A method for assembling stacked elastic units

By using a stacked elastic unit assembly method, data preprocessing and clustering algorithms are employed to optimize the selection and assembly of elastic unit groups, thereby solving the problems of poor stability and load-bearing capacity of elastic element groups and achieving higher stability and load-bearing capacity.

CN117390780BActive Publication Date: 2026-11-17713TH RES INST OF CHINA STATE SHIPBUILDING CORP LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202311116350.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-08-31
Publication Date
2026-11-17
Estimated Expiration
2043-08-31

AI Technical Summary

Technical Problem

The assembly of elastic element groups in the prior art lacks systematicity, resulting in poor stability and load-bearing capacity. Furthermore, the combination of multiple elastic elements is random, and there is a lack of reliable assembly methods.

Method used

A layered elastic unit assembly method is adopted. Through data preprocessing, clustering and index weight analysis, the K-medoids algorithm and Delphi method are used to optimize the selection and assembly of elastic unit groups to ensure performance consistency and stability.

Benefits of technology

It improves the stability and load-bearing capacity of the elastic element group, reduces the force required to complete the same action, and achieves a more uniform force distribution and stronger load-bearing capacity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117390780B_ABST
    Figure CN117390780B_ABST
Patent Text Reader

Abstract

This invention relates to a method for assembling stacked elastic unit cells, belonging to the field of quality classification. The method involves acquiring a pre-processed dataset of elastic unit cells, clustering the dataset according to a first preset value to obtain data clusters with the number of clusters equal to the first preset value, selecting a target data cluster, selecting data items from the target data cluster with the number of items equal to a second preset value, and forming an elastic unit cell group from the elastic unit cells corresponding to the selected data. The selected data is then deleted from the elastic unit cell dataset to obtain a new dataset. The current first preset value is subtracted by 1 to obtain a new first preset value. The above operation is repeated on the new dataset using the new first preset value until the number of data items in the largest data cluster of the new dataset is less than or equal to a preset threshold, at which point the selection of elastic unit cell groups stops. Each selected elastic unit cell group is then used to assemble different elastic body components, resulting in elastic body components with greater load-bearing capacity and stronger stability.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to a method for assembling a stacked elastic unit, belonging to the field of quality classification. Background Technology

[0002] Currently, flexible wrists are commonly used in some devices to compensate for misalignment between electrical plugs and sockets. The laminated elastomer involved in this invention is mainly used in the flexible wrists of these devices. The flexible wrist structure is as follows: Figure 1 , Figure 2 As shown, the flexible wrist's ability to compensate for misalignment between the plug and socket primarily relies on its elastic element assembly. The performance of the elastic element assembly directly determines the flexible wrist's compensation capability. Therefore, strictly controlling the performance of the elastic element assembly in the flexible wrist is essential.

[0003] An elastic element assembly consists of multiple elastic elements, so the overall performance of the assembly depends on the performance of individual elastic elements and the assembly performance of multiple elastic units. Current technologies primarily focus on the performance of individual elastic units. For example, Montemurro et al. utilized the mechanical properties of individual elastic elements in an elastic coordinate system and optimized the structure and optimal elastic modulus of the elastic element using mechanical calculations. Tkachenko et al. summarized the relationship between the physical and mechanical parameters of elastic elements and temperature using experimental methods and verified the results using practical examples. Du et al., starting from the deformation of elastic elements, used theoretical calculations and derivations to solve the maximum stress and strain equations of a certain type of elastic element. Tianjin University and Zhengzhou Institute of Mechanical and Electrical Engineering jointly developed a composite flexible wrist and studied the apparent elastic modulus, flexibility, and stiffness of elastic elements, laying the foundation for the performance research of domestically produced elastic elements. Although the above methods can improve the performance indicators of individual elastic elements to some extent, there is almost no research on how to assemble multiple elastic elements or which elastic elements to select to assemble into an elastic element assembly. Currently, elastic element assemblies are randomly assembled, without a defined and evidence-based assembly method. This results in poor stability and load-bearing capacity of the assembled elastic element group. Summary of the Invention

[0004] The purpose of this invention is to provide a method for assembling stacked elastic units to solve the problem of assembling elastic elements in an elastic element group. From the perspective of product quality, considering the performance characteristics of stacked elastic elements, this invention provides a reliable method for assembling elastic elements to improve the stability and load-bearing capacity of the elastic element group.

