A method for graded assembly and quality control of key components of a motorcycle engine
By calculating the predicted response characteristics of candidate fit combinations for motorcycle engines and matching them with the measured assembly process response curves, an individualized assembly response benchmark is generated. This solves the problem of identifying the correlation between the selection status of key components and the actual assembly quality in motorcycle engine assembly, and improves the accuracy and reliability of assembly quality control.
Patent Information
- Application Number
- CN202611260088.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-19
- Publication Date
- 2026-11-17
AI Technical Summary
Existing motorcycle engine assembly processes cannot effectively identify the correlation between the selection status of key components and the actual assembly quality, resulting in the ability to pass subsequent processes even when there are internal mismatches. Furthermore, the general quality benchmarks for fixed models are difficult to reflect the expected fit of the actual assembly.
By acquiring engine model identification and key component identity information, the predicted response characteristics of candidate fit combinations are calculated, individualized assembly response benchmarks are generated, and the measured assembly process response curves are matched with the predicted response characteristics. The comparison results between the estimated combination and the target combination are identified, and a release status is generated to control the permissions of subsequent processes.
It enables graded assembly and quality control of key components of motorcycle engines, reduces the impact of normal dimensional fluctuations on assembly quality judgment, identifies mis-graded assembly or failure to achieve the expected fit, and reduces the possibility of engines with abnormal internal fit obtaining the authority to proceed to subsequent processes.
Smart Images

Figure CN123113728A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of motorcycle engine assembly technology, and in particular to a method for graded assembly and quality control of key components of a motorcycle engine. Background Technology
[0002] Motorcycle engines typically include crankshafts, bearings, pistons, cylinder blocks, cylinder heads, timing mechanisms, and corresponding fasteners. Different engine models have varying key component models, dimensional tolerances, fit clearances, and assembly requirements. To reduce the machining costs of critical mating components, existing assembly processes usually allow for the production of components of the same model within a certain dimensional tolerance range. Appropriate components are selected for assembly based on nominal specifications, size grades, or operator experience. During assembly, pressing equipment is used to record pressing force and displacement, or tightening tools are used to record the final torque and final rotation angle. The assembly result is then judged according to a pre-set fixed threshold for the corresponding engine model.
[0003] However, even within the same machine model, the actual dimensions of the installed bearings, crankshafts, cylinder blocks, pistons, gaskets, and connected parts still exhibit normal dimensional fluctuations. Different combinations of components result in varying actual interference fits, clearances, axial clearances, and stack heights, thus causing corresponding changes in normal pressing force, displacement response, torque, and angular response. If a wider general acceptable range is set for the same machine model, process responses resulting from misaligned assembly, incorrect gasket selection, or components not forming the expected fit will still fall within this general acceptable range. Conversely, narrowing the general acceptable range will lead to the judgment of abnormal response differences caused by normal dimensional fluctuations.
[0004] Furthermore, existing component selection records, assembly process curves, and post-assembly inspection results are typically assessed separately. Fixed, universal quality benchmarks for specific engine models cannot reflect the process response that the actual selected combinations should produce for this particular engine. Therefore, it is difficult to verify whether the recorded combinations truly achieve the expected fit using the actual process response. Consequently, the existing process cannot correlate the selection status of key components with actual assembly quality, leading to the problem of gaining access to subsequent processes even when internal fit anomalies occur. Summary of the Invention
[0005] To address the aforementioned problems in the existing technology, this invention provides a graded assembly and quality control method for key components of a motorcycle engine. This method solves the problem that a fixed model's universal quality benchmark cannot correlate the selection status of key components with the actual assembly response, making it difficult to identify and prevent internal misalignment anomalies in a timely manner.
[0006] This invention provides a method for graded assembly and quality control of key components of a motorcycle engine, comprising: obtaining the model identifier and unique identifier of the engine to be assembled, calling the corresponding assembly process package, and obtaining the identity information and measured dimensions of the key components to be assembled; forming candidate fit combinations based on the measured dimensions, and calculating the predicted fit state value of each candidate fit combination; generating predicted response features of each candidate fit combination based on the measured dimensions, predicted fit state value, and basic response benchmark in the assembly process package; selecting a target combination from candidate fit combinations whose predicted fit state values are all within the corresponding target fit interval, and using the predicted response features of the target combination to form an individualized assembly response benchmark; assembling key components according to the target combination, collecting assembly process response curves and cross-process detection results used to characterize the fit state after assembly, matching the assembly process response curves with the predicted response features of each candidate fit combination to determine a presumed combination; comparing the presumed combination with the target combination, and generating a release state based on the comparison results, the consistency results of the assembly process response curves with the individualized assembly response benchmark, and the cross-process detection results, and controlling the permissions of subsequent processes according to the release state.
[0007] Preferably, each measured dimension is corrected for a reference temperature, and key components are divided into size grades according to the corrected measured dimensions. Candidate mating combinations are then formed from key components of different size grades. For the candidate mating combinations... c The j A coordination chain, predicted coordination state value for: ; When the measurement errors of each measured dimension and the dimensions of the compensation component are independent of each other, the combined standard uncertainty of the predicted fit state value is... for: ; And the predicted range for the coordination state is formed as follows: ; in, c Number the candidate combination combinations. j To match the chain number, n j For the first j The number of dimension items contained in a mating chain Candidate combination c The Middle i A measured dimension corrected for a reference temperature. For this measured dimension in the th... j The dimensions in each mating chain are transmitted in the direction of the chain and take a value of +1 or -1. Candidate combinationc In the j The dimensions of the compensating component in a mating chain, when no compensating component is provided. Take zero, and They are respectively and Standard uncertainty The candidate mate combination is defined as the mate state prediction interval. When all the mate state prediction intervals of a candidate mate combination are contained within the corresponding target mate interval, the candidate mate combination is determined as a valid candidate mate combination.
