Method and apparatus for order analysis

Abstract

The invention relates to a method comprising the following method steps: carrying out a rolling test of a toothing (2) by means of a rolling test bench (6, 8) for a single-flank rolling test and / or a double-flank rolling test; generating an order spectrum (10) from measurement data of the rolling test of the toothing (2); analyzing the order spectrum (10) by means of a computer-implemented classification (12), the classification (12) being arranged for determining a tooth deviation (I-X) of the tooth (2) and a test bench deviation (XI) of the rolling test bench (6, 8); and adjusting the manufacturing method of the tooth (2) on the basis of the determined tooth deviation (I-X) of the tooth, and / or adjusting the rolling test bench (6, 8) on the basis of the determined test bench deviation (XI, XII) of the rolling test bench (6, 8).

Description

Technical Field

[0001] This invention relates to a method and apparatus for performing and analyzing rolling tests. Background Technology

[0002] In motor vehicles operating on pure electric motors, the noise from the transmission is no longer masked by the noise of the internal combustion engine. The noise generated by the rolling gears is perceived as disturbing by occupants. Therefore, in addition to the usual requirements for the efficiency and service life of the gears, noise characteristics play a decisive role in the quality evaluation of the gears.

[0003] Therefore, in addition to quality assurance through traditional coordinate measurement techniques, rolling test benches are frequently used to evaluate the dynamic characteristics of tooth components, especially their noise characteristics. Furthermore, the results of rolling tests can provide conclusions regarding the quality of numerous geometric parameters of the tooth components, such as tooth pitch and runout.

[0004] The results of rolling tests are often presented in the form of an order spectrum. Based on the dominant order in the order spectrum, conclusions can be drawn about tooth deviations and the expected noise characteristics of the associated teeth in a fully assembled state.

[0005] Such order spectrum analysis is performed by machine operators who derive modifications to the manufacturing process based on empirical values ​​from previous evaluations. However, this approach relies heavily on the experience of the personnel involved and is therefore prone to errors. For example, misjudging the tooth geometry deviations that form the basis of the order to be corrected can lead to incorrect corrections in the manufacturing process. This may result in the need to perform multiple time-consuming and costly correction cycles until an acceptable order spectrum is obtained.

[0006] Manual review and evaluation of the order spectrum can also lead to the interpretation of test bench deviations that may affect the order spectrum as tooth deviations. For example, for a specific tooth configuration, it may be possible that the combination of tooth number and test speed generates vibrational excitation within a certain natural frequency range of the rolling test bench, which manifests as the dominant order in the order spectrum. Similarly, wear on the drive mechanism, bearings, or supports of the rolling test bench can also cause dominant orders in the order spectrum that are not caused by tooth deviations.

[0007] Such order spectrum distortion caused by the test bench itself is usually related to rotational speed. Therefore, as long as the order spectrum "migrates" with rotational speed—that is, the dominant order changes with rotational speed—the distortion of the order spectrum may be identifiable. However, this requires the personnel evaluating the order spectrum to have detailed knowledge of these effects.

[0008] The order spectrum is normalized relative to rotational speed. That is, the first order describes a measurement amplitude assigned to a test rotational speed, while all other orders describe multiples of the test rotational speed. Therefore, the dominant order, purely due to tooth geometry deviations, is the same at different rotational speeds—that is, these same orders always appear as the dominant orders regardless of the test rotational speed. Conversely, deviations caused by deviations in the test bench itself (and especially vibration excitation) tend to be rotational speed-dependent, and thus the order spectrum varies with rotational speed.

[0009] To evaluate whether the dominant order in the order spectrum is caused by geometric deviations of the teeth or by the influence of the test bench on the measurement results, the personnel responsible for reviewing and evaluating the corresponding order spectrum need extensive experience. Therefore, in practice, the following situation may occur: although the deviation to be corrected in the order spectrum is caused by the test bench itself and does not require adjustments to the manufacturing process, multiple ineffective corrections are still performed on the manufacturing process of the relevant teeth.