[0005] To achieve the above objectives, the present invention includes:

[0006] The present invention provides a method for assembling a stacked elastic unit, comprising the following steps:

[0007] 1) Obtain the original elastic element data, preprocess the original data to obtain the processed elastic element data, and use the processed elastic element data as the dataset of elastic elements.

[0008] 2) Cluster the dataset of elastic unit bodies according to the first set value to obtain data clusters with the number of classes equal to the first set value;

[0009] Determine the number of data in each data cluster, select a target data cluster from each data cluster based on the number of data, select data with a second set value from the target data cluster, and form an elastic unit group by combining the elastic unit bodies corresponding to the selected data to complete the selection of the elastic unit group.

[0010] The first setting value is the total number of elastic unit groups, and the second setting value is the number of elastic unit bodies required for one elastic unit group.

[0011] 3) Delete the selected data from the dataset of the elastic unit to obtain a new dataset. Subtract 1 from the current first setting value and use it as the new first setting value. Repeat steps 2) and 3) on the new dataset using the new first setting value to select elastic unit groups until the number of data in the largest data cluster of the new dataset is less than or equal to the set threshold. Stop selecting elastic unit groups. Use each selected elastic unit group to assemble different elastic elements.

[0012] Beneficial Effects: This invention proposes a feasible assembly method that benefits the performance of elastic element groups, providing a reliable basis for the assembly of elastic units. Based on the principle that the higher the isotropy of the elastic units, the better the performance of the elastic element group composed of them, this invention fully considers the performance indicators of the elastic units. By further selecting elastic units within the qualified range, i.e., using clustering to classify the data of each elastic unit, elastic units of the same category are grouped together for assembly decisions, providing effective support for the assembly of elastic unit groups. This invention further improves the performance of elastic elements, requiring less force and having a stronger load-bearing capacity to complete the same action. By assembling elastic elements according to the assembly decision results, the elastic elements experience more uniform force during operation, requiring less force and having a greater load-bearing capacity to complete the same action.

[0013] Furthermore, in step 2), the method for selecting data with a second set value from the target data cluster is as follows: sort the data in the target data cluster according to the size of the weight index, and select consecutive data with a second set value from the sorted data cluster. The weight index is obtained according to the Delphi method.

[0014] Beneficial Effects: This invention effectively resolves the potential conflict between clustering results and the required number of elastic units for each elastic unit group. The K-medoids algorithm, an unsupervised learning algorithm, results in an uncontrollable number of elastic units in each category after partitioning, while the required number of elastic units for each group is fixed, leading to a potential conflict. This invention comprehensively considers the influence of index weights. Through the above steps, using sorting and random selection methods, it can obtain elastic unit groups with higher isotropy, further improving the performance of elastic elements. The elastic element groups require less force to complete the same action, exhibit more uniform force distribution, and have stronger load-bearing capacity.

[0015] Furthermore, the preprocessing method for the raw data is as follows: the acquired raw elastic unit data are normalized respectively, the normalized data indicators are determined, the improved Delphi method is used to perform indicator weight analysis, and the normalized indicator data and corresponding weights are integrated to obtain the elastic unit dataset.

[0016] Beneficial Effects: This invention avoids the influence of dimensions and orders of magnitude in various indicators through preprocessing. Multiple indicators affect the quality of stacked elastic unit cells, and they have different dimensions and orders of magnitude. If these indicator data are directly used for clustering, indicators with larger values ​​will have a significant impact on the classification results, which is inconsistent with reality. This invention eliminates this influence through data preprocessing. By fully considering the influence of indicator weights, the data to be clustered is closer to reality. Multiple indicators affect the quality of stacked elastic unit cells, and these indicators have different effects on the quality of elastic unit cells. Data preprocessing can fully consider the weights of these indicators, making the data to be clustered more realistic, thereby making the clustering results more effective.