[0008] Preferably, for each valid candidate fit combination, the calculated fit center deviation is... D c Boundary margin M c and response discernibility R c Wherein, the mating center deviation D c The boundary margin is the normalized cumulative deviation of the predicted kinetic state value of each kinetic chain from the center value of the corresponding target kinetic interval. M c The response discriminability is the minimum normalized distance from the predicted interval of each coordination state to the boundary of the corresponding target coordination interval. R c The minimum normalized distance between the predicted response features of this effective candidate combination and the predicted response features of other effective candidate combination combinations; combination matching loss. J c for: ; in, D c , M c and R c The values of are all in the range [0,1]. w D , w M and w R These are the weights for fit center deviation, boundary margin, and response identifiability, respectively, and all are non-negative, summing to 1; when there is only one valid candidate fit combination, the response identifiability... R c Set to 1; adjust the combined matching loss. J c The smallest effective candidate combination is determined as the target combination.
[0009] Preferably, the predicted response features include the press-fitting force and displacement prediction curves and their individualized upper and lower envelopes; candidate fit combinations. c The predicted curves for pressing force and displacement are as follows: ; The corresponding individualized upper envelope and individualized lower envelope are as follows: , ; in, x The press-fit displacement is based on the press-fit contact point. Candidate combination c During press-fit displacement x Predicted pressing force at the location, The basic press-fit force and displacement curves in the basic response benchmark. Candidate combination c The predicted over-profit, This is the reference interference corresponding to the basic press-fit force and displacement curve. The interference fit is a displacement sensitivity coefficient related to the press-fit force. and Candidate combination c The individualized upper envelope and individualized lower envelope, Based on the allowable fluctuation amount, This is the displacement-related correction factor for the interference uncertainty on the envelope width. Candidate combination c The combined standard uncertainty of the predicted interference; the sensitivity coefficient and correction coefficient are determined by press-fitting data of qualified samples or calibration parts of the same model.
[0010] Preferably, the predicted response features include torque and angle prediction feature intervals; the generation process of the torque and angle prediction feature intervals includes: determining the actual stack height based on the measured dimensions of the connected parts in the candidate fit combination, and determining the actual equivalent compression based on the gasket grade; obtaining the reference stack height and reference equivalent compression corresponding to the basic response reference, and correcting the contact point angle in the basic response reference using the difference between the actual stack height and the actual equivalent compression and the difference between the reference stack height and the reference equivalent compression; correcting the slope of the elastic fastening segment after fitting and the target torque corresponding angle based on the bolt specifications, gasket grade, and tightening sequence; and combining the corrected contact point angle, elastic fastening segment slope, and target torque corresponding angle with the allowable intervals of the idle stroke angle, torque and angle integral, final torque, and final angle to constitute the torque and angle prediction feature intervals.
[0011] Preferably, assembling key components according to the target combination includes: reading the identification information and size level of the key component when it enters the assembly station, and verifying the reading results with the target combination; sending an unlock command to the assembly tool only when the key component model matches the process version, the key component's identification information and size level both belong to the target combination, the preceding process on which the target combination depends is in a valid completed state, and the assembly station has been loaded with an individualized assembly response benchmark; when any verification result is not satisfied, keeping the assembly tool locked, and associating the unsatisfactory verification result with the engine's unique identifier.
[0012] Preferably, the assembly process response curve includes a pressing force and displacement curve and a torque and angle curve. The contact position is determined and recorded from the original pressing force and displacement curve. Using the contact position as the segmentation benchmark, the pressing force and displacement curve is sequentially divided into an idle stroke segment, a contact establishment segment, a stable pressing segment, and a final positioning segment. The pressing force rise slope, average pressing force of the stable pressing segment, pressing work, final pressing force increment, and positioning displacement are extracted from each segment. The contact point angle is determined and recorded from the original torque and angle curve. Using the contact point as the segmentation benchmark, the torque and angle curve is sequentially divided into a screw-in segment, a contact segment, an elastic fastening segment, and a final holding segment. The idle stroke angle, elastic fastening segment slope, target torque corresponding angle, torque and angle integral, final torque, and final angle are extracted from each segment. The contact position, contact point angle, and curve features extracted from each segment constitute a measured response feature vector. .
[0013] Preferably, the matching of the assembly process response curve with the predicted response features includes: determining whether the pressing force, pressing displacement, torque and rotation angle in the assembly process response curve exceed the hard quality boundary specified in the assembly process package; if any one of them exceeds the hard quality boundary, a hard quality boundary overrun result is generated. When none of the parameters exceed the hard quality boundary, the measured response feature vector... The predicted response feature vector of candidate combination c Matching distance between d c for: ; The proposed combination is determined as follows: ; in, m In response to the number of features, r For the response feature number, z r The measured response feature vector The r Each response feature To predict the response feature vector The r-th predicted response feature in the data. For the first r The allowable fluctuation scale of each response feature is greater than zero. For the first r Each response feature has a non-negative weight, and the sum of the weights of all response features is 1. For the set of candidate matching combinations, c For candidate combinations, The estimated combination is used; the matching distance corresponding to the target combination is recorded as the assembly response consistency quantity; when the estimated combination is different from the target combination, an abnormal combination result is generated.
