[0138] P oily / pumpHasLeakage =N oily / pumpHasLeakage / N hydraulicSystem / hose / oily =5 / 6=0,83

[0139] P pumpHasLeakage / pumpSealingWronglyAss. =N pumpHasLeakage / pumpSealingWronglyAss. /

[0140] N hydraulicSystem / hose / oily / pumpHasLeakage =2 / 5=0,4

[0141] P pumpHasLeakage / systemPressureLimitExceeded =N pumpHasLeakage / systemPressureLimitExceeded /

[0142] N hydraulicSystem / hose / oily / pumpHasLeakage =2 / 5=0,4

[0143] P pumpHasLeakage / installNewPump =1 / 5=0,2

[0144] P systemPressureLimitExceeded / replacePressureReliefValve =1 / 2=0,5

[0145] P systemPressureLimitExceeded / checkParameter2 =1 / 2=0,5.

[0146] In the second step, the statistical probability P stat and the initial probability P defined by expert knowledge expert Combination, thus obtaining the weighted probability P w .

[0147] P w =x×P expert +(1-x)×P stat

[0148] The weighting factor x = f(N path ) depends on the number of paths N path .

[0149] Examples of the function f may be a linear function, a Heaviside function or a tanh.

[0150] Since the number of available historical records is rather low in this example, the weighting factor is chosen high x = 0.9, which means that the expert knowledge has a high weight.

[0151] In the final step, the weighted probability P w Scaling is done so that the sum of the probabilities for each object equals 1.

[0152]

[0153] This produces Table 1:

[0154]

[0155] The resulting probability P for each relationship rel exist Figure 6 As shown in the figure.

[0156] The probabilities can be updated with each new confirmed path or for example every day, after 10 new paths, ... Preferably, the calculated probabilities and intermediate results are stored in a statistical database to produce a small overall computational effort. However, it may also be feasible to calculate the probabilities for each solution finding process online.

[0157] exist Figure 7 In the example, the shortest path algorithm is used to calculate the most likely technical cause. Figure 5 and Figure 6 For clarity, in Figure 7 In the figure, the related objects of the knowledge graph are marked with A to I.

[0158] In the first step, Figure 6 The graph with probabilities shown in is converted into a matrix of distances represented by Table 2, where the distance d is calculated as d=1-P rel .

[0159] Table 2:

[0160] A B C D E F G H I A 0 0.5 0.5 B 0 0.33 0.83 0.83 C 0 0 D 0 0.2 0.8 E 0 F 0 G 0 H 0 I 0

[0161] Based on the matrix of distances, the most likely technical cause can be calculated as the cause with the shortest path from "oil" (A) to each cause. For example, the shortest path can be calculated using the Dijkstra algorithm. The system can then output the direct cause and also the indirect cause to the user as an example.

[0162] Direct cause:

[0163] a. Hose tear (total distance d = 0.5)

[0164] b. Pump leaks (d = 0.5)

[0165] Indirect causes:

[0166] a. Pump seal is incorrectly assembled (d = 0.67)

[0167] b. Exceeding the system pressure limit (d = 1.17)

[0168] Note that the shortest path algorithm produces the most likely causes but does not calculate the total probability of each cause.

[0169] In the same way, the shortest path algorithm can be used to calculate the most likely solution to the "Oil" problem:

[0170] a. Replace the hose (d = 0.5)

[0171] b. Check parameter 2 (d = 1.03)

[0172] c. Install a new pump (d = 1.33)

[0173] d. Replace the pressure relief valve (d = 1.63)

[0174] If some objects in the knowledge graph are not relevant to a specific machine type (eg, failure of a hydraulic pump at an all-electric machine), they may be ignored in step 1 for generating the matrix of distances.

