Abstract
A method for the automated identification of a product defect of a product and/or for the automated identification of a product defect cause of the product defect, includes producing the product from a plurality of product elements via a plurality of manufacturing steps, and gathering a number n of items of test information by at least one product test, wherein the n items of test information form an n-dimensional test value. The method also includes carrying out a dimension reduction of the n-dimensional test value by at least one statistics process to obtain a dimension-reduced test value, comparing the dimension-reduced test value with a multitude of learned reference values, assigning the dimension-reduced test value to at least one group of reference values that are similar to each other, and identifying, in an automated manner, the product defect and/or the product defect cause on the basis of the assignment.
Claims
1-15: (canceled)
16. A method for the automated identification of a product defect of a product (1, 2, 3, 40, 41, 42, 43, 44, 45) and/or for the automated identification of a product defect cause of the product defect, comprising: producing the product (1, 2, 3, 40, 41, 42, 43, 44, 45) from a plurality of product elements (4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18) via a plurality of manufacturing steps; gathering a number n of items of test information by at least one product test (101), the n items of test information forming an n-dimensional test value; carrying out a dimension reduction of the n-dimensional test value (102) by at least one statistics process to obtain a dimension-reduced test value; comparing the dimension-reduced test value (103) with a multitude of learned reference values (46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60); assigning the dimension-reduced test value to at least one group of reference values (46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 104) that are similar to each other; and identifying, in an automated manner, the product defect (105) and/or the product defect cause (106) on the basis of the assignment.
17. The method of claim 16, wherein the plurality of reference values (46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60) is classified according to product defects and/or product defect causes during a learning process.
18. The method of claim 16, wherein the assignment takes place (104) in accordance with a distance matrix.
19. The method of at least one of claim 16, wherein the reference values (46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60) are dimension-reduced by the at least one statistics process to a dimension number that is identical to that of the dimension-reduced test value.
20. The method of at least one of claim 16, wherein the dimension-reduced test value has at least one hundred dimensions.
21. The method of claim 16, wherein further comprising making available an assignability of the multitude of product elements (4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18) to the product (1, 2, 3, 40, 41, 42, 43, 44, 45) and/or a traceability of the product (1, 2, 3, 40, 41, 42, 43, 44, 45) across all manufacturing steps.
22. The method of claim 16, further comprising adjusting an open-loop control of a manufacturing process of the product (1, 2, 3, 40, 41, 42, 43, 44, 45) based at least in part on identified product defects and product defect causes (107).
23. The method of claim 16, wherein the items of test information comprises one or more of acoustic items of information, mechanical items of information, and electrical items of information.
24. The method of claim 16, wherein determining a repair measure as wells as one or more of a probability of success, a cost, and a time required for the repair measure of the product based at least in part on an identified product defect.
25. The method of claim 16, wherein further comprising adapting the method, in an automated manner, to a plurality of products (1, 2, 3, 40, 41, 42, 43, 44, 45).
26. The method of claim 16, wherein outputting one or more of a notification regarding identified product defects and/or product defect causes, the probability of success, cost, and time required for the repair of the product in an automated manner.
27. The method of claim 16, wherein the method is performed out by a knowledge-based artificial intelligence, wherein the artificial intelligence retrains itself.
28. The method of claim 16, wherein the method is carried out after completion of the product (1, 2, 3, 40, 41, 42, 43, 44, 45).
29. A device for automated identification of a product defect of a product (1, 2, 3, 40, 41, 42, 43, 44, 45) and/or for the automated identification of a product defect cause of the product defect, comprising: means for producing the product (1, 2, 3, 40, 41, 42, 43, 44, 45) from a plurality of product elements (4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18) by a plurality of manufacturing steps; means for gathering a number n of items of test information by at least one product test, the n items of test information forming an n-dimensional test value; means for carrying out a dimension reduction of the n-dimensional test value by at least one statistics process to obtain a dimension-reduced test value; means for comparing the dimension-reduced test value with a multitude of learned reference values (46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60); means for assigning the dimension-reduced test value to at least one group of reference values (46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 104) that are similar to each other; and means for identifying, in an automated manner, the product defect and/or the product defect cause based on the assignment.