[0014] Preferably, the cross-process inspection results include at least one of the inspection results for crankshaft axial clearance, crankshaft rotational resistance, and timing phase; generating the release state includes: when all parameters are within the hard quality boundary, the estimated combination is the same as the target combination, and the matching distance corresponding to the target combination is... When the data is not greater than the first consistency threshold, the necessary process data is complete, and the test results of each cross-process are all within the corresponding qualified range, a direct release status is generated. When all of the following basic conditions are met and at least one review trigger condition is met, a review status is generated, and at least one of the crankshaft axial clearance, crankshaft rotational resistance, and timing phase is re-tested. The basic conditions include: all parameters are within the hard quality boundary, the estimated combination is the same as the target combination, and the matching distance corresponding to the target combination is [not specified]. The verification triggering conditions include: the matching distance corresponding to the target combination; the data of the necessary process steps are complete; and the detection results of each cross-process step do not exceed the corresponding preset allowable limits. The result is greater than the first consistency threshold, or at least one cross-process test result is within the corresponding review interval; when a hard quality boundary is exceeded, the presumed combination differs from the target combination, or the matching distance corresponding to the target combination is exceeded. A quality lockout state is generated when the result exceeds the second consistency threshold, any cross-process inspection result exceeds the corresponding preset allowable limit, or necessary process data is missing; among these, For the target combination, the first consistency threshold is less than the second consistency threshold; the assembly process package pre-sets qualified intervals, verification intervals, verification qualified intervals and preset allowable limits for each cross-process inspection result. The qualified intervals and verification intervals do not overlap, and any cross-process inspection result can only be judged as being in the qualified interval, in the verification interval or exceeding the preset allowable limit; the engine in the verification state is switched to the direct release state when all specified verification inspection results are in the corresponding verification qualified interval, and switched to the quality lock-up state when any specified verification inspection result is not in the corresponding verification qualified interval.
[0015] Preferably, the assembly process package further includes quality dependencies, which characterize the dependency direction between key components, mating chains, target combinations, individualized assembly response benchmarks, assembly process response curves, cross-process inspection results, and release status. When an assembled key component is disassembled or replaced, the affected mating chain is determined based on the disassembled or replaced key component, and along the dependency direction of the quality dependencies, the target combination, individualized assembly response benchmark, assembly process response curve, cross-process inspection results, and release status corresponding to the affected mating chain are marked as failed, while the corresponding subsequent process permissions are locked. The identity information and measured dimensions of the reinstalled or replaced key component are obtained, the target combination and individualized assembly response benchmark are redefined, and the affected assembly and inspection processes are re-executed. The subsequent process permissions are restored only after all failed items have regenerated qualified results.
[0016] The beneficial effects of this invention are as follows: This invention generates predicted response features for each candidate fit combination by utilizing the measured dimensions and predicted fit states of key components, and uses the predicted response features of the target combination to form an individualized assembly response benchmark. This allows the assembly quality benchmark to change accordingly with the actual installed components and their fit states, reducing the impact of normal dimensional fluctuations on assembly quality judgment.
[0017] The measured assembly process response curves are further matched with the predicted response characteristics of each candidate fit combination to determine the estimated combination that is closest to the actual response. The estimated combination is then compared with the target combination. The assembly process response is used to back-verify the recorded component combination and identify mis-grade assembly or failure to form the expected fit state within the general qualified range of the model.
[0018] Then, by combining the comparison results, assembly response consistency, and cross-process inspection results to generate the release status, the component selection, process assembly, and post-assembly inspection form a related quality chain, reducing the possibility that an engine with a single final value that is qualified but has internal mismatch will obtain the authority for subsequent processes.
[0019] Furthermore, during the selection process, the center deviation, boundary margin, and response identifiability are considered simultaneously. From multiple combinations that meet the target fit range, the target combination is selected that has both fit quality margin and is easy to identify through subsequent assembly response.
[0020] This invention addresses the disassembly or replacement of components by invalidating relevant quality data and release status along quality-dependent relationships, thus preventing the continued use of pre-repair quality data after motorcycle engine repair. Attached Figure Description
[0021] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0022] Figure 1 This is a flowchart of the graded assembly and quality control method for key components of a motorcycle engine according to the present invention. Figure 2 This is a flowchart illustrating the generation of candidate mating combinations, selection of target combinations, and generation of individualized assembly response benchmarks for the present invention. Figure 3 This is a flowchart for determining the segmentation and estimated combination of the assembly process response curve in the present invention. Figure 4 This is a schematic diagram illustrating the quality dependency relationship and rework-related failure control of the present invention. Detailed Implementation
[0023] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the specific implementation methods, features, and effects of the present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments. The specific dimensions, thresholds, and weights mentioned below are used to illustrate the implementation process of the present invention and are calibrated or set according to engine type, component materials, machining tolerances, equipment precision, and quality control requirements; the following values are calculated parameters.
[0024] Example 1 See Figure 1 As shown, this embodiment provides a graded assembly and quality control method for key components of a motorcycle engine. This embodiment is used to illustrate the basic implementation method of establishing an assembly response benchmark based on the actual component assembly, estimating the component assembly using the actual assembly response, and controlling quality release accordingly.
[0025] First, when an engine to be installed enters the critical component assembly section, its model identification and unique engine identifier are read. The model identification is used to distinguish the engine product model and process version, while the unique engine identifier is used throughout the component selection, assembly, testing, rework, and release processes. The unique engine identifier includes the engine serial number.
[0026] The corresponding assembly process package is invoked based on the model identifier. The assembly process package records at least the types of key components that can be used, the target mating range, the basic response benchmark, the assembly sequence, the assembly tool program, the process quality boundary, and the permission conditions for subsequent processes, and is version-managed according to the model and process version.
[0027] Critical components are those whose dimensional combinations can affect the internal fit or assembly response of the engine, including one or more of the following: crankcase, bearings, crankshaft, pistons, cylinder block, adjusting shims, connected parts, and bolts. The identification information and measured dimensions of each critical component to be installed are obtained. The identification information includes at least one of the following: unique component number, batch number, model, manufacturing process version, and size class.