[0010] Furthermore, even highly experienced personnel may encounter situations where they are facing previously unknown order spectra or failure modes for the first time—for example, when a new or modified tooth geometry is used for the first time in mass production, or when a specific deviation or test bench deviation occurs for the first time. Summary of the Invention

[0011] Against this backdrop, the technical problem upon which this invention is based is to propose a method capable of reliably analyzing the order spectrum of rolling tests. In particular, it should be able to reliably correct and distinguish between tooth deviations and test bench deviations. Furthermore, an apparatus for performing this method is also proposed.

[0012] The aforementioned technical problems are solved by the features of the independent claims. Other embodiments of the invention are derived from the dependent claims and the following description.

[0013] According to a first aspect, the present invention relates to a method comprising the following steps: performing a rolling test on a tooth using a rolling test bench for single-tooth-surface rolling tests and / or double-tooth-surface rolling tests; generating an order spectrum from measurement data of the tooth rolling test; analyzing the order spectrum using a computer-implemented classification, wherein the classification is configured to determine tooth deviations and test bench deviations of the rolling test bench; adjusting the manufacturing method of the tooth based on the determined tooth deviations, and / or adjusting the rolling test bench based on the determined test bench deviations of the rolling test bench.

[0014] On the one hand, the computer-implemented classification can identify tooth deviations based on the analyzed order spectrum. Furthermore, this classification can distinguish between tooth deviations and test bench deviations by identifying test bench deviations.

[0015] This avoids the situation where the manufacturing method is incorrectly corrected even though the cause of the identifiable deviation in the order spectrum is test bench bias. In other words, it avoids the situation where the dominant order is incorrectly assigned to a tooth deviation that does not actually exist, and the correction of this non-existent tooth deviation does not improve the order spectrum for subsequent measurements.

[0016] Furthermore, computer-based classification provides a vast database that exceeds most users' knowledge of order spectrum interpretation. Therefore, even rare order spectrum variations can be reliably evaluated and specifically corrected. In particular, the feasibility of distinguishing between tooth deviations and test bench deviations via computer avoids the lengthy trial-and-error process required in actual operation.

[0017] If the order spectrum analysis determines that the gear teeth deviate from their target geometry and this deviation must be corrected, then the manufacturing method can be adjusted computer-based. This means, for example, adjusting the kinematics of a gear machining method (such as grinding) and / or adjusting the geometry of the tools used for gear machining (such as grinding tools) to reduce the deviation of the gear teeth from their target geometry, or in other words, improving the noise characteristics of the gear teeth. The correction parameters required to adjust the kinematics and / or tool geometry can be automatically generated by software.

[0018] Therefore, a closed-loop control system for quality assurance can be provided, wherein the unacceptable amplitudes of the dominant order in the order spectrum are corrected by adjusting the manufacturing method to reduce these amplitudes to an acceptable level.

[0019] The above operations can also be performed automatically for test bench deviations. For example, it can prompt the maintenance needs of specific components of the rolling test bench, suggest alternative test speeds, or suggest the damping or stiffness of specific components of the rolling test bench.

[0020] In this article, the related concept of "adjusting" the test bench should be interpreted broadly and includes any influence on the rolling test bench that would reduce or eliminate its impact on the measurement results. For example, this could be replacing, repairing, or maintaining components of the rolling test bench, or adjusting test procedure settings such as the test speed or variations in test speed, or the clamping force in gear contact.

[0021] In this article, the terms "test bench" and "rolling test bench" are used synonymously.

[0022] In this article, when single-tooth-surface rolling test is mentioned, it refers to a known test method used to test transmission teeth. In this method, the tooth to be tested rolls with a corresponding mating gear or standard gear at a fixed shaft spacing. Single-tooth-surface rolling test can detect, for example, the following tooth deviations: circular runout, rolling deviation, circular runout error, inter-tooth amplitude, maximum rolling deviation, transmission error and dynamic backlash, noise characteristics, and surface errors.

[0023] In this article, when referring to the double-tooth-surface rolling test, it refers to a known test method used to test transmission teeth. In this method, the tooth to be tested rolls with a corresponding mating gear or standard gear at a non-fixed shaft spacing, wherein a defined clamping force is applied between the rolling meshing teeth. The double-tooth-surface rolling test can detect, for example, the following tooth deviations: circular runout, rolling runout, rolling deviation, double ball size, and noise characteristics.