[0175] List of reference numerals:

[0176] PC production cell

[0177] IMM Injection Molding Machine

[0178] HMI Human Machine Interface

[0179] ECU Electronic Computing Unit

[0180] DB database

[0181] COMP Components

Claims

1. A device for finding possible technical causes or solutions for a failure of a specific production unit, the specific production unit being one of a plurality of possible production units, wherein: The specific production unit is composed of a plurality of different objects and contains at least one forming machine operating in a cycle, the device comprising: - at least one electronic computing unit; - at least one machine interface operatively coupled to said at least one electronic computing unit; at least one knowledge model providing information for all production units of the plurality of possible production units, the information specifying, for a plurality of possible failures, which objects present in one or more of the plurality of production units may have caused a given failure, The at least one electronic computing unit is configured to: - accepting input from an operator of said specific production unit, said input specifying observations observed by said operator regarding a malfunction of said specific production unit; - based on said input, accessing said at least one knowledge model and identifying, for those objects forming part of said specific production unit, at least one object that is likely to exhibit a fault consistent with observations of an operator of said specific production unit; - using an algorithm to calculate at least one technical cause or solution to an observation provided as input by said operator; - providing the calculated at least one technical cause or solution to the operator via the at least one machine interface.

2. The device according to the preceding claim, wherein The at least one electronic computing unit is configured to accept input to identify the forming machine of the specific production unit and / or specific cycle operation.

3. A device according to any one of the preceding claims, wherein: The at least one electronic computing unit is configured to provide a list of possible observations to the operator via the at least one machine interface and to accept input selecting an observation from the list of possible observations that best matches an observation observed by the operator regarding a failure of the particular production unit.

4. Device according to the preceding claim, wherein The at least one electronic computing unit is configured to provide a list of objects present in the specific production unit and to provide the list of possible observations based on a selection of a specific object made by the operator, wherein the list of possible observations is preferably ordered according to frequency of occurrence.

5. The device according to at least one of the preceding claims, wherein The at least one electronic computing unit is configured to be operably connected to at least one statistical database, which contains the frequency of occurrence of each observation result, so as to calculate at least the most likely technical cause or solution of the observation result provided as input by the operator, and provide the calculated at least one technical cause or solution to the operator via the at least one machine interface, and wherein, preferably, the at least one electronic computing unit is also configured to calculate a limited number of possible technical causes or solutions, and provide the calculated possible technical causes or solutions to the operator via the at least one machine interface.

6. Device according to the preceding claim, wherein The at least one electronic computing unit is configured to calculate the probability of the possible technical causes or solutions by using the frequency or probability of occurrence of each technical cause or solution from the statistical database, and preferably to calculate the total probability of the combination of selected observations and objects by using a shortest path algorithm or a maximum flow algorithm.

7. The device according to at least one of the preceding claims, wherein The at least one electronic computing unit is configured to provide at least one suggestion of a solution based on the calculated at least one technical cause to the operator via the at least one machine interface.

8. Device according to the preceding claim, wherein The at least one electronic computing unit is configured to calculate combined probabilities or frequencies of different solutions based on the calculated technical reasons.

9. The device according to at least one of the preceding claims, wherein: The at least one electronic computing unit is configured to provide at least one possible cause or solution to an observation provided as input by the operator to the operator via the at least one machine interface.

10. The device according to at least one of claims 7 to 9, wherein: The at least one electronic computing unit is configured to accept input from the operator specifying whether the proposed solution has worked, and to use the input to update the statistical database.

11. The device according to at least one of the preceding claims, wherein The at least one electronic computing unit is configured to accept input from the operator to: Build and / or update specific knowledge models, and / or Constructing and / or updating the statistical database.

12. The device according to at least one of the preceding claims, wherein: The at least one electronic computing unit is configured to: - identifying at least two different objects that may exhibit a fault consistent with observations of an operator of said specific production unit; - providing the identified at least two objects to the operator via the at least one machine interface; - accepting user input of the operator selecting one of the provided at least two identified objects; - At least one technical reason or solution that uses an algorithm to calculate a specific combination of the following: Observations provided by the operator as user input, and • Objects selected by the operator.

