Assembly quality early warning and process self-optimization system based on data fingerprint matching

Abstract

This invention discloses an assembly quality early warning and process self-optimization system based on data fingerprint matching, belonging to the field of mechanical assembly quality control and process optimization technology. The system includes: a data acquisition module, a fingerprint construction and storage module, a matching and early warning module, a dynamic self-adjustment module, and a feedback verification module. This invention establishes a multi-dimensional data acquisition system to form process-level data fingerprints, builds a fingerprint database, and completes algorithm parameter calibration, enabling comprehensive calculation and hierarchical early warning of assembly quality deviations. It reduces missed and false alarms, achieves process control, and dual-end storage ensures secure and efficient data retrieval. Through a dynamic adjustment algorithm for process parameters, it achieves process parameter self-optimization, constructs a closed-loop linkage mechanism, and completes the entire process operation without manual intervention, adapting to complex assembly processes.

Description

Technical Field

[0001] This invention relates to the field of mechanical assembly quality control and process optimization technology, specifically to an assembly quality early warning and process self-optimization system based on data fingerprint matching. Background Technology

[0002] Intelligent manufacturing is a core trend in the transformation and upgrading of the manufacturing industry. The precision and complexity of mechanical equipment assembly processes are continuously increasing, and assembly quality directly determines the overall performance and lifespan of the product, becoming a core link in the entire production process control. While current assembly production incorporates sensor acquisition and process parameter monitoring technologies to obtain basic production data, the integration and utilization of multi-dimensional data is insufficient. Assembly quality control still relies primarily on post-production inspection, making it difficult to achieve early warning during the process. Furthermore, process optimization is mostly done through manual adjustments after the fact, lacking real-time and dynamic capabilities. Therefore, there is an urgent need to construct an integrated quality early warning and process self-optimization system based on data fusion and algorithm technology to meet the real-time control requirements of intelligent manufacturing.

[0003] Existing assembly quality early warning technologies mostly rely on single-dimensional sensing or process parameters for deviation judgment, failing to construct a multi-dimensional process-level data fingerprint system. Deviation calculations lack comprehensiveness and accuracy, and the grading and triggering mechanisms for early warning signals are incomplete, easily leading to missed and false alarms. Regarding process optimization, existing technologies largely employ passive optimization modes using manual adjustment or fixed parameters, lacking quantifiable dynamic adjustment algorithms for process parameters. This prevents precise, grading adjustment of parameters based on the degree of quality deviation, and the absence of a closed-loop linkage mechanism of early warning-adjustment-verification. The effectiveness of adjusted process parameters cannot be verified in a timely manner, causing process optimization to lag behind the production process. This makes it difficult to adapt to the real-time control requirements of complex assembly processes and fails to fully leverage the value of production data.

[0004] In summary, existing assembly quality control and process optimization technologies suffer from low data fusion levels, inaccurate deviation judgment, lack of quantitative algorithms for process adjustment, and absence of closed-loop verification mechanisms. These limitations prevent the achievement of process-based early warning of assembly quality and dynamic self-optimization of process parameters, easily leading to potential assembly quality issues and reducing production efficiency and product yield. As the manufacturing industry undergoes a deep upgrade towards intelligence and precision, the industry demands higher levels of real-time control, accurate early warning, and dynamic optimization of the assembly process. Developing an assembly quality early warning and process self-optimization system based on multi-dimensional data fusion and the construction of data fingerprints is crucial for addressing existing industry pain points and improving the level of intelligent assembly production. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide an assembly quality early warning and process self-optimization system based on data fingerprint matching. It can construct process-level three-dimensional data fingerprints and a standardized fingerprint library through multi-dimensional data collection, and use a three-dimensional fingerprint comprehensive deviation algorithm to realize quality deviation calculation and graded early warning; it adopts a process parameter dynamic adjustment algorithm to realize precise dynamic adjustment of process parameters; and it constructs a feedback verification closed-loop linkage mechanism to re-collect data for secondary verification until the deviation meets the standard, thereby realizing process control and process self-optimization of assembly quality.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an assembly quality early warning and process self-optimization system based on data fingerprint matching, the system comprising: a data acquisition module, a fingerprint construction and storage module, a matching and early warning module, a dynamic self-adjustment module, and a feedback verification module;

[0007] The data acquisition module collects various sensor data, process parameter data, and process execution time data throughout the entire assembly process, and binds all collected data with assembly process identifiers to form multi-dimensional assembly data.