[0024] Methods for rolling tests with rotational error analysis, or more specifically, single-tooth-surface rolling tests and double-tooth-surface rolling tests, are existing technologies and are well known. The core of this invention is not the rolling test itself, or the single-tooth-surface rolling test or double-tooth-surface rolling test, but rather the computer-implemented identification and differentiation of tooth deviations and test bench deviations in the order spectrum, which presents the results of such rolling tests.

[0025] Here, order analysis first generates an order spectrum from the results of the rolling test. The individual orders and / or order ranges of this spectrum can then be assigned to test characteristics of the tooth, such as circular runout error; wobble; first-order and / or higher-order pitch errors; surface waviness; tooth surface shape errors, etc. Specifically, the results of the rolling test are generated by providing rotational speed-dependent shaft data from the rolling test bench (e.g., via FFT) as an order spectrum. The abbreviation "FFT" here stands for "Fast Fourier Transform."

[0026] Here, the order is several times the rotational speed of the teeth on the rolling test bench, so the measured deviation or measurement value is plotted as amplitude at each order.

[0027] The teeth to be tested are transmission teeth used for power transmission, which have speed and torque conversion capabilities. The method of the present invention can also be applied, for example, to spur gears and bevel gears.

[0028] It can be specified that the classification is performed by an artificial intelligence model (such as a neural network or the like) that is trained based on training data.

[0029] According to one embodiment of the method, multiple categories of indicative tooth deviations for the classification can be specified, wherein the training data has a variation process of at least one order amplitude and / or at least one order spectrum for each of these categories, the amplitude and / or variation process having the characteristics of one or more tooth deviations, wherein these categories are different from each other in terms of at least one assigned tooth deviation.

[0030] It can be specified that, regarding the classification, at least one category of indicator test bench bias is provided, wherein the training data has a variation process of at least one order of amplitude and / or at least one order of spectrum, the amplitude and / or variation process having the characteristics of one or more test bench biases.

[0031] Alternatively or additionally, it may be specified that, with respect to the classification, at least one category of indicator test bench deviation, the training data has a variation process of at least one order amplitude and / or at least one order spectrum, which amplitude and / or variation process does not have the characteristics of one or more tooth deviations.

[0032] It can be stipulated that the artificial intelligence model is trained based on training data, which can optimize the classification of dental deviations in terms of the noise characteristics of the teeth, especially the psychoacoustic characteristics.

[0033] Alternatively or additionally, it may be stipulated that the artificial intelligence model is trained based on training data that can optimize the classification of tooth deviations in terms of tooth lifespan.

[0034] Alternatively or supplementally, it may be stipulated that the artificial intelligence model is trained based on training data that can optimize the classification of tooth deviations in terms of tooth efficiency.

[0035] According to one embodiment of the method, multiple artificial intelligence models can be specified, wherein these artificial intelligence models are trained based on training data that can optimize the classification of tooth deviations in one or more of the following aspects: the noise characteristics of the teeth, especially the psychoacoustic characteristics; the lifespan of the teeth; and the efficiency of the teeth.

[0036] It can be stipulated that after adjusting the manufacturing method of the tooth and / or adjusting (especially maintaining) the test bench, rolling tests and analyses are performed again on the same or another identical tooth to identify the effectiveness or ineffectiveness of the adjustment and / or maintenance, wherein the data on the effectiveness or ineffectiveness of the adjustment and / or maintenance are used as additional training data for the artificial intelligence model to improve the artificial intelligence model.

[0037] According to one implementation of the method, it can be stipulated that training data, artificial intelligence models, and tuning schemes to be assigned to rolling test benches with the same structure are collected and stored on a server.

[0038] It can be stipulated that, based on the results of the analysis, predefined action suggestions or predefined evaluations are output to operators, especially in text form.

[0039] It can be stipulated that an AI-based dialogue be conducted with the operator, during which the effectiveness or ineffectiveness of the suggested measures are inquired.