[0017] Furthermore, the target data cluster selected in step 2) is the data cluster with the largest number of data among all data clusters.

[0018] Beneficial effect: The target data cluster is the data cluster with the largest number of data among all data clusters, thereby ensuring that the requirements of the second set value can be met when selecting data from the data cluster.

[0019] Furthermore, the K-medoids algorithm is used to cluster the dataset of elastic units.

[0020] Beneficial effects: The K-medoids algorithm is used to cluster the dataset of elastic units. This algorithm is simple and efficient, can quickly perform clustering, and the data differences within the clusters are small.

[0021] Furthermore, the weights of each indicator are obtained using the Delphi method as follows:

[0022] ① Analyze the various factors affecting the quality of the elastic unit, determine the influencing indicators and design an evaluation table, and invite experts to score each indicator according to the evaluation table;

[0023] ② Based on the scoring results of each index corresponding to each elastic unit, obtain the normal distribution function corresponding to the score of each index, and solve the maximum likelihood estimate of the mathematical expectation and variance in the normal distribution function;

[0024] ③ Based on the expected value and variance in the normal distribution function, the scores of each indicator are divided into strongly differential data and weakly differential data. The strongly differential data are corrected based on the maximum likelihood estimate of the expected value to obtain the corrected dataset of scores for each indicator.

[0025] ④ Calculate the total score of each indicator after correction and the total score of all indicators. The quotient of the total score of a single indicator and the total score of all indicators is the weight of that indicator.

[0026] Beneficial effects: This invention further improves upon the traditional Delphi method. The index weights obtained by this invention are more objective and accurate, avoiding the influence of the evaluator's subjective consciousness, knowledge structure, and personal experience on the evaluation results of the existing Delphi method.

[0027] Furthermore, the method for correcting the scores of each indicator is to replace the strongly differential data in the score of an indicator with the expected value of the score of that indicator if there is data with strong differences.

[0028] Beneficial effect: Effectively avoids the impact of highly variable data.

[0029] Furthermore, the raw data of the elastic unit includes the total length, outer diameter, total weight, length of the elastic part, hardness of the rubber sheet, lateral stiffness, tilting stiffness, axial compressive stiffness, and torsional stiffness.

[0030] Beneficial effects: This invention comprehensively considers various important influencing factors of the elastic unit, thus obtaining more reliable assembly decision results. Furthermore, the method for obtaining the original data of the elastic unit is as follows: elastic units with appearance quality grades of excellent and good, corresponding to 1 and 0.6 respectively, are selected; the total length, outer diameter, total weight, length of the elastic part, and hardness of the rubber sheet are extracted from the product inspection records; the lateral stiffness, tilting stiffness, axial compressive stiffness, and torsional stiffness are expressed using the force values ​​corresponding to the fixed deformation in the product test report.

[0031] Beneficial Effects: This invention proposes a feasible method for extracting raw data from stacked elastic element bodies. Considering the performance characteristics of elastic components and combining relevant reports and materials, a feasible method is proposed for extracting raw data from stacked elastic element bodies. Qualitative indicators are transformed into quantitative indicators, making the relevant indicators more operable. Some performance-influencing indicators of stacked elastic element bodies are qualitative, such as appearance quality indicators. Converting these qualitative indicators into quantitative indicators makes them more operable. Attached Figure Description

[0032] Figure 1 This is a schematic diagram of a flexible wrist in existing technology;

[0033] Figure 2 This is a schematic diagram of a flexible wrist structure in the prior art;

[0034] Figure 3 This is a flowchart illustrating the overall technical solution in the method embodiments of the present invention;

[0035] Figure 4 This is a schematic diagram of the elastomer quality classification results in the method embodiment of the present invention;

[0036] Figure 5 These are graphs of various sets of test data in the method embodiments of the present invention;

[0037] Figure 6 This is a comparison chart of performance indicators before and after classification in the method embodiment of the present invention;

[0038] Figure 7 This is a flowchart of the quality classification method in an embodiment of the present invention. Detailed Implementation

[0039] The present invention will now be described in further detail with reference to the accompanying drawings.