[0008] The fingerprint construction and storage module receives multi-dimensional assembly data, constructs real-time three-dimensional data fingerprints at the process level, builds and stores a standard three-dimensional data fingerprint library corresponding to each assembly process, and completes the calibration and storage of the initial parameters of the three-dimensional fingerprint comprehensive deviation algorithm.

[0009] The matching and early warning module receives real-time 3D data fingerprints, standard 3D data fingerprint database data, and initial calibration parameters; calculates a comprehensive deviation value using a 3D fingerprint comprehensive deviation algorithm; compares the comprehensive deviation value with a preset threshold; archives the matching result if the deviation value does not exceed the threshold; and outputs a corresponding quality early warning signal if the deviation value exceeds the threshold. Simultaneously, the deviation data and the corresponding assembly process identifier are transmitted to the dynamic self-adjustment module.

[0010] The dynamic self-adjustment module: Based on the transmitted deviation data and assembly process identifier, it uses a process parameter dynamic adjustment algorithm to calculate the adjustment amount of each process parameter, and dynamically adjusts the process parameters of the assembly execution equipment according to the adjustment amount.

[0011] The feedback verification module receives the adjusted process parameters, collects the corresponding assembly data and generates a new real-time three-dimensional data fingerprint, and sends it back to the matching and early warning module for secondary matching verification; based on the verification result, it triggers data archiving or secondary adjustment to form a closed-loop linkage.

[0012] Furthermore, the data acquisition module includes a sensor acquisition unit, a process parameter acquisition unit, a time acquisition unit, and a data binding unit;

[0013] The sensing and acquisition unit collects pressure sensing data, torque sensing data, displacement sensing data, and temperature sensing data throughout the entire assembly process, and generates a sensing and acquisition dataset with millisecond-level timestamps.

[0014] The process parameter acquisition unit collects process parameter data such as rotational speed, feed rate, pressing force, and holding time during equipment operation, and generates a process parameter acquisition dataset with millisecond-level timestamps.

[0015] The time acquisition unit captures and records the start time of the assembly process, the start and end times of each sub-step, and the overall completion time of the process, generating a process execution time dataset with a unique time identifier.

[0016] The data binding unit assigns a unique assembly process identifier to each executed assembly process. The identifier includes the product model, process number, production line number, and execution batch information. The sensor acquisition dataset, process parameter acquisition dataset, and process execution time dataset are associated and bound one-to-one with the assembly process identifier according to a unified millisecond-level timestamp. The bound multi-dimensional data is then standardized and structured to form multi-dimensional assembly data.

[0017] Furthermore, the fingerprint construction and storage module includes a fingerprint construction unit, a fingerprint database construction unit, a parameter calibration unit, and a data storage unit;

[0018] The fingerprint construction unit uses sensor data as the X-axis dimension, process parameter data as the Y-axis dimension, and process execution time data as the Z-axis dimension. After normalizing the data in each dimension, feature values ​​are extracted and fused to form a process-level real-time three-dimensional data fingerprint.

[0019] The fingerprint database construction unit selects multi-dimensional assembly data from the assembly process to construct a three-dimensional data fingerprint, which is then processed to obtain a standard three-dimensional data fingerprint. The standard three-dimensional data fingerprint database is then constructed by classifying the product model and process number.

[0020] The parameter calibration unit calibrates the initial parameters of the three-axis weight coefficients and deviation correction coefficients of the three-dimensional fingerprint comprehensive deviation algorithm based on the data distribution characteristics of the standard three-dimensional fingerprint database.