[0028] Multiple candidate fit combinations are formed based on the available key components and their measured dimensions. Each candidate fit combination is associated with a specific component to be installed in the engine. For each candidate fit combination, a predicted fit condition is calculated using the dimensional transfer relationships between the relevant component dimensions. The predicted fit condition includes at least one of the following: interference fit between the bearing outer ring and bearing bore; fit between the crankshaft journal and bearing inner ring; clearance between the piston and cylinder bore; crankshaft axial clearance; and stack height of fasteners.
[0029] The target combination is selected from candidate fit combinations whose predicted fit states fall within the target fit range specified by the assembly process package. For each candidate fit combination, a predicted response feature is generated based on the measured dimensions, predicted fit state, and baseline response. The predicted response feature describes the assembly process response that the corresponding candidate fit combination should produce when assembly is completed normally.
[0030] Once the target combination is selected, its predicted response characteristics form the individualized assembly response benchmark for the engine. Therefore, different engine instances of the same model correspond to different individualized assembly response benchmarks due to the different measured dimensions of the selected components.
[0031] The assembly station performs assembly according to the target combination and collects the assembly process response curves. For the press-fitting process, the assembly process response curve reflects the change of press-fitting force with press-fitting displacement; for the bolt tightening process, the assembly process response curve reflects the change of torque with rotation angle. After assembly, cross-process inspection results characterizing the actual fit state are obtained.
[0032] The measured assembly process response curves are matched with the predicted response characteristics of each candidate fit combination, and the candidate fit combination with the smallest matching distance is determined as the presumed fit combination. The presumed fit combination is compared with the target fit combination recorded before assembly, and it is also determined whether the measured assembly process response curve is consistent with the individualized assembly response benchmark of the target fit combination. The release status is generated by combining the cross-process inspection results.
[0033] Therefore, the target combination is not only used to guide the selection of components, but also to determine the response that should be generated during the assembly process; the actual assembly response is used to verify whether the target combination has truly formed the expected fit state, and the cross-process test results further verify the actual state after assembly, forming a closed loop of quality release that is related to the preceding and following processes.
[0034] Example 2 See Figure 2 As shown in Example 1, if the determination of whether a candidate combination is within the target fit range is based solely on the measured dimensional values, then for components close to the dimensional grade boundary or the target fit range boundary, measurement repeatability, gauge error, and temperature changes cause the actual fit state to exceed the target fit range. Therefore, this example transfers the measurement uncertainty to the fit state prediction range.
[0035] For the measured temperature is t Actual measured dimensions Correct to the reference temperature according to the following formula. t 0: ; in, X i Corrected dimensions at the reference temperature. For the first i The coefficient of linear expansion of the materials of each component.
[0036] For candidate combinations c The j For each coordination chain, the predicted coordination state value is calculated according to the following formula: ; When the measurement errors of each input dimension are independent, the combined standard uncertainty is: ; This forms the prediction interval for the coordination state: ; When measurement errors of different sizes are correlated, the corresponding covariance term is added inside the square root to avoid underestimating the uncertainty of the predicted value of the fit state.
[0037] The following example uses crankshaft axial clearance. Candidate fit combinations. c In equation 1, the temperature-corrected crankcase fit space dimension is 58.020 mm, the crankshaft assembly width is 57.740 mm, and the adjusting shim thickness is 0.120 mm. The standard uncertainties for these three dimensions are 0.008 mm, 0.006 mm, and 0.004 mm, respectively. Therefore: mm; mm; therefore: mm; When the target fit interval is [0.140, 0.180] mm, the predicted interval is entirely contained within the target fit interval; therefore, the candidate fit combinations... c1 is a valid candidate combination.
[0038] When multiple valid candidate mate combinations exist, selecting only the combination closest to the target center can lead to choosing combinations with insufficient boundary margins or whose predicted responses are too similar to other combinations, thus reducing the ability to identify correct and incorrect assemblies in subsequent processes. Therefore, further calculations are performed on the mate center deviation, boundary margin, and response identifiability.
[0039] Let the first j The target coordination interval is: Its central value is Half width is The lower and upper limits of the candidate combination prediction interval are respectively and One normalization method is: ; ; in, q The number of cooperating chains participating in the evaluation.
[0040] Let the normalized distance between the predicted response feature vectors of different candidate combinations be . The preset identifiable reference distance is The identifiability of the response is then expressed as: ; When there is only one valid candidate combination R c Take 1.
[0041] Pick w D =0.4、 w M =0.3、 w R =0.3. For candidate combinations c 1. Let ,
[0042] , ,but: ; For candidate combinations c 2. Let , , ,but: ; because Select candidate combinations c1 is chosen as the target combination. This selection takes into account both fit center deviation, boundary margin, and the identifiability of subsequent process responses.
[0043] Example 3 After selecting the target combination in Example 2, different target combinations of the same model may still produce different normal pressing forces due to different actual interference fits. If a fixed pressing force and displacement envelope are used, it is easy to misjudge normal combinations or miss abnormal combinations. To address this, this example generates pressing force and displacement prediction curves and their individualized upper and lower envelopes based on the predicted interference fits of candidate fit combinations.
[0044] Candidate combination c The predicted curves for pressing force and displacement are as follows: ; The individualized upper envelope and the individualized lower envelope are respectively: ; On multiple calibration parts with known interference fits, the press-fit curves were aligned according to the contact points, and sampling points were taken at each discrete displacement. x k The regression is performed according to the following formula: ; in, l For calibration part number, For the first l Each calibration component is in x k The pressing force at the point, Given its known over-excess amount, For the regression intercept, The residuals are used to obtain the displacement correlation sensitivity coefficient based on the regression results. .