[0040] If these suggestions are ineffective, alternative action recommendations can be generated based on artificial intelligence. If these recommendations are also ineffective, relevant measurement data of the unimproved teeth can be used as input data to improve the AI ​​model.

[0041] If the action recommendations are effective, the improved tooth measurement data can be used as input data to improve the artificial intelligence model.

[0042] According to a second aspect, the present invention relates to an apparatus for rolling tests of teeth, wherein the apparatus is configured to perform the following method steps: performing a rolling test of the teeth by single-tooth-surface rolling test and / or double-tooth-surface rolling test; generating an order spectrum from the measurement data of the rolling test of the teeth; and analyzing the order spectrum by a computer-implemented classification, wherein the classification is configured to determine the tooth deviation and the test table deviation of the rolling test table. Attached Figure Description

[0043] The present invention will now be described in more detail with reference to the accompanying drawings illustrating embodiments. The drawings are shown schematically:

[0044] Figure 1 The method steps of the present invention are shown;

[0045] Figure 2A The order spectrum corresponding to the first rotational speed is shown;

[0046] Figure 2B The order spectrum corresponding to the second rotational speed is shown;

[0047] Figure 2C The order spectrum corresponding to the third rotational speed is shown. Detailed Implementation

[0048] Figure 1 The method steps of the present invention are shown. First, in the first method step (i), the teeth 2 are machined by cutting using a gear machining tool 4. The gear machining tool 4 can be, for example, a gear grinding machine.

[0049] After the cutting process, in method step (ii), a rolling test is performed on the tooth 2—that is, a single-tooth-surface rolling test and / or a double-tooth-surface rolling test. For this purpose, a test stand 6 for the single-tooth-surface rolling test and a test stand 8 for the double-tooth-surface rolling test are shown exemplary and schematically. Test stands capable of simultaneously performing single-tooth-surface rolling tests and double-tooth-surface rolling tests are known in the prior art.

[0050] In method step (iii), an order spectrum 10 is generated from the measurement data of the rolling test of the tooth 2. Furthermore, in method step (iii), the order spectrum 10 is analyzed by a computer-implemented classification 12, wherein the classification 12 is set to determine the tooth deviation IX of the tooth 4 and the test bench deviations XI and XII of the rolling test benches 6 and 8.

[0051] According to method step (iii), the measured rotational error is plotted in μrad at each order, thereby generating an order spectrum 10. Therefore, the deviation determined by the rolling test is provided in the form of an order spectrum.

[0052] The computer-implemented classification 12 assigns each order or order range of the order spectrum 10 to the tooth deviation of the tooth section 2.

[0053] Rolling test anomalies occurring in order range I are caused, for example, by pitch errors in tooth 2. Order range I extends, for example, from order 1 to approximately order 160.

[0054] Rolling test anomalies occurring in order range II are caused, for example, by deviations in the tooth profile or tooth surface of tooth 2. Order range II extends, for example, from approximately order 160 up to order 430.

[0055] Rolling test anomalies occurring in order range III are caused, for example, by tooth surface ripples in tooth 2. Order range III extends, for example, from order 290 up to orders 500 and above.

[0056] The given numerical values ​​for the extended order range are merely illustrative and are used to illustrate the method and process of the present invention.

[0057] In addition to the order ranges I, II, and III mentioned above, each order can also be specifically assigned to a particular tooth deviation in tooth 2.

[0058] For example, according to the rolling test, the first order of the order spectrum describes the circular runout error of the tooth, where the rotational error corresponding to the first order is denoted as IV. According to the rolling test, the second order of the order spectrum corresponds to the oscillation of the tooth, where the rotational error corresponding to the second order is denoted as V.

[0059] The order range marked VI extends from the third order of the rolling test to the first gear meshing order, where the dominant order in this range indicates the presence of periodically occurring pitch errors.

[0060] The order range marked VII is assigned to those orders in the rolling test that cannot be assigned to the meshing frequency or its harmonics or harmonic sidebands, wherein these orders may also be collectively referred to as ghost orders.

[0061] In this article, the gear meshing orders corresponding to the first, second, third, and fourth gear meshing orders are marked as VIII.