[0040] Example of a method for determining the elastic unit group of an elastomer element:

[0041] Based on the concept that higher isotropy of elastic unit cells leads to better assembly performance, and consequently, superior performance of the elastic element composed of them (i.e., elastic elements composed of elastic unit cells with similar performance indicators exhibit more stable performance), and considering the performance characteristics of stacked elastic elements from a product quality perspective, the following is proposed: Figure 3 The method for assembling a stacked elastic unit body provided here offers a reliable approach for assembling elastic elements. Specific implementation details are as follows:

[0042] 1) Obtain the original elastic element data, preprocess the original elastic element data to obtain the processed elastic element data, and use the processed elastic element data as the dataset of elastic elements.

[0043] In this embodiment, the characteristics of laminated elastic elements are analyzed during the research process. Analysis shows that in actual working conditions, a laminated elastomer can be considered as a rubber spring, mainly subjected to tensile force, shear force, and torsional force. From a practical application perspective, the quality of a laminated elastomer should primarily consider its lateral stiffness, torsional stiffness, axial compressive stiffness, and torsional stiffness. Therefore, this embodiment summarizes the key indicators for classifying the quality of elastic elements based on the basic requirements of product quality control and the perspective of actual product use, as follows: appearance quality grade, lateral stiffness, torsional stiffness, axial compressive stiffness, torsional stiffness, total length, outer diameter, total weight, and length of the elastic part (excluding the length of the metal parts at both ends). Then, a method for extracting raw data is developed based on feature analysis. The original data acquisition method is as follows: the appearance quality grades of excellent and good correspond to 1 and 0.6 respectively; the total length, outer diameter, total weight, length of the elastic part (excluding the length of the metal parts at both ends) and hardness of the rubber sheet are extracted from the product inspection record; the lateral stiffness, tilting stiffness, axial compressive stiffness and torsional stiffness are expressed by the force values ​​corresponding to the fixed deformation in the product test report.

[0044] Based on the above research and analysis, it can be seen that the original dataset in this embodiment is obtained by selecting elastic units within the qualified range from the product inspection records and extracting the data to obtain the original dataset. The dataset includes the total length, outer diameter, total weight, elastic part length (excluding the length of the metal parts at both ends) of each elastic unit, and the hardness, lateral stiffness, tilting stiffness, axial compressive stiffness and torsional stiffness of the rubber sheet.

[0045] The requirements for obtaining the original dataset are as follows: the appearance quality grades of the elastic unit body are excellent and good, corresponding to 1 and 0.6 respectively; the total length, outer diameter, total weight, length of the elastic part (excluding the length of the metal parts at both ends), and hardness of the rubber sheet are obtained from the product inspection records; and the lateral stiffness, tilting stiffness, axial compressive stiffness, and torsional stiffness are represented by the force values ​​corresponding to the fixed deformation in the product test report.

[0046] Secondly, the original data of the elastic unit is preprocessed. The preprocessing steps include data normalization, index weighting, and dataset integration. The stacked elastic unit assembly method proposed in this embodiment is mainly for stacked elastic units in a qualified state, so the data to be processed is relatively stable and does not contain "dirty data". Therefore, this invention uses the most classic and commonly used extreme value normalization to process the relevant data. The specific method is as follows:

[0047] Let the original dataset of elastic elements from 1 to m be X = (X1, X2, ..., Xm). a , ..., X m The index set A = (A1, A2, ..., A...) iA p The original data set X for extracting the index of the a-th elastic element. a =(t A1 , t A2 , ..., t Ai , ..., t Ap The normalized dataset Y = (Y1, Y2, ..., Y...) a , ..., Y m The normalized data set Y of the a-th elastic element. a =(T A1 T A2 ,…,T Ai ,…,T Ap If the normalized data T is then... Ai It can be processed using equation (1):

