METHOD FOR COMPUTER-ASSISTED PREDICTION OF FUTURE OPERATING CONDITIONS OF MACHINE COMPONENTS

Description

[0001] The invention relates to a method with which future states, such as an impending damage event, a maintenance requirement, reaching the maximum service life and the like, can be predicted with computer assistance.

[0002] The term "machine component" is used below to refer to individual machine elements or groups of interacting machine elements combined into specific assemblies. Each of these machine elements represents the smallest technical functional unit capable of fulfilling the technical and functional requirements arising from the respective technical application under consideration. Accordingly, the term "machine component" also encompasses the individual parts used in the manufacture of machine components or their machine elements.

[0003] In many applications, machine components of the type discussed here are subjected to high loads of various kinds during practical use. These loads can result from forces absorbed by the respective component or from environmental conditions (e.g., ambient temperature, atmosphere) under which the machine component operates.Machine components subjected to such loads include, in particular, moving parts such as wheels, rollers, cylinders, shafts, rings, chains, belts such as toothed belts, V-belts, flat belts or multi-V-belts, which are used, for example, in drives, discs or disc profiles, seals, gears, springs, as well as hoses, or assemblies that are themselves moved under load or whose components move under load during operation, such as gears, transmissions, drive units in general, especially belt drives such as toothed belt drives, motors, couplings in general, especially slip clutches, clamping sets / shaft-hub connections, freewheels, rolling bearings, plain bearings, brakes, drive trains and the like.

[0004] Each of these machine components is typically essential to the functionality of the machine in which it is used. Consequently, operational disruptions regularly occur when a particular machine component malfunctions or even fails. The economic losses caused by such disruptions can be considerable, especially in large-scale industrial manufacturing. Therefore, machine components are usually subjected to regular maintenance and are replaced or repaired in a timely manner to minimize the risk of spontaneous failure. Maintenance work generally follows a maintenance plan established by the manufacturer of the machine or machine component, taking into account empirical data, design criteria, and operating conditions.

[0005] Furthermore, for particularly critical machine components, it is common practice to systematically monitor certain parameters, such as temperature development, loads, and the like, to prevent overloading of the component and premature failure. If a monitored parameter reaches a specific limit, this is typically considered a warning signal, triggering an action. This action might, for example, involve reducing the load that caused the limit to be exceeded, thus bringing the monitored parameter back within its normal range, or it might involve performing maintenance on the machine component to prevent wear-related failure and ensure its continued operation.The limit values, the reaching of which triggers the respective action, are also generally based on empirical values, which are either communicated to the machine operator by the machine manufacturer or found by the machine operator based on practical experience in use.

[0006] The problem is that both the acquisition of the parameter(s) essential for the operating state of the respective machine component and the determination of the respective limit value always require an interpretation of measured values ​​and operating conditions. The result of this interpretation depends either on the experience of the machine operator or the machine driver, i.e., the person who operates and monitors the machine, or on the quality of the empirical data and the interpretation model that the manufacturer of the respective machine component or the machine equipped with it uses as the basis for the recommendations for action that they provide in the event that the monitored parameter reaches the limit value considered critical.

[0007] "An Intelligent Predictive Engine for Milling Machine Prognostic Monitoring" (XIANG Li et al., IEEE INTERNATIONAL CONFERENCE ON INDUSTRIAL INFORMATICS, IEEE, PI, August 1, 2006 (2006-08-01), pages 1075-1080, ISBN: 978-0-7803-9700-2) presents an intelligent predictive engine (IPE) for applications in the prognostic monitoring and failure prediction of machines. The IPE was designed and developed using embedded data processing and analysis tools based on multiple regression models and artificial neural networks. A case study on predicting the remaining service life of milling machines is presented to demonstrate the applicability of the IPE in the tooling industry. The case study compares the predictive performance of different models built with the same set of experimental data.The neural backpropagation network demonstrates significantly better performance than other models for solving the prognostic problem of predicting tool life on a milling machine. The algorithms in the IPE are generic and can be used for various application scenarios requiring predictive equipment analysis.