[0045] Basic allowable volatility The correction coefficient is determined based on the preset quantile values of the regression residuals. One implementation method for converting the uncertainty of the predicted interference amount into an envelope expansion amount in the direction of the press-fit force is as follows: ; in, k u The coverage factor is set according to the envelope coverage requirements.
[0046] For example, in press-fit displacement x At 10mm, the reference interference. mm, basic pressing force kN. When using calibration parts with interference fits of 0.025mm, 0.030mm, and 0.035mm, the corresponding average pressing forces are 4.30kN, 4.82kN, and 5.31kN, respectively. Therefore, the sensitivity coefficient at this displacement is: kN / mm; For the predicted over-excess amount Candidate mating combinations for mm: kN; like kN, mm, coverage factor k u =2, then kN / mm, with corresponding upper and lower envelopes as follows: kN, kN; Performing the same calculations at each sampling point throughout the pressing process will generate an individualized upper and lower envelope corresponding to the target combination.
[0047] Example 4 When the gasket grade, the stack height of the connected parts, or the bolt specifications in the target assembly change, the screwing and fitting process may still differ even if the final torque is the same. Therefore, generating the torque and angle prediction feature range corresponding to the target assembly is particularly important in this embodiment.
[0048] The actual stack height is calculated based on the measured thickness of each connected component in the candidate mating combination. H c The actual equivalent compression is determined based on the actual gasket grade. q c Obtain the reference stack height from the basic response reference. H 0. Baseline equivalent compression q 0 and reference bonding point angle .
[0049] The actual effective stack height is The baseline effective stack height is The predicted corner of the bonding point is: ; in, k H This is the rotation correction coefficient corresponding to the effective stack height change.
[0050] For example, H 0 = 32.00 mm q 0 = 0.01 mm , H c=32.06mm, q c =0.02mm, ,but: ; Based on this, the slope of the elastic tightening segment and the corresponding rotation angle of the target torque are corrected according to the bolt specifications, gasket grade and tightening sequence. Combined with the idle stroke rotation angle, torque and rotation angle integral, and the allowable range of final torque and final rotation angle, the torque and rotation angle prediction characteristic range is formed.
[0051] The correction amounts for the slope of the aforementioned elastic fastening segment and the rotation angle corresponding to the target torque are pre-collected using qualified samples or calibration parts of the same model, grouped according to bolt specifications, gasket grades, and tightening sequence. Torque and rotation angle curves are collected for each group of samples. The elastic fastening segment after each curve is fitted is then fitted to obtain the correction amount for the slope of the elastic fastening segment. The rotation angle at which the target torque is reached is also statistically analyzed to obtain the correction amount for the rotation angle corresponding to the target torque. Each group of correction amounts or correction coefficients is stored according to the process version of the assembly process package. When generating the predicted response characteristics of candidate fit combinations, the corresponding correction amounts or correction coefficients are retrieved according to the bolt specifications, gasket grades, and tightening sequence corresponding to that candidate fit combination.
[0052] Example 5 Even after determining the target combination and individualized assembly response benchmark according to Examples 1 to 4, it is still possible for parts of the same model but different size grades to enter the assembly station; during assembly, only the part model is checked, and parts of incorrect size grades may still be assembled incorrectly. To address this, this embodiment performs station authorization before the assembly tools are unlocked.
[0053] Once critical components arrive at the assembly station, their identity information and size class are read and verified against the target combination bound to the engine's unique identifier. The unlocking conditions for assembly tools include: the component model matches the current assembly process package; the component process version matches the assembly process package version; both the component's identity information and size class belong to the target combination; the preceding processes that the target combination depends on are in a valid completion state; and the current assembly station has loaded the individualized assembly response benchmark corresponding to the target combination.
[0054] Only when all the above conditions are met will the workstation controller of the control system overseeing the automated production process send an unlocking command to the pressing equipment, tightening tools, or tooling fixtures. If any condition is not met, the assembly tools will remain locked, and the engine's unique identifier, the identity of the abnormal component, the target assembly requirements, the actual reading results, the workstation, and the time will be recorded.
[0055] Example 6 See Figure 3As shown, due to differences in the initial loading position, tool idle stroke, and initial screw-in position of different workpieces, directly comparing the initial position offset of the unprocessed complete curve will affect the matching result. Therefore, this embodiment saves the contact position and contact point angle from the original curve, and then segments the curve based on the corresponding positions.
[0056] For the original pressing force and displacement curve, the location where the pressing force exceeds the contact threshold and the rate of change of pressing force is consistently positive at multiple consecutive sampling points is determined as the contact position and saved. Subsequently, based on the contact position, the curve is divided into an idle stroke segment, a contact establishment segment, a stable pressing segment, and a terminal positioning segment.
[0057] The following parameters were extracted from each stage: contact position, pressure rise slope, average pressure in the stable pressure section, pressure work, final pressure increment, and displacement at the end of the pressure process. The pressure work is as follows: ; For the original torque and angle curves, the position where the torque continuously exceeds the bonding threshold and the subsequent torque growth rate reaches the preset condition is determined as the bonding point. The corresponding bonding point angle in the original curve is saved. Then, based on the bonding point as the segmentation benchmark, the curve is divided into the screw-in segment, bonding segment, elastic fastening segment, and terminal holding segment.
[0058] The following parameters are extracted from each stage: idle stroke angle, contact point angle, slope of the elastic fastening section, angle corresponding to the target torque, torque and angle integral, final torque, and final angle. The torque and angle integral are: ; The above features are arranged in the same order as the predicted response features to form the measured response feature vector: ; By first saving the original contact position and the corner of the bonding point, and then implementing segmentation, the problem of the contact position or the corner of the bonding point being always zero after curve alignment is avoided.