30. A device configured for implementing the method of claim 16.
Description
BRIEF DESCRIPTION OF THE DRAWINGS
[0055] Example aspects of invention are explained by way of example in the following with reference to embodiments represented in the figures, wherein
[0056] FIG. 1 shows, by way of example and diagrammatically, a manufacturing process of a product,
[0057] FIG. 2 shows, by way of example, a simplification achievable by the method according to the invention as compared to a comparison process that is typical from the prior art,
[0058] FIG. 3 shows, by way of example and diagrammatically, different variants of products and groups of reference values assigned thereto, in the form of a table,
[0059] FIG. 4 shows, by way of example and diagrammatically, several groups of reference values, and
[0060] FIG. 5 shows, by way of example, one possible embodiment of the method according to the invention for the automated identification of a product defect of a product and/or for the automated identification of a product defect cause of the product defect in the form of a flow chart.
[0061] Identical objects, functional units, and comparable components are marked with the same reference characters in all figures. These objects, functional units, and comparable components are identically designed with regard to their technical features, as long as nothing else results, explicitly or implicitly, from the description.
DETAILED DESCRIPTION
[0062] Reference will now be made to embodiments of the invention, one or more examples of which are shown in the drawings. Each embodiment is provided by way of explanation of the invention, and not as a limitation of the invention. For example, features illustrated or described as part of one embodiment can be combined with another embodiment to yield still another embodiment. It is intended that the present invention include these and other modifications and variations to the embodiments described herein.
[0063] FIG. 1 shows, by way of example and diagrammatically, a manufacturing process of a product and the associated complexity of the identification of a product defect of the product and of the identification of the underlying product defect cause. Within the scope of the manufacturing process shown by way of example, three different variants 1, 2, 3 of a product 1, 2, 3 designed as a vehicle transmission 1, 2, 3 are manufactured, according to the example. The vehicle transmissions 1, 2, 3 are each manufactured from a multitude of product elements 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18 in a multitude of manufacturing steps, wherein a first portion of product elements 4, 5, 6, 7, 8 is supplied and a second portion of product elements 9, 10, 11, 12, 13, 14, 15, 16, 17, 18 is manufactured in-house. The supplied product elements 4, 5, 6, 7, 8 as well as the in-house manufactured product elements 9, 10, 11, 12, 13, 14, 15, 16, 17, 18 can have a product defect. According to the example, the supplied product element 5 and the in-house manufactured product element 13 both have a product defect. The multitude of product elements 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18 is assembled in one assembly step, a pre-assembly according to the example, to form assemblies 19, 20, 21, 22, 23. According to the example from FIG. 1, a product defect arises during the assembly of the assembly 20. The assemblies 19, 20, 21, 22, 23 are then combined, in a further assembly step, the final assembly according to the example, to form the complete products 1, 2, 3, namely the vehicle transmissions 1, 2, 3, wherein the vehicle transmissions 1 and 2 are produced, according to the example, in larger numbers than the vehicle transmission 3. A product defect also arises during the final assembly of the vehicle transmission 3, according to the example. After the final assembly, the vehicle transmissions 1, 2, 3 are inspected according to the method according to the invention for the automated identification of a product defect of a product and/or for the automated identification of a product defect cause of the product defect. The advantage of the method according to example aspects of the invention is, primarily, that only the fully assembled products 1, 2, 3, e.g., the vehicle transmissions 1, 2, 3, are inspected, and a series of individual inspections does not need to be carried out after each assembly step and/or after the delivery or the manufacture of the multitude of product elements 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18. The method according to example aspects of the invention includes [0064] producing the product 1, 2, 3 from a multitude of product elements 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18 by a multitude of manufacturing steps, and [0065] gathering a number n of items of test information by at least one product test, wherein the n items of test information form an n-dimensional test value, [0066] carrying out a dimension reduction of the n-dimensional test value by at least one statistics process to obtain a dimension-reduced test value, [0067] comparing the dimension-reduced test value to a multitude of learned reference values, [0068] assigning the dimension-reduced test value to at least one group of reference values that are similar to each other, and [0069] identifying, in an automated manner, the product defect and/or the product defect cause on the basis of the assignment.