[0021] The data storage unit adopts a dual storage mode of cloud server and local server to synchronously store real-time 3D data fingerprint, standard 3D data fingerprint library, and initial calibration parameters of 3D fingerprint comprehensive deviation algorithm on both ends.

[0022] Furthermore, the specific steps for the matching and early warning module to calculate the comprehensive deviation value using the three-dimensional fingerprint comprehensive deviation algorithm are as follows:

[0023] Based on the assembly process identifier, retrieve the corresponding standard 3D data fingerprint from the standard 3D data fingerprint database. Let the triaxial feature value of the standard 3D data fingerprint be... The triaxial feature values ​​of real-time 3D data fingerprints are The initial calibration parameters of the three-dimensional fingerprint comprehensive deviation algorithm include the three-axis weight coefficients. , , and deviation correction factor ,and ;

[0024] Calculate the three-axis single-dimensional deviation value using the formula: , , ,in, This represents the single-dimensional deviation value along the X-axis. This represents the single-dimensional deviation value along the Y-axis. This represents the single-dimensional deviation value along the Z-axis.

[0025] The comprehensive deviation value is calculated using the formula of the three-dimensional fingerprint comprehensive deviation algorithm. , ,in, The deviation correction coefficient is determined based on the data distribution characteristics of the standard three-dimensional data fingerprint database, and its value range is (0,1].

[0026] The calculated comprehensive deviation value Compare with a preset threshold, when If the value is less than or equal to a preset threshold, the matching results and real-time 3D data fingerprints will be sent back to the fingerprint construction and storage module for archiving; if the value is less than or equal to a preset threshold, the matching results and real-time 3D data fingerprints will be sent back to the fingerprint construction and storage module for archiving. If a preset threshold is set, a corresponding quality warning signal will be output, and... Deviation data and corresponding assembly process identifiers are synchronously transmitted to the dynamic self-adjustment module.

[0027] Furthermore, the preset thresholds in the matching and early warning module correspond one-to-one with the assembly process identifiers, and are synchronously calibrated by the fingerprint construction and storage module when building the standard three-dimensional data fingerprint database; the quality early warning signal includes the assembly process identifier, deviation information, and risk level, with the risk level based on the comprehensive deviation value. Size division, A level one warning is issued when the threshold value is between 1.2 times the preset threshold value, and a yellow warning signal is output. A level 2 warning is issued when the value is between 1.2 and 1.5 times the preset threshold, and an orange warning signal is output. If the risk exceeds 1.5 times the preset threshold, a Level 3 warning is issued, a red warning signal is output, and an audible and visual alarm and a pop-up reminder from the background system are triggered simultaneously. Different risk levels correspond to different priorities for adjusting process parameters.

[0028] Furthermore, the dynamic self-adjustment module includes a parameter retrieval unit, an adjustment amount calculation unit, and a device adjustment unit;

[0029] The parameter retrieval unit is used to retrieve the list of process parameters for the corresponding assembly process and the association rules between process parameters and assembly quality according to the assembly process identifier; the association rules are a quantitative correspondence between the change in process parameters and the change in the deviation value of three-dimensional data fingerprint, established based on the process characteristics of the assembly process and historical assembly quality data.

[0030] The adjustment calculation unit is used to receive deviation data and assembly process identifiers, and to calculate the adjustment amount of each process parameter using a dynamic adjustment algorithm for process parameters.

[0031] The equipment adjustment unit is used to communicate with the control system of the assembly execution equipment and adjust the process parameters of the assembly execution equipment according to the adjustment amount.

[0032] Furthermore, the specific steps for calculating the adjustment amount of each process parameter using the dynamic adjustment algorithm for process parameters are as follows:

[0033] Let the overall deviation value transmitted by the matching and early warning modules be... The preset process parameter quality qualification threshold is: The basic coefficient for adjusting process parameters is: The value is determined by the process characteristics of the corresponding assembly process and ranges from (0,2].