[0048]

[0049] In the formula: T Ai Let A be the index of the a-th elastic body. i Normalized data, t Ai Let A be the index of the a-th elastic body. i The corresponding original data, t Aimin For m elastic elements, index A i The smallest original data, t Aimax For m elastic elements, index A i The largest raw data

[0050] In this embodiment, the preprocessing step also includes determining the index weights after data normalization, integrating each data index and its corresponding weight to obtain the processed elastic unit data, and using the processed elastic unit data as the final elastic unit dataset.

[0051] Specifically, the indicator weight determination method in this embodiment is the Delphi method. Considering that the evaluation results of the Delphi method are easily influenced by the evaluator's subjective awareness, knowledge structure, and personal experience, this embodiment improves the subsequent data processing when using the Delphi method. The entire process is as follows:

[0052] ① Each elastic unit is scored according to the factors affecting it. Specifically, this embodiment designs an evaluation table and invites relevant experts to score the evaluations. The evaluation table is shown in Table 1. Let the k-th expert score index A. i The evaluation dataset is F ik .

[0053] Table 1

[0054]

[0055] ② Based on the scoring results of each index corresponding to each elastic unit, obtain the normal distribution function corresponding to the score of each index, and solve for the maximum likelihood estimate of the mathematical expectation and variance in the normal distribution function corresponding to the score of each index. Specifically, obtain the score of each index, and solve for the normal distribution function N(μ, σ) of the score of each index. 2 Indicator A i The evaluation dataset is F Ai =(F i1 F i2 F ik F im μ can be obtained through equation (2). Ai and σ Ai 2 Maximum likelihood estimate:

[0056]

[0057] ③ Based on the expected value and variance of the normal distribution function, the scores of the indicators corresponding to the elastic unit data are divided into strongly variable data and weakly variable data. The strongly variable data are then corrected based on the maximum likelihood estimate of the expected value. The data obtained for each indicator are further divided into strongly variable data (sample data falling outside the μ±3σ region) and weakly variable data (sample data falling within the μ±3σ region) according to the 3σ criterion. The evaluation dataset F is then... Ai Replace the strongly dissimilar data in the table with the corresponding mathematical expectation μ. Ai To correct for highly disparate data, the corrected dataset F of scores for each indicator is obtained. Ai '=(F i1 ', F i2 ',…,F ik ',…,F im ').

[0058] ④ Calculate the total score of each indicator after correction, and the total score of all indicators. The quotient of the total score of a single indicator and the total score of all indicators is the weight of that indicator. Indicator A i weight U Ai It can be calculated using equation (3).

[0059]

[0060] The normalized index data and their corresponding weights are integrated to obtain the final elastic unit dataset. Let the database after data integration be V = (V1, V2, ..., V...). a , ..., V m ), the dataset V after integrating the a-th elastic element a =(TA1 'T A2 ′, T A1 ′,…,T Ai ′,…,T Ap If T''), then T' ... Ai ′ can be processed using equation (4):

[0061] T Ai ′=T Ai *U Ai (4)

[0062] After completing the extraction of raw data, data processing before quality classification, and dataset integration, each data point is numbered to correspond with the actual elastic element, thus obtaining the final dataset.

[0063] 2) Cluster the dataset of elastic unit bodies according to the first set value K to obtain data clusters with the number of classes equal to the first set value;

[0064] Determine the number of data in each data cluster, select a target data cluster from each data cluster based on the number of data, select data with a number of data points equal to the second set value U from the target data cluster, and form an elastic unit group by combining the elastic unit bodies corresponding to the selected data to complete the selection of the elastic unit group.