[0070] Therefore, a reliable identification not only of the existing product defects, but also of the product defect causes underlying the product defects after the complete manufacture of the products 1, 2, 3 is made possible. According to the exemplary embodiment from FIG. 1, the vehicle transmissions 1, 2, 3 are subjected to an acoustic test, wherein a total of 170.Math.10.sup.6 items of test information are gathered. By the statistics process, a dimension reduction of the 170.Math.10.sup.6-dimensional test value to a dimension-reduced test value, namely, for example, a 1200-dimensional test value, is carried out. According to the example, it becomes apparent that one of the vehicle transmissions 1, both vehicle transmission 2, and the vehicle transmission 3 each have a product defect. Due to the method according to example aspects of the invention, it also becomes apparent that the product defect of the vehicle transmission 1 results from the faulty supplied product element 5 as the product defect cause. The product defects of the two vehicle transmissions 2 result from the faulty in-house manufactured product element 13 as well as from a faulty pre-assembly of the assembly 20, as the product defect cause. Finally, with respect to the product defect of the vehicle transmission 3, it becomes apparent that it results from a faulty final assembly of the vehicle transmission 3.
[0071] FIG. 2 shows, by way of example, a simplification achievable by the method according to example aspects of the invention as compared to a comparison process that is typical from the prior art, in the form of a flow chart. The known method is represented at the top in FIG. 2 and the method according to example aspects of the invention is represented at the bottom in FIG. 2. According to the prior art, in a method step 30, initially a product defect of a product 1, 2, 3 is identified. This also takes place within the scope of the method according to example aspects of the invention in step 30. In order to identify the product defect cause, the product is now disassembled, according to the prior art, by skilled persons in step 31 and, in step 32, the multitude of product elements 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18 are individually examined and defects are analyzed, which is associated with a comparatively high time requirement and corresponding cost. The high time requirement results primarily from the fact that a multitude of individual tests must be carried out in order to identify the product defect cause. The product defect cause is first identified in step 33 as the result of the examination and analysis of the multitude of product elements 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18. The method according to example aspects of the invention, however, allows for an automated identification not only of the product defect in step 30, but also of the product defect cause in step 33 by a comparison of the dimension-reduced test value with a multitude of groups of learned reference values. Due to the fact that the dimension-reduced test value is still characterized by all n items of test information, i.e., has similarities to appropriate groups of reference values, despite the dimension reduction by the at least one statistics process, the product defect cause can be finally identified by subsequently assigning the dimension-reduced test value to at least one group of reference values that are similar to each other. Therefore, significant savings of time and cost can be achieved as compared to the method that is typical from the prior art.
[0072] FIG. 3 shows, by way of example and diagrammatically, in the form of a table, different variants 40, 41, 42, 43, 44, 45 of products 40, 41, 42, 43, 44, 45 and, assigned thereto, groups of reference values 46, 47, 48, 49, 50, to which the test values are compared, in order to make an assignment possible. The groups of reference values 46, 47, 48, 49, 50 each describe different technical features and/or properties, some of which can be identical for several or all variants 40, 41, 42, 43, 44, 45 of products 40, 41, 42, 43, 44, 45. This identification of the particular variants 40, 41, 42, 43, 44, 45 to be inspected and of the groups of reference values 46, 47, 48, 49, 50 to be utilized takes place in an automated manner. Therefore, the method according to example aspects of the invention for the automated identification of a product defect of a product and/or for the automated identification of a product defect cause of the product defect can also be carried out in an easy way in the presence of different variants 40, 41, 42, 43, 44, 45 of products 40, 41, 42, 43, 44, 45.
[0073] FIG. 4 shows, by way of example and diagrammatically, several groups 51, 52, 53, 54, 55, 56, 57, 58, 59, 60 of reference values, which were dimension-reduced to two dimensions in each case by at least one statistics process. The reference values are also characterized, in their two-dimensional representation, by all dimensions and/or items of test information taken into account in the at least one statistics process, which is why reference values that describe similar product properties and/or product defects are sorted into groups 51, 52, 53, 54, 55, 56, 57, 58, 59, 60. Neither the x-axis nor the y-axis has a unit, since the units are dispensed with anyway due to the dimension reduction. Similarities between the reference values and/or between the groups of reference values are represented exclusively by their particular spacing in the coordinate system. According to the example, the reference values of the group 51 represent a faulty clutch return mechanism of a product designed as a vehicle transmission. The product defect cause with respect to the reference values of the group 51 is a mechanical return spring that was inadvertently not installed during the assembly. If a test value that has also been reduced to two dimensions can now be assigned to the group 51 due to the proximity of the test value to this group 51, it can be detected, on the basis thereof, that the vehicle transmission, from which the test value originated, also has a faulty clutch return mechanism, in which the return spring was inadvertently not installed. The group 52 represents, according to the example, a fully operable and faultless transmission. The further groups 53, 54, 55, 56, 57, 58, 59, 60 each represent further specific product defects as well as the product defect causes underlying them.