[0034] Let the process parameters to be adjusted be: , The number of process parameters to be adjusted; the correlation coefficient between the process parameters and the overall deviation value is... ,and The correlation coefficient is determined by the degree of influence of process parameters on assembly quality; the higher the degree of influence, the higher the correlation coefficient. The larger;

[0035] The adjustment amount of each process parameter is calculated using a dynamic adjustment algorithm. , , among which, when ,but A negative value indicates that the process parameters need to be adjusted downwards; when ,but A positive value indicates that the process parameters need to be adjusted upwards;

[0036] Calculate the adjusted process parameter values and will As the target value for adjusting the process parameters of the assembly execution equipment.

[0037] Furthermore, the secondary matching verification process in the feedback verification module is as follows: after receiving the adjusted process parameters, the data acquisition module re-acquires multi-dimensional assembly data corresponding to the assembly process under the adjusted process parameters; through the three-dimensional data fingerprint construction method of the fingerprint construction and storage module, a new process-level real-time three-dimensional data fingerprint is generated based on the acquired multi-dimensional assembly data; the new process-level real-time three-dimensional data fingerprint is sent back to the matching and early warning module, which performs secondary matching verification on the new real-time three-dimensional data fingerprint according to the initial comparison process.

[0038] Furthermore, the closed-loop linkage process in the feedback verification module is as follows: receiving the secondary matching verification result and comprehensive deviation value from the matching and early warning module; when the comprehensive deviation value does not exceed the preset threshold, triggering process adjustment and archiving of corresponding assembly data; when the comprehensive deviation value still exceeds the threshold, sending a secondary adjustment command to the dynamic self-adjustment module; after the dynamic self-adjustment module completes the command adjustment, the feedback verification module repeatedly collects assembly data, generates new real-time three-dimensional data fingerprints, and sends them back for verification until the deviation value meets the standard, thus forming a closed-loop linkage.

[0039] Compared with existing technologies, this assembly quality early warning and process self-optimization system based on data fingerprint matching has the following advantages:

[0040] I. This invention establishes a multi-dimensional data acquisition system to achieve accurate collection and association of various data throughout the assembly process. It combines a three-dimensional data fingerprint construction method to form process-level data fingerprints, while simultaneously building a standardized fingerprint library and calibrating algorithm parameters. Based on the three-dimensional fingerprint comprehensive deviation algorithm, it achieves comprehensive calculation of assembly quality deviations. By comparing deviation values ​​with process-specific thresholds, it enables tiered triggering of quality warnings, breaking through the limitations of judging quality based on single-dimensional data. This makes quality warnings more aligned with the actual characteristics of the process, effectively reducing missed and false alarms, and achieving process-based control of assembly quality. Furthermore, the dual-end storage mode ensures the secure storage and efficient retrieval of various data and algorithm parameters.

[0041] Second, this invention utilizes a dynamic process parameter adjustment algorithm, combined with the correlation between assembly quality deviation and process parameters, to achieve quantitative calculation of process parameter adjustment amounts. Based on the equipment communication mechanism, it completes precise dynamic adjustment of the process parameters of the assembly execution equipment. Simultaneously, it constructs a closed-loop linkage mechanism for feedback verification. After parameter adjustment, data is re-collected to generate new data fingerprints, and secondary matching verification is completed. Based on the verification results, data archiving or secondary adjustment is implemented until the deviation value meets the standard. This breaks through the traditional manual adjustment and passive optimization model, achieving automated and precise self-optimization of process parameters. The entire process of early warning, adjustment, and verification can be completed without manual intervention, adapting to the control needs of complex assembly processes.

[0042] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0044] Figure 1 A block diagram showing the module composition of an assembly quality early warning and process self-optimization system based on data fingerprint matching;

[0045] Figure 2 The flowchart shows an assembly quality early warning and process self-optimization system based on data fingerprint matching.