[0008] Against the background of the above-described procedure, which is common in operational practice, the task arose to create a method with which an upcoming event in the life cycle of a machine component can be predicted with a high degree of certainty and reproducibility.

[0009] The invention has solved this problem by the method specified in claim 1.

[0010] Advantageous embodiments of the invention are specified in the dependent claims and are explained in detail below.

[0011] The inventive method for computer-aided prediction of future operating states of machine components therefore comprises the following steps: A) For each machine component under consideration, at least one parameter is selected that is representative of the machine component's condition. B) The parameter is recorded as a measured value during the machine component's operation. C) Taking into account the application in which the machine component is used, the recorded measured value is fed into an automated process for generating a recommendation for action, based on the following procedure: Ca) For each machine component under consideration, properties and the influencing factors that determine these properties have been determined. C.b) For each machine component under consideration, the relevant applications are determined, the influencing factors to which the machine component is exposed in each application are identified, the influencing factors are formulated, and it is determined which influencing factors must be considered when interpreting the recorded measurement and the resulting recommendations for action. cc) Based on statements regarding the influencing factors obtained during the practical use of the machine component under consideration, or based on statements related to the influencing factors obtained through systematic experimental investigations, the interactions and effects of the influencing factors are determined and related to specific damage events.If, during the ongoing use of the procedure, it becomes apparent that such previously unidentified influencing factors and effects exist, these newly identified influencing factors or effects are added to the group of factors to be considered and interact with other influencing factors or related damage events. If the product- or application-specific interactions can be described formulaically by a rule, this rule is formulated as soon as the interactions become known. Alternatively or additionally, the interactions can also be captured in the form of fuzzy descriptions ("fuzzy logic," artificial neural networks, other artificial intelligence methods) of the relationships. Cd) The in step C.c) Information (influencing factors, rules, and vague descriptions of interactions) determined regarding the interactions and the resulting damage events is fed into a database as machine-readable data. This database can comprise one, two, or more individual databases, each containing machine-readable data related to specific use cases and the associated damage events. Information determined regarding mutually influencing requirements and influencing factors is optionally linked in the database in such a way that changes to one piece of information have a direct impact on the other piece(s) linked to it. In this way, the database creates a network of interrelated information linked to specific machine components, applicable to a multitude of different applications and the associated damage events.e) The respective application, the recorded measurement value, and the associated influencing factors are fed as input variables into a selection algorithm running on a computer. This algorithm, taking into account the input variables and the recorded measurement value, uses the information stored in the database to select the most suitable recommendation for action for the respective application. Thus, the recorded measurement value is interpreted based on the information stored in the database, which is linked to each other and based on practical experience. This results in a recommendation for action that considers the specific characteristics of the respective application. The influencing factors relevant to the application in question have been formulated for the derivation of an action recommendation specific to the application and taking the recorded measurement value into account (step Cb)).

[0012] To continuously improve the database created by work steps Ca) - Cd) for the selection step (work step Ce)), the following additional work steps are completed: Cf) The machine component for which the respective measured value is recorded and interpreted in the manner specified in step Ce) is uniquely identified with regard to the application in which it is used. Cg) For this purpose, the selected machine component is identified and monitored in its practical application using the identification means. The identification means can not only carry information that uniquely identifies the respective machine component, but also information about the characteristics of the machine component, in order to, for example, easily assign the measured values ​​obtained through monitoring to a specific combination of properties of the machine component under consideration.In its simplest form, monitoring provides not only the recorded measurement but also information on at least one other parameter, such as the time elapsed between the start of operation and the recording of the respective measurement. Additionally, monitoring also provides information on two or more further parameters relating to external influences (temperature, atmosphere) to which the machine component was exposed during operation. Furthermore, the identification device itself can be equipped to provide information about the internal condition and the stresses on the machine element during operation. Of course, other known measurement methods can also be used to record the forces acting on the machine element during operation, its changes in length, temperature development, and the like (see...).For the example of a belt drive, see US 6,264,577 B1, DE 10 2010 002 551A1, JP 2009-007078 A, DE 44 44 263 C1, US 5,843,258 A, AU 2009203049 A1). Ch) The statements recorded in step Cg) are fed into the process beginning with step Cc) as influencing factors.