[0059] Example 7 Even if the pressing force, pressing displacement, final torque, and final rotation angle do not exceed the safety boundaries, the fit state reflected by the actual curve may still be closer to another candidate fit combination. Therefore, this embodiment first performs a hard quality boundary judgment, and then determines the matching distance between the measured response and each candidate fit combination.
[0060] When the hard quality boundary is not exceeded, the matching distance is calculated as follows: ; and from the candidate combination set Determine the presumed combination: ; The following is the calculation process using four response features. Let: ; Candidate Combinations c 1 and c The predicted response feature vectors for 2 are as follows: , ; The permissible fluctuation scale is: ; If the weights of all four features are 0.25, then: ; ; Therefore, the presumed combination for c 1. If the target combination selected before assembly is also c 1. If the combination is consistent, then the target combination recorded before assembly is... c 2. Even if each final value is still within the hard quality boundary, a combined anomalous result will still be generated.
[0061] Example 8 If the above embodiments only set two states, "qualified" and "unqualified," engines approaching the consistency threshold or crossing the inspection boundary may experience unnecessary rework or be released directly without verification during production. Therefore, this embodiment sets up three states: direct release, verification, and quality lockout.
[0062] Cross-process inspection results include at least one of the following: crankshaft axial clearance, crankshaft rotational resistance, and timing phase inspection results. Let the target combination be... The first consistency threshold is d 1. The second consistency threshold is d 2, and d 1< d 2.
[0063] When all parameters are within the hard quality boundaries, and the assumed combination is the same as the target combination, When the necessary process data is complete and the test results of each cross-process are within the corresponding qualified range, a direct release status is generated.
[0064] When all parameters are within the hard quality boundaries, and the assumed combination is the same as the target combination, The necessary process data is complete, and the test results across each process do not exceed the corresponding preset allowable limits, and meet the requirements. Alternatively, a review status is generated when at least one cross-process test result falls within the corresponding review interval.
[0065] When hard quality boundaries are exceeded, or the putative combination differs from the target combination, When any cross-process inspection result exceeds the corresponding preset allowable limit or when necessary process data is missing, a quality lockout state is generated.
[0066] The qualified range, verification range, verification qualified range and preset allowable limit corresponding to the inspection results of each cross-process are pre-stored in the assembly process package; the qualified range and verification range do not overlap, and any cross-process inspection result can only be judged as being in the qualified range, in the verification range or exceeding the preset allowable limit.
[0067] For example, setting d 1 = 0.35 d 2 = 0.75. If the target combination is... c 1. The matching distance is 0.284 mm, the measured crankshaft axial clearance is 0.160 mm, the crankshaft rotational resistance is 0.32 N·m, and the timing phase deviation is... If all three items are within the corresponding qualified range, then a direct release status will be generated.
[0068] In another example, if the target combination is the same as the expected combination, the matching distance is 0.52, or the first measured crankshaft axial clearance is 0.179 mm and enters the verification range, then a verification state is generated. After cleaning the measurement reference again and loading in the specified direction, if the crankshaft axial clearance is measured again as 0.163 mm, and all specified verification items enter the corresponding verification pass range, then the verification state is changed to the direct release state; if any specified verification item is still unqualified, it is changed to the quality lock-up state.
[0069] Example 9 See Figure 4 As shown, after an engine is determined to be abnormal and critical components are disassembled or replaced, the dimensional conditions corresponding to the original target assembly and the original individualized assembly response benchmark have changed. If the pre-rework pass curves, inspection results, or release status are still retained, the reworked engine may gain access to subsequent processes based on the failure data. To address this, this embodiment sets quality dependencies in the assembly process package.
[0070] Quality dependencies are represented using directed relation tables or directed graphs. Nodes include critical components, mating chains, target combinations, individualized assembly response benchmarks, assembly process response curves, cross-process inspection results, and release status; directed edges indicate that a change in one node will cause the failure of the next node.
[0071] For example, the bearing identity and measured dimensions point to the bearing outer ring interference fit chain; this fit chain points to the target combination and the individualized assembly response benchmark for press-fit force and displacement; the target combination and the individualized assembly response benchmark together point to the bearing press-fit process response curve; the bearing press-fit process response curve and the target combination point to the crankshaft axial clearance and crankshaft rotational resistance test results; the above test results and combination consistency results together point to the release status.
[0072] When a bearing is disassembled and reinstalled or replaced during the rework process, the affected mating chain is determined based on the bearing node. Then, along the dependent direction, the original target combination, the original individualized assembly response benchmark, the original press-fit curve, the original crankshaft axial clearance result, the original crankshaft rotational resistance result, and the original release status are marked as failures, and the permissions for subsequent processes are locked.
[0073] Obtain the identification information and measured dimensions of the reinstalled or replaced bearing, recalculate the predicted values and ranges of the fit condition, reselect the target combination, and generate an individualized assembly response benchmark; then re-execute the bearing press-fit and re-inspect the crankshaft axial clearance and crankshaft rotational resistance. Only after all failed items have obtained qualified results can the subsequent process permissions be restored.
[0074] After accumulating qualified sample data for the same model, the basic response benchmark is updated under restrictions. Only data that is in direct release status, has passed engine terminal inspection, and has not undergone rework is included in the benchmark update sample set. Before updating, it should also be verified that the stable pressing force does not decrease in the opposite direction when the predicted interference increases, and that the direction of change of the bonding point angle does not oppose the calibration relationship when the effective stack height increases. If the corresponding physical monotonicity is not met, the basic response benchmark is not updated and equipment or gauge calibration is triggered. Each update is limited to a preset range, and the process versions before and after the update are retained to avoid the basic response benchmark drift caused by abnormal samples.