[0046] Figure 3 This is a flowchart of the dynamic adjustment and secondary matching verification of process parameters for an assembly quality early warning and process self-optimization system based on data fingerprint matching. Detailed Implementation

[0047] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0048] Example 1

[0049] The data acquisition module is the first to conduct comprehensive data acquisition throughout the entire cylinder and piston press-fitting process. The sensor acquisition unit accurately collects pressure, displacement, and temperature sensor data during the press-fitting process. All acquired data generates a sensor acquisition dataset with millisecond-level timestamps. These millisecond-level timestamps ensure that each sensor data point accurately corresponds to the real-time operation of the press-fitting process, allowing subsequent data analysis to accurately pinpoint the change points of various physical quantities during the press-fitting process. The process parameter acquisition unit simultaneously collects process parameter data for this process, including press-fitting force, feed rate, and holding time, also generating a process parameter acquisition dataset with millisecond-level timestamps. These millisecond-level timestamps ensure that the process parameter data matches the operating status of the press-fitting equipment in real time, clearly reflecting the relationship between the equipment's operating parameters and the actual operating status. Synchronization of pressing actions: The time acquisition unit accurately captures and records the start time of the pressing process, the execution time of the pressing advancement sub-step, the execution time of the pressure holding sub-step, and the overall completion time of the process, generating a process execution time dataset with a unique time identifier. The unique time identifier enables full-time traceability of the pressing process and can quickly locate time-series anomalies during process execution. The data binding unit assigns a unique assembly process identifier to the pressing process, and associates and binds the sensor acquisition dataset, process parameter acquisition dataset, and process execution time dataset with the assembly process identifier according to a unified millisecond-level timestamp. At the same time, the bound multi-dimensional data is standardized and structured, ultimately forming multi-dimensional assembly data that comprehensively reflects the operating status of the pressing process.

[0050] After receiving the multi-dimensional assembly data, the fingerprint construction and storage module normalizes the data according to the rule of using sensor data as the X-axis dimension, process parameter data as the Y-axis dimension, and process execution time data as the Z-axis dimension, and extracts feature values ​​for each dimension. Normalization eliminates the differences in the dimensions of the data, making different types of data comparable. Feature value extraction allows for the condensation of core information from the multi-dimensional data and the elimination of invalid and redundant data. Subsequently, the fingerprint is fused to construct a process-level real-time 3D data fingerprint for the pressing process, ensuring that the generated real-time 3D data fingerprint accurately and comprehensively reflects the actual operating status of the pressing process. Simultaneously, this module selects standard multi-dimensional assembly data for this process to construct a standard 3D data fingerprint, completing the construction of a standard 3D data fingerprint library according to the engine product model and cylinder block piston pressing process number. The system establishes a standard library categorized by product model and process number, providing accurate and process-specific references for subsequent matching and verification. Based on the data distribution characteristics of the standard 3D fingerprint database, the initial parameters of the three-axis weighting coefficients and deviation correction coefficients for the 3D fingerprint comprehensive deviation algorithm are calibrated. These initial parameters, calibrated according to the data characteristics of the standard library, ensure that the initial calculation logic of the 3D fingerprint comprehensive deviation algorithm better aligns with the process characteristics of the pressing procedure, improving the accuracy of subsequent deviation calculations. Subsequently, the module employs a dual-storage mode, combining cloud and local servers, to synchronously store real-time 3D fingerprint data, the standard 3D fingerprint database, and the initial calibration parameters of the 3D fingerprint comprehensive deviation algorithm. This dual-storage mode achieves both remote data sharing and secure cloud backup, while also ensuring efficient data retrieval by local assembly equipment.