[0065] The selected data is deleted from the dataset of elastic units to obtain a new dataset. The first setting value K is the number of elastic elements composed of data from the dataset of elastic units (i.e., the total number of groups that can be composed of elastic units), and the second setting value U is the number of elastic units required to form a single elastic element (i.e., the number of elastic units required for a group of elastic units).

[0066] Specifically, in this embodiment, when selecting the target data cluster, the data in the target data cluster are sorted according to the size of the weight index, and the number of consecutive data items with the second set value is selected from the sorted data clusters. The target data cluster can be the data cluster with the most data items among all data clusters, or it can be a data cluster with the number of data items exceeding the number of elastic unit bodies required to form a single elastic element (e.g., a flexible wrist). In this embodiment, the target data cluster is the data cluster with the most data items among all data clusters.

[0067] The weight index corresponding to each elastic unit data can be calculated in the preprocessing step or when sorting the data in the target data cluster.

[0068] Specifically, the target data cluster with the largest number of data points is sorted from smallest to largest according to the index with the largest weight (which is obtained from the preprocessed index weight processing result). U consecutive data points are randomly extracted from the sorted target data cluster. The elastic unit bodies corresponding to the U data points are grouped into a group of elastic unit bodies using the data number. Each of the elastic unit body groups can be assembled into an elastic element.

[0069] The K-medoids algorithm is used to cluster the dataset of elastic units. Other implementation methods include the K-means algorithm.

[0070] 3) Delete the selected data from the dataset of the elastic unit body to obtain a new dataset. Subtract 1 from the current first setting value and use it as the new first setting value. Repeat steps 2) and 3) on the new dataset using the new first setting value to select elastic unit body groups until the number of data in the largest data cluster of the new dataset is less than or equal to the set threshold. Stop selecting elastic unit body groups. Use each selected elastic unit body group to assemble different elastic body elements.

[0071] Specifically, such as Figure 7 As shown, a total category K-1 for quality classification is set based on the data in the new dataset, and clustering is performed using the K-medoids algorithm. Step 2) is repeated on the new dataset using the new initial value K-1 until the number of data points in the largest cluster of the new dataset is less than or equal to a set threshold. At this point, the selection of elastic unit groups stops, and the classification of the new dataset is completed. Each selected elastic unit group is then used to assemble different elastic body components. If the number of data points in the largest cluster of the new dataset is insufficient to select an elastic unit group, the selection of elastic unit groups is terminated.

[0072] To verify the feasibility of the proposed mass classification method, this embodiment randomly selected 72 laminated elastic elements and classified them according to the method of this embodiment. The results show that the proposed method is entirely feasible. Specific classification results are as follows: Figure 4 As shown.

[0073] The effectiveness of the assembly decision was also preliminarily verified in this invention. The results show that, based on the results obtained in this embodiment, the elastic unit group requires less force and has a greater load-bearing capacity to complete the same action. In this embodiment, one randomly selected group of elastic units from the same batch (group 0) and 12 groups of classified elastic units (groups 1-12) were assembled with the front and rear ends of the flexible wrist, and yaw, torsion, pitch, and ultimate load-bearing tests were conducted according to the actual working conditions of the flexible wrist. Specific test results are as follows: Figure 5 , Figure 6As shown. Therefore, assembling elastic elements according to the mass classification results of the stacked elastomers proposed in this embodiment can further improve the performance of the elastic elements. The elastic element group requires less force to complete the same action, the force is more uniform, and the load-bearing capacity is stronger.

[0074] Specific implementation methods have been given above, but the present invention is not limited to the described implementation methods. The basic idea of ​​the present invention lies in the above basic scheme. For those skilled in the art, designing various modified models, formulas, and parameters based on the teachings of the present invention does not require creative effort. Changes, modifications, substitutions, and variations made to the implementation methods without departing from the principles and spirit of the present invention still fall within the protection scope of the present invention.