[0075] In the above embodiments, the relevant calculations and status judgments are performed by a workstation controller, an industrial computer, a manufacturing execution system, or a combination thereof. The workstation controller is communicatively connected to dimensional measuring equipment, identification reading equipment, pressing equipment, tightening tools, displacement detection equipment, torque detection equipment, and phase detection equipment, thereby converting the data processing results into assembly tool unlocking, workstation release, verification, or locking control commands.
[0076] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any equivalent changes or modifications made by those skilled in the art based on the above disclosure without departing from the scope of the present invention should fall within the scope of the present invention.
Claims
1. A method for graded assembly and quality control of key components of a motorcycle engine, characterized in that, include: Obtain the engine model identifier and engine unique identifier of the engine to be installed, call the corresponding assembly process package, and obtain the identity information and actual measured dimensions of the key components to be installed; Candidate fit combinations are formed based on the measured dimensions, and the predicted fit state value of each candidate fit combination is calculated. Based on the measured dimensions, predicted values of fit states, and basic response benchmarks in the assembly process package of each candidate fit combination, predictive response features of each candidate fit combination are generated. From the candidate fit combinations in which each predicted value of fit state is located within the corresponding target fit range, a target combination is selected, and the predictive response features of the target combination constitute an individualized assembly response benchmark. Assemble key components according to the target combination, collect assembly process response curves and cross-process detection results used to characterize the fit state after assembly, and match the assembly process response curves with the predicted response characteristics of each candidate fit combination to determine the estimated combination. The estimated combination is compared with the target combination, and a release status is generated based on the comparison results, the consistency results of the assembly process response curve with respect to the individualized assembly response benchmark, and the cross-process detection results. The permissions of subsequent processes are controlled according to the release status.
2. The graded assembly and quality control method according to claim 1, characterized in that, Calculating the predicted fit state includes: correcting each measured dimension by a reference temperature, classifying key components into size levels according to the corrected measured dimensions, and forming candidate fit combinations from key components of different size levels; For candidate combinations c The j A coordination chain, predicted coordination state value for: ; When the measurement errors of each measured dimension and the dimensions of the compensation component are independent of each other, the combined standard uncertainty of the predicted fit state value is... for: ; And the predicted range for the coordination state is formed as follows: ; in, c Number the candidate combination combinations. j To match the chain number, n j For the first j The number of dimension items contained in a mating chain Candidate combination c The Middle i A measured dimension corrected for a reference temperature. For this measured dimension in the th... j The dimensions in each mating chain are transmitted in the direction of the chain and take a value of +1 or -1. Candidate combination c In the j The dimensions of the compensating component in a mating chain, when no compensating component is provided. Take zero, and They are respectively and Standard uncertainty To match the state prediction interval; When the predicted intervals of each commutation state of a candidate commutation combination are all contained within the corresponding target commutation interval, the candidate commutation combination is determined as a valid candidate commutation combination.
3. The graded assembly and quality control method according to claim 2, characterized in that, Selecting the target combination includes: calculating the fit center deviation for each valid candidate fit combination. D c Boundary margin M c and response discernibility R c Wherein, the mating center deviation D c The boundary margin is the normalized cumulative deviation of the predicted kinetic state value of each kinetic chain from the center value of the corresponding target kinetic interval. M c The response discriminability is the minimum normalized distance from the predicted interval of each coordination state to the boundary of the corresponding target coordination interval. R c The minimum normalized distance between the predicted response features of this effective candidate combination and the predicted response features of other effective candidate combination combinations; Combination matching loss J c for: ; in, D c , M c and R c The values of are all in the range [0,1]. w D , w M and w R These are the weights for fit center deviation, boundary margin, and response identifiability, respectively, and all are non-negative, summing to 1. When there is only one valid candidate combination, the response identifiability R c Set to 1; adjust the combined matching loss. J c The smallest effective candidate combination is determined as the target combination.
4. The graded assembly and quality control method according to claim 3, characterized in that, The predicted response features include the press-fitting force and displacement prediction curves and their individualized upper and lower envelopes; candidate fit combinations. c The predicted curves for pressing force and displacement are as follows: ; The corresponding individualized upper envelope and individualized lower envelope are as follows: 、 ; in, x The press-fit displacement is based on the press-fit contact point. Candidate combination c During press-fit displacement x Predicted pressing force at the location, The basic press-fit force and displacement curves in the basic response benchmark. Candidate combination c The predicted over-profit, This is the reference interference corresponding to the basic press-fit force and displacement curve. The interference fit is a displacement sensitivity coefficient related to the press-fit force. and Candidate combination c The individualized upper envelope and individualized lower envelope, Based on the allowable fluctuation amount, This is the displacement-related correction factor for the interference uncertainty on the envelope width. Candidate combination c The combined standard uncertainty of the predicted interference; the sensitivity coefficient and correction coefficient are determined by press-fitting data of qualified samples or calibration parts of the same model.
5. The graded assembly and quality control method according to claim 3, characterized in that, The predicted response features include torque and steering angle prediction feature ranges; The process of generating the torque and angle prediction feature range includes: determining the actual stack height based on the measured dimensions of the connected parts in the candidate fit combination, and determining the actual equivalent compression based on the gasket grade; Obtain the reference stack height and reference equivalent compression corresponding to the basic response reference. Use the difference between the actual stack height and the actual equivalent compression and the difference between the reference stack height and the reference equivalent compression to correct the mating point rotation angle in the basic response reference. Based on the bolt specifications, gasket grade, and tightening sequence, adjust the slope of the elastic fastening section after bonding and the rotation angle corresponding to the target torque; The corrected contact point angle, the slope of the elastic fastening section, and the angle corresponding to the target torque, together with the idle stroke angle, the torque and angle integral, and the allowable range of the final torque and final angle, constitute the torque and angle prediction feature range.