[0051] After receiving real-time 3D data fingerprints, standard 3D data fingerprint library data, and initial calibration parameters, the matching and early warning module accurately retrieves the corresponding standard 3D data fingerprint from the standard 3D data fingerprint library based on the assembly process identifier of the pressing process. This ensures that the reference object for deviation calculation precisely matches the currently executed pressing process, eliminating cross-process and cross-model reference deviations. Subsequently, the 3D fingerprint comprehensive deviation algorithm is called to calculate the comprehensive deviation value of the process. The formula is: ,in, This is the overall deviation value; This refers to the deviation correction coefficient calibrated based on the data distribution characteristics of the standard three-dimensional data fingerprint database; This represents the single-dimensional deviation value along the X-axis. This represents the single-dimensional deviation value along the Y-axis. This represents the single-dimensional deviation value along the Z-axis. , , The three-axis weighting coefficients are used; the application of the three-dimensional fingerprint comprehensive deviation algorithm allows the deviation calculation between real-time three-dimensional data fingerprints and standard three-dimensional data fingerprints to take into account the characteristics of sensing, process, and time dimensions, so that the comprehensive deviation value can comprehensively and accurately reflect the actual deviation of the pressing process. This module compares the calculated comprehensive deviation value with a pre-calibrated corresponding preset threshold for the pressing process. The calibration method, where the preset threshold corresponds one-to-one with the assembly process identifier, makes the threshold judgment more in line with the actual quality accuracy requirements of the pressing process. When the comprehensive deviation value does not exceed the preset threshold, the module sends the matching result and real-time three-dimensional data fingerprint back to the fingerprint construction and storage module for archiving. The archiving of qualified process data can gradually form a traceable assembly quality database, providing real and effective data support for the subsequent optimization and iteration of the standard three-dimensional data fingerprint library; when the comprehensive deviation value exceeds the preset threshold, the module... The ratio of the overall deviation value to a preset threshold is used to classify risk levels and output corresponding quality warning signals. A comprehensive deviation value between the preset threshold and 1.2 times the preset threshold triggers a Level 1 warning with a yellow warning signal; a comprehensive deviation value between 1.2 and 1.5 times the preset threshold triggers a Level 2 warning with an orange warning signal; and a comprehensive deviation value exceeding 1.5 times the preset threshold triggers a Level 3 warning with a red warning signal. Simultaneously, audible and visual alarms and background system pop-up reminders are triggered. This tiered warning system makes quality risk management more targeted, allowing staff to quickly assess the risk level based on the warning signal's severity. The Level 3 warning's audible and visual alarms and background pop-up promptly alert staff to high-level quality risks, preventing major quality problems. Different risk levels correspond to different process parameter adjustment priorities, ensuring subsequent process parameter adjustments focus on core issues and are carried out according to priority, improving adjustment efficiency. Simultaneously, this module transmits the comprehensive deviation value, deviation data, and assembly process identifier for the pressing process to the dynamic self-adjustment module. Precise process identifiers and comprehensive deviation data allow the dynamic self-adjustment module to quickly locate the pressing process requiring adjustment.

[0052] After receiving the transmitted deviation data and assembly process identifier, the dynamic self-adjustment module retrieves the process parameter list and the correlation rules between process parameters and assembly quality for the cylinder piston press-fitting process from the database based on the assembly process identifier. This dedicated process parameter list and correlation rules ensure that the process parameter adjustments are more closely aligned with the quality requirements of the press-fitting process, avoiding adjustment deviations caused by general rules. After receiving the deviation data and assembly process identifier, the adjustment calculation unit uses the dynamic adjustment algorithm to calculate the adjustment amounts for process parameters such as press-fitting force, feed speed, and holding time. The formula is: ,in, These are the adjustment amounts for each process parameter; The basic coefficient for adjusting process parameters; The correlation coefficient between process parameters and overall deviation value; This is the overall deviation value; The preset process parameter quality qualification threshold is used; the application of the process parameter dynamic adjustment algorithm makes the calculation of the adjustment amount of each process parameter more scientific, and can accurately match the degree of deviation with the adjustment range of the process parameter, avoiding secondary quality deviation caused by blind adjustment. The equipment adjustment unit establishes a stable communication connection with the control system of the pressing execution equipment, so that the adjustment command of the process parameter can be transmitted to the pressing execution equipment without deviation. Then, the equipment adjustment unit dynamically adjusts the pressing force, feed speed and holding time of the pressing execution equipment according to the calculated adjustment amount of each process parameter, so that the process parameters of the pressing execution equipment can adapt to the quality requirements of the pressing process in real time.