Claims

1. A method for assembling a stacked elastic unit, characterized in that, Includes the following steps: 1) Obtain the original elastic element data, preprocess the original elastic element data to obtain the processed elastic element data, and use the processed elastic element data as the dataset of elastic elements. 2) Cluster the dataset of elastic unit bodies according to the first set value to obtain data clusters with the number of classes equal to the first set value; Determine the number of data in each data cluster, select a target data cluster from each data cluster based on the number of data, select data with a second set value from the target data cluster, and set the elastic unit corresponding to the selected data into an elastic unit group to complete the selection of each elastic unit in the elastic unit group. The first setting value is the total number of elastic unit groups, and the second setting value is the number of elastic unit bodies required to form a set of elastic unit groups. 3) Delete the selected data from the dataset of the elastic unit to obtain a new dataset. Subtract 1 from the current first setting value and use it as the new first setting value. Repeat steps 2) and 3) on the new dataset using the new first setting value to select elastic unit groups until the number of data in the data cluster with the largest data volume in the new dataset is less than or equal to the set threshold, and then stop selecting elastic unit groups. The selected groups of elastic unit bodies are used to assemble different elastic body components.

2. The method for assembling a stacked elastic unit according to claim 1, characterized in that, In step 2), the method for selecting data with a second set value from the target data cluster is as follows: sort the data in the target data cluster according to the size of the weight index, and select consecutive data with a second set value from the sorted data cluster. The weight index is obtained according to the Delphi method.

3. The method for assembling a stacked elastic unit according to claim 2, characterized in that, The preprocessing method for the raw data is as follows: the raw elastic unit data are normalized, the normalized data indicators are determined, the improved Delphi method is used to perform indicator weight analysis, and the normalized indicator data and corresponding weights are integrated to obtain the elastic unit dataset.

4. The method for assembling a stacked elastic unit according to claim 2, characterized in that, The target data cluster selected in step 2) is the data cluster with the largest number of data among all data clusters.

5. The method for assembling a stacked elastic unit according to claim 3, characterized in that, The K-medoids algorithm is used to cluster the dataset of elastic units.

6. The method for assembling a stacked elastic unit according to claim 2 or 3, characterized in that, The weights of each indicator are obtained using the Delphi method as follows: ① Analyze the various factors affecting the quality of the elastic unit, determine the influencing indicators and design an evaluation table, and invite experts to score each indicator according to the evaluation table; ② Based on the scoring results of each index corresponding to each elastic unit, obtain the normal distribution function corresponding to the score of each index, and solve the maximum likelihood estimate of the mathematical expectation and variance in the normal distribution function; ③ Based on the expected value and variance in the normal distribution function, the scores of each indicator are divided into strongly differential data and weakly differential data. The strongly differential data are corrected based on the maximum likelihood estimate of the expected value to obtain the corrected dataset of scores for each indicator. ④ Calculate the total score of each indicator after correction and the total score of all indicators. The quotient of the total score of a single indicator and the total score of all indicators is the weight of that indicator.

7. The method for assembling a stacked elastic unit according to claim 6, characterized in that, The correction method for the index scores corresponding to the elastic unit data is as follows: if there are highly differential data in a certain index score, replace the highly differential data in the index score with the mathematical expectation of the index score.

8. The method for assembling a stacked elastic unit according to claim 1 or 3, characterized in that, The raw data of the elastic unit include total length, outer diameter, total weight, length of the elastic part, hardness of the rubber sheet, lateral stiffness, tilting stiffness, axial compressive stiffness, and torsional stiffness.

9. The method for assembling a stacked elastic unit according to claim 8, characterized in that, The method for obtaining the original data of the elastic unit is as follows: the appearance quality grades of excellent and good correspond to 1 and 0.6 respectively; the total length, outer diameter, total weight, length of the elastic part and hardness of the rubber sheet are extracted from the product inspection records; Lateral stiffness, tilting stiffness, axial compressive stiffness, and torsional stiffness are expressed as the force values ​​corresponding to the fixed deformation in the product test report.

Citation Information

Patent Citations

  • Clustering-based approximate duplicate record detection method and apparatus

    CN108268876A