[0013] With the additional work steps Cf) to Ch), a self-learning selection system is created that continuously improves and increasingly reflects practical reality. Of course, not every machine component selected according to the invention needs to undergo the additional work steps Cf) to Ch); it may be sufficient if at least one machine component, or ideally a representative number of machine components, is monitored in the respective application.

[0014] The invention is therefore based on the same principles as the method for selecting a machine component presented in European patent application EP 3 292 520 A1.

[0015] The invention extends the aforementioned method described in European patent application EP 3 292 520 A1 by adding the possibility of determining a recommendation for action based on a measured value recorded for a specific machine component, taking into account the influencing factors that are typical for the respective application and systematically linked.

[0016] The method according to the invention thus enables, starting from the respective measured value and on the basis of the influencing factors and their relationships determined for the respective application and identified through systematic test series or the recording of practical experience, a reliable, automatically generated prediction of the position at which the respective machine component under consideration is located in its life cycle.

[0017] The resulting recommendation for action accurately reflects the actual need for action. Thus, the recommendation for action determined according to the invention for maintenance to be carried out within a specific period with the aim of maintaining the normal condition of the machine component can consist of a recommendation for replacement by a specific, named point in time to prevent a spontaneous failure that is highly likely to occur after this point in time, or of waiting if the recorded measurement has proven to be non-critical, i.e., confirming a normal condition.

[0018] Since the recommended course of action determined according to the invention is given at a time when no damage event has yet occurred, the inventive procedure allows for a high degree of planning certainty. Maintenance work or the replacement of a machine component can then be carried out at a time when a shutdown of the respective machine is planned, such as on a weekend or during routine maintenance work scheduled for other reasons.

[0019] To assign influencing factors or measured values ​​to specific uses of machine components, a so-called "application key" can be defined. This key is applied in a standardized form and links the assigned influencing factors together. Such an application key simplifies the automated processing of the process to determine the most appropriate course of action.

[0020] The essential aspect of the inventive method is that, in an automated process based on empirically determined or theoretically formulated information stored in a database, the machine element optimally suited for the respective application is accurately identified, starting from the specific application itself. The information used for this purpose, stored in the database, describes the relationships between the respective application, the influencing factors applicable to it, and their interactions. In the sense of a self-learning system, the quality and reliability of this information can be continuously improved by repeatedly feeding in newly acquired data.

[0021] The system according to the invention can be designed such that, for new applications for which there is no directly assigned information in the database, it can, for example, use the influencing factors assigned to this new application and the interactions determined or formulated for each of them from the information stored in the database to identify the next most likely application and, based on this next most likely application, suggest a suitable machine element. In this way, an optimally suited machine element can be quickly determined for a new application in an iterative process, without the need for complex and time-consuming trials.

[0022] The advantages of the method according to the invention therefore consist in a detailed, application-specific analysis and dimensioning of the design according to any criteria, optimal utilization of the product or material properties actually inherent in the respective machine element under consideration and the resulting actual service life taking into account the specific characteristics of the respective application, the targeted avoidance of premature or delayed replacement of the machine element and the associated optimal utilization of the resource that the respective machine element represents, the avoidance of uncontrolled failures that can be caused in the prior art due to the selection of machine elements that are sufficient according to the selection criteria considered so far, but prove to be unsuitable in practice, or due to unrecognized interactions of influencing factors,the continuous improvement of predictions for application-specific questions, and the targeted, economical product development for specific problems, whereby the particular advantage of the inventive procedure lies in the fact that not only individual components or groups of components can be considered in the selection of machine components, but that the result of the inventive procedure in determining the component optimized for the respective application can also be suggestions for the selection of specific components, materials or process parameters with which a machine element or an assembly ("aggregate") formed from several interacting machine elements can be manufactured, each of which is optimally adapted to the requirements arising from the intended use.

[0023] Application-related influencing factors can be mechanical loads (speed, torque, preload forces, drive geometry, friction pairing, load changes, load cycles, speeds, accelerations, tensile or compressive stresses, other dynamic or static loads); environmental conditions (atmosphere in which the application takes place, ambient humidity and temperature, contamination, material pairing, pressure), properties of the material from which the machine component is made (material behavior when heated, under load, under load changes, under friction), geometric shape and quality of the machine component.