13. A computer-implemented method for finding possible technical causes or solutions for a failure of a specific production unit, the specific production unit being one of a plurality of production units, wherein: The specific production unit is composed of a plurality of different objects and contains at least one forming machine operating in a cycle, the method using: - at least one electronic computing unit; - at least one machine interface operatively coupled to said at least one electronic computing unit; at least one knowledge model providing information for all production units of the plurality of production units, the information specifying, for a plurality of possible failures, which objects present in one or more of the plurality of production units may cause a given failure, And the method at least comprises the following steps: - accepting input from an operator of said specific production unit, said input specifying observations observed by said operator regarding a malfunction of said specific production unit; - based on said input, accessing said at least one knowledge model and identifying, for those objects forming part of said specific production unit, at least one object that is likely to exhibit a fault consistent with observations of an operator of said specific production unit; - using an algorithm to calculate at least the most likely technical cause or solution to an observation provided as input by said operator; - providing the calculated at least one technical cause or solution to the operator via the at least one machine interface.

14. The method according to the preceding claim, wherein: The at least one electronic computing unit receives input to identify the specific production unit.

15. The method according to any one of the two preceding claims, wherein: The at least one electronic computing unit provides a list of possible observations to the operator via the at least one machine interface and accepts input selecting an observation from the list of possible observations that best matches an observation observed by the operator regarding a failure of the particular production unit.

16. The method according to the preceding claim, wherein: The at least one electronic computing unit provides a list of objects present in the specific production unit and based on the selection of a specific object made by the operator provides the list of possible observations, wherein the list of possible observations is preferably ordered according to frequency of occurrence.

17. The method according to at least one of claims 13 to 16, wherein: The at least one electronic computing unit is connected to at least one statistical database, which contains the frequency or probability of occurrence of each observation result, so as to calculate at least the most likely technical cause or solution of the observation result provided by the operator as input, and provide the calculated at least one technical cause or solution to the operator via the at least one machine interface, and wherein, preferably, the at least one electronic computing unit also calculates a limited number of possible technical causes or solutions, and provides the calculated possible technical causes or solutions to the operator via the at least one machine interface.

18. The method according to the preceding claim, wherein: The at least one electronic computing unit calculates the probability of the possible technical causes or solutions by using the probability or frequency of each technical cause or solution from the statistical database, and preferably calculates the total probability of the selected combination of observations and objects by using a shortest path algorithm or a maximum flow algorithm.

19. The method according to at least one of claims 13 to 18, wherein: The at least one electronic computing unit provides at least one suggestion for a solution based on the calculated at least one technical cause to the operator via the at least one machine interface.

20. The method according to the preceding claim, wherein: The at least one electronic computing unit calculates the combined probabilities or frequencies of different solutions based on the calculated technical reasons.

21. The method according to at least one of claims 13 to 20, wherein: The at least one electronic computing unit provides at least one possible cause of an observation provided by the operator as user input and an object selected by the operator to the operator via the at least one machine interface.

22. The method according to at least one of claims 13 to 21, wherein: The at least one electronic computing unit accepts input from the operator specifying whether the proposed solution has worked, and uses the input to update the statistical database.

23. The method according to at least one of claims 13 to 22, wherein: The at least one electronic computing unit receives input from the operator to: Build and / or update specific knowledge models, and / or Constructing and / or updating the statistical database.

24. The method according to at least one of claims 13 to 23, wherein: - identifying at least two different objects that may exhibit a fault consistent with said operator's observations of said specific production unit; - providing the identified at least two objects to the operator via the at least one machine interface; - accepting input from the operator selecting one of the at least two identified objects provided; - Calculate at least one technical reason or solution for a specific combination of the following: observations provided as input by the operator, and • Objects selected by the operator.

25. A computer program, when the program is executed by a computer, the computer program causes the computer to be configured as the apparatus according to at least one of claims 1 to 12 or to perform the method according to claims 13 to 24.

26. A data carrier signal carrying a computer program according to the preceding claim.

Citation Information

Patent Citations

  • Method and device for visualizing or evaluating a process state

    EP3551420B1

  • Device and method for visualising or assessing a process state

    EP3754447A1

  • Method for the automatic process monitoring and process diagnosis of a piece-based process (batch production), in particular an injection-molding process, and machine that performs the process or set of machines that performs the process

    EP3774267A1

  • Knowledge engineering tool

    US4658370A