[0074] FIG. 5 shows, by way of example, one possible example embodiment of the method according to the invention for the automated identification of a product defect of a product and/or for the automated identification of a product defect cause of the product defect in the form of a flow chart. In the method step 100, initially a production of the product 1, 2, 3, 40, 41, 42, 43, 44, 45 from a multitude of product elements 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18 takes place by a multitude of manufacturing steps. In step 101, after a completion of the product 1, 2, 3, 40, 41, 42, 43, 44, 45, a gathering of a number n of items of test information takes place by at least one product test, wherein the n items of test information form an n-dimensional test value. Acoustic items of information, mechanical items of information, and electrical items of information are gathered as items of test information. In the following step 102, a dimension reduction of the n-dimensional test value is carried out by at least one statistics process to obtain a dimension-reduced test value, which, in step 103, is compared to a multitude of learned reference values 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, wherein the reference values have a number of dimensions that is identical to that of the dimension-reduced test value. In the method step 104, an assignment of the dimension-reduced test value then takes place in accordance with a distance matrix to at least one group of reference values 51, 52, 53, 54, 55, 56, 57, 58, 59, 60 that are similar to each other, which, finally, in step 105, permits an identification of the product defect and, simultaneously in step 106, an identification of the product defect cause on the basis of the assignment. On the basis of the product defect and/or the product defect cause identified in step 106, an open-loop control of a manufacturing process of the product 1, 2, 3, 40, 41, 42, 43, 44, 45 takes place in the following step 107 in the sense that the manufacturing process is influenced, modified, and/or corrected in such a way that the identified product defect cause is avoided and the identified product defect therefore no longer arises in the subsequently manufactured products 1, 2, 3, 40, 41, 42, 43, 44, 45. Simultaneously with step 107, in step 108, a notification regarding the identified product defect, the product defect cause, the probability of success of a repair, the cost of the repair, and the time required for the repair of the product 1, 2, 3, 40, 41, 42, 43, 44, 45 is output, in an automated manner, to a group of human operators. According to the example, the method is carried out by a knowledge-based artificial intelligence, which retrains itself on the basis of the reference values fed thereto.
[0075] Modifications and variations can be made to the embodiments illustrated or described herein without departing from the scope and spirit of the invention as set forth in the appended claims. In the claims, reference characters corresponding to elements recited in the detailed description and the drawings may be recited. Such reference characters are enclosed within parentheses and are provided as an aid for reference to example embodiments described in the detailed description and the drawings. Such reference characters are provided for convenience only and have no effect on the scope of the claims. In particular, such reference characters are not intended to limit the claims to the particular example embodiments described in the detailed description and the drawings.
REFERENCE CHARACTERS
[0076] 1, 2, 3 product, vehicle transmission [0077] 4, 5, 6, 7, 8, 9, 10 product element [0078] 11, 12, 13, 14, 15 product element [0079] 16, 17, 18 product element [0080] 19, 20, 21, 22, 23 assembly [0081] 30 identification of a product defect [0082] 31 identification of a product defect cause by skilled persons [0083] 32 examination and analysis of a multitude of product elements by skilled persons [0084] 33 identification of a defect cause [0085] 34 drive of the vehicle drive system additionally first electric motor [0086] 40, 41, 42 product, vehicle transmission [0087] 43, 44, 45 product, vehicle transmission [0088] 46, 47, 48, 49, 50 group of reference values [0089] 51, 52, 53, 54, 55 group of reference values [0090] 56, 57, 58, 59, 60 group of reference values [0091] 100 production of a product [0092] 101 gathering a number n of items of test information [0093] 102 carrying out a dimension reduction [0094] 103 comparison with a multitude of learned reference values [0095] 104 assigning the dimension-reduced test value to at least one group of reference values that are similar to each other [0096] 105 identifying the product defect [0097] 106 identifying the product defect cause [0098] 107 open-loop control of a manufacturing process [0099] 108 outputting a notification