[0024] The measurement result recorded at the respective monitored machine component can be transmitted to a central computer system using any conceivable data transmission device. This system stores the information used according to the invention and performs the respective interpretation of the measured value. The central computer system then returns the action recommendation derived from the measured value to the machine operator, also via a suitable data transmission device, or automatically initiates any necessary steps.

[0025] A typical embodiment of the invention is described below.

[0026] Fig. 2 This shows a conveyor system F in a top view.

[0027] In modern automobile manufacturing, the individual components, the body, or the partially assembled vehicle are transported on sleds along conveyor tracks. The sleds S of such conveyor systems F, also known in practice as "SKID systems," run on rollers 1, which are mounted on shafts 2, 3.

[0028] One shaft 2 is driven by an electric motor 4, while the other shafts 3 are coupled to the driven first shaft 2 via a belt drive 5.

[0029] The rollers 1 have a casing 6 around their circumference made of a polyurethane material ("PU material") or another elastomer. The casing 6 accordingly exhibits both a certain elasticity and high friction to ensure a secure and sufficiently damped coupling of the carriages to be driven (http: / / www.logsystems.de / skid.html).

[0030] The rollers (1) of SKID systems are subject to high loads and correspondingly high wear. At the same time, the unforeseen failure of such a roller (1) can mean the shutdown of an entire production line. To avoid this, the rollers (1) in conventional SKID systems are replaced at regular, fixed intervals.

[0031] In a skid system F according to the invention, at least one of the rollers 1 is equipped with a sensor 7 which detects the temperature development in the PU coating 6 of the roller 1.

[0032] In the course of the inventive method, the known applications in which, for example, the rollers 1 considered here as examples are used are recorded. For this purpose, customer data, applications and products used, production conditions, operating conditions, etc., are recorded in detail in an application key. The typical application scenario for the respective application is then simulated on suitable test benches, and, for this specific application, tests such as lifetime tests are carried out. The insights gained in this way from practical customer experience or from the test results become part of the pool of interconnected information stored in the database.

[0033] This information represents the expected service life under the respective operating conditions as well as the expected operating behavior of the rollers 1 in the application case "SKID system" and correlates this, for example, with the temperature development in the PU material of the PU coating 6 of the roller 1.

[0034] Sensor 7 sends the measured values ​​it has recorded to a central data processing system (not shown here), where the measured values ​​are evaluated and a recommendation for action is derived from this evaluation, based on which a replacement or general maintenance of roller 1 may subsequently be carried out.

[0035] Accordingly, the temperature of the PU coating 6 measured by sensor 7 is normally within a specific normal temperature range. With increasing operating time, the rolling behavior of the carriages S on the rollers 1 changes due to various influencing factors, such as aging of the coating, material variations or fatigue, geometric errors in the alignment of the individual components relative to each other (e.g., misalignment), and changes in the condition of the circumferential surface of the rollers 1 that comes into contact with the carriages S due to damage or abrasion. Wear and flexing increase, so the temperature in the coating 6 also rises. Consequently, the temperature measured by sensor 7 increases.

[0036] The temperature detected by sensor 7 is used as an input variable in the data processing unit running in Fig. 1The interpretive system is fed in and compared with the stored information on the influencing factors that are essential for the "SKID system" application and the correlated temperature development of the casing 6 of roller 1. If this interpretation shows, for example, that the temperature increase is progressing at a rate that deviates from a standard curve, indicating complete failure, i.e., reaching the service life limit, within a certain, near period, the system recommends replacing roller 1 at a safe distance before reaching that period. However, if the recorded temperature shows that the temperature development is due to, for example, seasonal warming of the ambient temperature or similar factors, the recommendation "Normal operation" is issued, meaning no action is required.

[0037] The temperature profile recorded by sensor 7 and any other available information about the operating conditions under which the rollers 1 are used are compared with the information stored in the interpretation system. The actual wear condition of each replaced roller 1 can then be checked on a sample basis.