[0062] I and X are used to mark the ranges of the sidebands that affect the harmonic meshing frequency, which are modulated by periodic pitch deviation.

[0063] The gear meshing order corresponding to the fifth or higher gear meshing order is marked as X and is usually caused by the surface waviness of the tooth surface.

[0064] In this document, order spectrum 10' is exemplarily and schematically assigned to test bench deviations XI and XII, which is determined for a different rotational speed than order spectrum 10, and whose dominant order is different from that of order spectrum 10.

[0065] This is Figure 2A , Figure 2B and Figure 2C This is explained in the text. Figure 2A The order spectrum 10 is shown for the test rotational speed i1. Figure 2B The order spectrum 10' is shown for the test rotational speed i2, which is different from the test rotational speed i1. Figure 2C The order spectrum 10'' is shown for a test rotational speed i3, which is different from both test rotational speeds i1 and i2. Therefore, it is true that i1 ≠ i2 ≠ i3.

[0066] like Figure 2A , Figure 2B and Figure 2C As shown, the dominant order differs for different speeds i1, i2, and i3. In other words, the dominant order "shifts" with the speed. This indicates that test bench deviations XI and / or XII distort the measurement results of the rolling test. Therefore, operators can be advised to first eliminate test bench deviations XI and / or XII (e.g., by maintaining or repairing the test bench's bearings, supports, or drive mechanism) before meaningfully identifying and evaluating tooth deviations.

[0067] The designation of test bench deviations as XI and XII is understood as exemplary. Furthermore, test bench deviations independent of rotational speed can also be identified and considered.

[0068] Therefore, according to Figure 1 In step (iii) of the method, the manufacturing method of the tooth 2 is adjusted based on the determined tooth deviation IX of the tooth 2, and / or the rolling test tables 6 and 8 are adjusted based on the determined test table deviations XI and XII of the rolling test tables 6 and 8; or a suggestion to adjust the manufacturing method of the tooth 2 based on the determined tooth deviation IX of the tooth 2 is output, and / or a suggestion to adjust the rolling test tables 6 and 8 based on the determined test table deviations XI and XII of the rolling test tables 6 and 8 is output.

[0069] The computer-implemented classification 12, as well as the adjustment of the manufacturing method and / or the adjustment of the rolling test benches 6 and 8, can be performed by the control and evaluation unit 14.

[0070] The classification is performed by an artificial intelligence model 16, specifically by a neural network trained on training data 18.

[0071] The specification states that, regarding the classification of multiple categories of tooth deviation IX, the training data 18 has a variation process of at least one order amplitude and / or at least one order spectrum for each of these categories, the amplitude and / or variation process having the characteristics of one or more tooth deviation IX, wherein these categories are different from each other in at least one assigned tooth deviation IX.

[0072] Furthermore, it is specified that, regarding at least one category of the classification indicator test bench deviations XI and XII, the training data 18 has a variation process of at least one order amplitude and / or at least one order spectrum, the amplitude and / or variation process having the characteristics of one test bench deviation XI, XII or multiple test bench deviations XI, XII; and / or, regarding at least one category of the classification indicator test bench deviations XI, XII, the training data 18 has a variation process of at least one order amplitude and / or at least one order spectrum, the amplitude and / or variation process not having the characteristics of one tooth deviation IX or multiple tooth deviations IX.

[0073] After adjusting the manufacturing method of tooth 2 and / or, for example, maintaining test benches 6 and 8 and / or adjusting the rolling test, the rolling test and analysis are performed again on the same or another identical tooth to identify the effectiveness or ineffectiveness of the adjustment and / or maintenance, wherein the data on the effectiveness or ineffectiveness of the adjustment and / or maintenance are used as additional training data for artificial intelligence model 16 to improve the artificial intelligence model 16.

[0074] Based on the analysis results, predefined action suggestions 20 and / or predefined evaluations 22 are output to the operators, especially in text form.