6. The graded assembly and quality control method according to claim 1, characterized in that, Assembling key components according to the target combination includes: reading the identification information and size grade of the key components when they enter the assembly station, and verifying the reading results with the target combination; An unlock command is sent to the assembly tool only when the model of the key component matches the process version, the identity information and size level of the key component belong to the target combination, the preceding process on which the target combination depends is in a valid completed state, and the assembly station has loaded the individualized assembly response benchmark. If any verification result is not met, the assembly tool is kept locked, and the unsatisfactory verification result is associated with the engine's unique identifier and recorded.
7. The graded assembly and quality control method according to claim 1, characterized in that, The assembly process response curves include pressing force and displacement curves and torque and rotation angle curves; The contact position is determined and recorded from the original pressing force and displacement curve. Using the contact position as the segmentation benchmark, the pressing force and displacement curve is divided into the idle stroke segment, the contact establishment segment, the stable pressing segment, and the terminal positioning segment. The pressing force rise slope, the average pressing force of the stable pressing segment, the pressing work, the terminal pressing force increment, and the positioning displacement are extracted from each segment. The contact point angle is determined and recorded from the original torque and angle curves. Using the contact point as the segmentation reference, the torque and angle curves are divided into the screw-in segment, the contact segment, the elastic fastening segment, and the terminal holding segment. The idle stroke angle, the slope of the elastic fastening segment, the angle corresponding to the target torque, the integral of torque and angle, the final torque, and the final angle are extracted from each segment. The contact position, the angle of the contact point, and the curve features extracted from each segment are used to construct the measured response feature vector. .
8. The graded assembly and quality control method according to claim 7, characterized in that, The matching of the assembly process response curve with the predicted response features includes: determining whether the pressing force, pressing displacement, torque and rotation angle in the assembly process response curve exceed the hard quality boundary specified in the assembly process package; if any one of them exceeds the hard quality boundary, a hard quality boundary overrun result is generated. When none of the parameters exceed the hard quality boundary, the measured response feature vector... The predicted response feature vector of candidate combination c Matching distance between d c for: ; The proposed combination is determined as follows: ; in, m In response to the number of features, r For the response feature number, z r The measured response feature vector The r Each response feature To predict the response feature vector The r-th predicted response feature in the data. For the first r The allowable fluctuation scale of each response feature is greater than zero. For the first r Each response feature has a non-negative weight, and the sum of the weights of all response features is 1. For the set of candidate matching combinations, c For candidate combinations, For the presumed combination; The matching distance corresponding to the target combination is recorded as the assembly response consistency quantity; when the presumed combination is different from the target combination, an abnormal combination result is generated.
9. The graded assembly and quality control method according to claim 8, characterized in that, The cross-process inspection results include at least one of the inspection results for crankshaft axial clearance, crankshaft rotational resistance, and timing phase; generating the release state includes: when all parameters are within the hard quality boundary, the estimated combination is the same as the target combination, and the matching distance corresponding to the target combination is... When the value is not greater than the first consistency threshold, the necessary process data is complete, and the cross-process test results are all within the corresponding qualified range, a direct release status is generated. When all of the following basic conditions are met and at least one verification trigger condition is met, a verification state is generated, and at least one of the crankshaft axial clearance, crankshaft rotational resistance, and timing phase is re-detected; the basic conditions include: all parameters are within the hard quality boundary, the estimated combination is the same as the target combination, and the matching distance corresponding to the target combination is... The verification triggering conditions include: the matching distance corresponding to the target combination; the data of the necessary process steps are complete; and the detection results of each cross-process step do not exceed the corresponding preset allowable limits. The result is greater than the first consistency threshold, or at least one cross-process test result is in the corresponding review interval; When hard quality boundaries are exceeded, the assumed combination differs from the target combination, or the matching distance corresponding to the target combination occurs... When the result exceeds the second consistency threshold, or when any cross-process test result exceeds the corresponding preset allowable limit, or when necessary process data is missing, a quality lockout state is generated. in, For the target combination, the first consistency threshold is less than the second consistency threshold; the assembly process package pre-sets a qualified range, a review range, a review qualified range and a preset allowable limit for each cross-process inspection result, the qualified range and the review range do not overlap, and any cross-process inspection result can only be judged as being in the qualified range, in the review range or exceeding the preset allowable limit; An engine in the review process will be put into direct release mode if all specified review test results are within the corresponding review pass range, and into quality lock-up mode if any specified review test result is not within the corresponding review pass range.
10. The graded assembly and quality control method according to claim 9, characterized in that, The assembly process package also includes quality dependencies, which are used to characterize the dependency direction between key components, mating chains, target combinations, individualized assembly response benchmarks, assembly process response curves, cross-process inspection results, and release status. When a critical component that has been installed is disassembled or replaced, the affected mating chain is determined based on the disassembled or replaced critical component. Along the dependency direction of the quality dependency relationship, the target combination, individualized assembly response benchmark, assembly process response curve, cross-process inspection results and release status corresponding to the affected mating chain are marked as failed, and the corresponding subsequent process permissions are locked. Obtain the identification information and measured dimensions of the key components after reinstallation or replacement, redetermine the target combination and individualized assembly response benchmarks, and re-execute the affected assembly and inspection processes; restore the permissions for subsequent processes only after all failed items have regenerated qualified results.