[0038] If a significant deviation of the predicted state from the actual state is found, the recorded temperature profile and the associated further information about the operating conditions are included in the database underlying the interpretation system, and the information stored there on influencing factors and rules for coupling and mutually influencing these influencing factors is updated accordingly.

[0039] Similarly, information about unforeseen damage events that were not predicted or were incorrectly predicted by the interpretation system is fed into the system and used to refine the rules and influencing factors stored there.

[0040] In this way, the interpretation system is continuously improved, so that the accuracy of the forecast results leading to the respective recommendations for action is also continuously improved. REFERENCE MARK

[0041] F Conveyor equipment S Sled 1 Rollers 2, 3 Shafts 4 Electric motor 5 Belt drive 6 Roller sheathing 1 7 Sensor

Claims

1. Method for computer-assisted prediction of future operating states of machine components, comprising the following steps: A) For a respective machine component under consideration, at least one parameter is selected that is representative of the condition of the machine component under consideration; B) The parameter is captured as a measured value during ongoing use of the machine component under consideration; C) The captured measured value is fed into an automatic process for finding an action recommendation, while taking into account the application in which the machine component under consideration is used, based on the following procedure: C.a) For the respectively considered machine component, properties and the influencing factors that determine these properties have been determined; C.b) For the respectively considered machine component, the relevant applications have been determined, the influencing factors to which the machine component is exposed in the respective application have been identified, and it has been determined which influencing factors are to be taken into account when interpreting the captured measurement value and the resulting action recommendation; C.c) Based on statements obtained from the practical use of the machine component under consideration regarding the influencing factors to be considered, or from statements relating to the influencing factors to be considered obtained through systematic experimental investigations, the interactions and effects of the influencing factors to be considered are determined and linked to specific damage events; C.d) The information determined in work step C.c) on the interactions and damage events influenced by them is fed into a database as machine-readable data; C.e) The respective application, the recorded measurement value and the influencing factors to be considered are fed as input variables into a selection algorithm running on a computer, which selects the most suitable action recommendation for use in the respective application based on the information stored in the database, taking into account the input variables and; C.f) At least one of the machine components under consideration, for which the respective measured value is captured and interpreted in the manner specified in the work step C.e), is uniquely identified with regard to the application in which it is used; C.g) Thereto, the at least one of the machine components under consideration is identified and monitored in its practical use by means of an identification device, whereby the monitoring serves to capture the measured value and at least three further characteristic values, whereby (i) one of the at least three further characteristic values provides information about the time that has elapsed between the start of use and the recording of the respective measured value, and (ii) two further characteristic values of the at least three characteristic values each provide information about a respective external influence, in particular temperature and / or atmosphere and / or ambient humidity and / or contamination and / or pressure, to which the at least one of the machine components under consideration was exposed during use, and wherein not every machine component under consideration is monitored; C.h) The statements about the measured value and the other characteristic values captured by the monitoring in step C.g) are fed into the process beginning with steps C.c) as influencing variables.

2. Method according to claim 1, characterised in that that if, during ongoing use of the method it transpires in step C.c) that there are still unidentified influencing factors, these newly identified influencing factors are added to the group of influencing factors to be considered and being in interaction with other influencing factors or with damage events being put into association therewith.

3. Method according to one of the preceding claims, characterised in that the database into which the previously determined information is fed in the step C.d) comprises one, two or more individual databases, which respectively contain machine-readable data relating to specific use cases.

4. Method according to one of the preceding claims, characterised in that a representative number of machine components in the respective application is monitored.

5. Method according to one of the preceding claims, characterised in that the identification means contains information about the nature of the machine component in addition to information that uniquely identifies the respective machine component.

6. Method according to one of the preceding claims, characterised in that, in step C.g), the time period between the start of use and a replacement necessitated by wear is recorded as a further characteristic value.

7. Method according to one of the preceding claims, characterised in that the monitoring in step C.g) provides information about a set of two or more further characteristic values.

8. Method according to one of the preceding claims, characterised in that the identification means is equipped in such a way that it provides information about the internal state and the loads on the machine element during operation.

9. System for computer-assisted prediction of future operating states of machine components, comprising a database and further means for carrying out the method according to one of claims 1 to 8.