[0075] therefore, Figure 1 A device 24 for tooth rolling test is generally shown, wherein the device 24 is configured to perform the following method steps: performing a rolling test on tooth 2 by single tooth surface rolling test and / or double tooth surface rolling test; generating an order spectrum 10 from the measurement data of the rolling test on tooth 2; and analyzing the order spectrum by a computer-implemented classification 12, wherein the classification is configured to determine the tooth deviation IX of tooth 2 and the test table deviations XI and XII of rolling test tables 6 and 8.

Claims

1. A method comprising the following method steps: - The rolling test of the tooth (2) is performed by a rolling test stand (6, 8) for single-tooth rolling test and / or double-tooth rolling test; - An order spectrum (10) is generated from the measurement data of the rolling test of the teeth (2). Its features are, - The order spectrum (10) is analyzed by a computer-implemented classification (12), wherein the classification (12) is set up to determine the tooth deviation (IX) of the tooth (2) and the test bench deviation (XI) of the rolling test bench (6, 8); and - The manufacturing method of adjusting the tooth (2) based on the determined tooth deviation (IX) and / or the rolling test table (6,8) based on the determined test table deviation (XI, XII) of the rolling test table (6,8).

2. The method according to claim 1, characterized in that, The classification (12) is performed by an artificial intelligence model (16), such as a neural network (16) or the like, which is trained based on training data.

3. The method according to claim 2, characterized in that, Regarding the classification of multiple categories of indicator tooth deviations (IX), the training data has a variation process of at least one order amplitude and / or at least one order spectrum for each of these categories, the amplitude and / or variation process having the characteristics of one tooth deviation (IX) or multiple tooth deviations (IX), wherein these categories are different from each other in terms of at least one assigned tooth deviation (IX).

4. The method according to claim 2 or claim 3, characterized in that, - At least one category of the indicated test bench bias (XI, XII) for the classification, wherein the training data has a variation process of at least one order amplitude and / or at least one order spectrum, the amplitude and / or variation process having the characteristics of one test bench bias (XI, XII) or multiple test bench biases (XI, XII); and / or - At least one category of the indicated test bench deviation (XI, XII) for the classification, wherein the training data has a variation process of at least one order amplitude and / or at least one order spectrum, the amplitude and / or variation process not having the characteristics of one or more tooth deviations.

5. The method according to any one of claims 2-4, characterized in that, - The artificial intelligence model (16) is trained on training data that optimizes the classification of dental deviations (IX) in terms of the noise characteristics of the teeth, especially the psychoacoustic characteristics; and / or - The artificial intelligence model (16) is trained based on training data that optimizes the classification of tooth deviations (IX) in terms of tooth lifespan; and / or - The artificial intelligence model (16) is trained based on training data that optimizes the classification of tooth deviations (IX) in terms of tooth efficiency.

6. The method according to any one of claims 2-5, characterized in that, Multiple artificial intelligence models (16) are set up, which are trained based on training data that can optimize the classification of tooth deviations in one or more of the following aspects: - Noise characteristics of the teeth (2), especially psychoacoustic characteristics; - Service life of the teeth (2); - Efficiency of the tooth section (2).

7. The method according to any one of claims 2-6, characterized in that, After adjusting the manufacturing method and / or maintenance test bench (6, 8) of the tooth (2), rolling tests and analyses are performed again on the same or another identical tooth to identify the effectiveness or ineffectiveness of the adjustment and / or maintenance, wherein the data on the effectiveness or ineffectiveness of the adjustment and / or maintenance are used as additional training data for the artificial intelligence model (16) to improve the artificial intelligence model (16).

8. The method according to any one of claims 2-7, characterized in that, Training data, AI models, and tuning schemes to be assigned to rolling test benches with the same structure are collected and stored on the server.

9. The method according to any one of the preceding claims, characterized in that, Based on the results of the analysis, predefined action suggestions or predefined evaluations are output to operators, especially in text form.

10. An apparatus for rolling test of teeth, wherein, The apparatus is configured to perform the following method steps: performing a rolling test on a tooth by means of a single-tooth-surface rolling test and / or a double-tooth-surface rolling test; generating an order spectrum from the measurement data of the rolling test on the tooth; and analyzing the order spectrum by means of a computer-implemented classification, wherein the classification is configured to determine the tooth deviation and the test bench deviation of the rolling test bench.

Images