[0053] After receiving the adjusted process parameters from the dynamic self-adjustment module, the feedback verification module triggers the data acquisition module to re-acquire multi-dimensional assembly data of the pressing process under the adjusted process parameters. This re-acquired multi-dimensional assembly data accurately and truthfully reflects the actual operating status of the pressing process under the adjusted process parameters, providing a reliable data foundation for subsequent secondary verification. Then, using the 3D data fingerprint construction method of the fingerprint construction and storage module, a new process-level real-time 3D data fingerprint is generated based on the newly acquired multi-dimensional assembly data. The unified fingerprint construction method allows for effective and accurate comparison between the newly generated real-time 3D data fingerprint and the standard 3D data fingerprint, avoiding comparison errors caused by different construction methods. Subsequently, this new real-time 3D data fingerprint is sent back to the matching and early warning module. Following the complete process of the initial comparison, the matching and early warning module performs secondary matching verification on the new real-time 3D data fingerprint using the 3D fingerprint comprehensive deviation algorithm. Using the same process as the initial comparison makes the results of the secondary matching verification more consistent and reliable. The feedback verification module receives the data from the matching and early warning module in real time. The feedback verification module obtains the secondary matching verification results and the corresponding comprehensive deviation value. When the comprehensive deviation value does not exceed the preset threshold, the feedback verification module immediately triggers the archiving operation of the relevant information of this process adjustment and the corresponding assembly data. The archiving of effective process adjustment schemes and corresponding data can form reusable process optimization data, providing a reference for the quality control of subsequent similar pressing processes. When the comprehensive deviation value still exceeds the preset threshold, the feedback verification module sends a secondary adjustment command to the dynamic self-adjustment module, allowing the incompletely eliminated deviation to be further adjusted in process parameters. After receiving the secondary adjustment command, the dynamic self-adjustment module recalculates the adjustment amount of each process parameter based on the new deviation data and completes the adjustment of the process parameters of the pressing equipment. The feedback verification module then repeats the operation of collecting assembly data, generating new real-time three-dimensional data fingerprints, and sending them back to the matching and early warning module for matching verification. This process is continuously looped until the calculated comprehensive deviation value reaches the preset threshold requirement, ultimately forming a closed-loop linkage between quality early warning and process optimization for the pressing assembly process of the automobile engine block and piston. Figure 1As shown. This closed-loop linkage allows for continuous dynamic optimization of the quality of the press-fitting process, ensuring that the operation of each press-fitting process accurately meets the standard requirements, fundamentally improving the consistency and stability of the assembly quality of the engine block and piston press-fitting.

[0054] Example 2

[0055] The data acquisition module first conducts full data acquisition for the bolt tightening assembly process. The sensor acquisition unit accurately collects torque, pressure, and displacement sensor data during the tightening process, generating a sensor acquisition dataset with millisecond-level timestamps for all acquired data. The process parameter acquisition unit simultaneously collects process parameter data such as rotational speed, torque value, and holding time for this process, generating a process parameter acquisition dataset with millisecond-level timestamps. The time acquisition unit accurately captures and records all process execution time data, including the start time of the bolt tightening process, the start and end times of the single bolt tightening sub-step, and the completion time of tightening all bolts in the pump, generating a process execution time dataset with a unique time identifier. The data binding unit assigns a unique assembly process identifier to the bolt tightening process, and associates and binds the sensor acquisition dataset, process parameter acquisition dataset, and process execution time dataset with the assembly process identifier one by one according to a unified millisecond-level timestamp. At the same time, the bound multi-dimensional data is standardized and structured, ultimately forming multi-dimensional assembly data that can comprehensively represent the actual operating status of the bolt tightening process.

[0056] After receiving the multi-dimensional assembly data, the fingerprint construction and storage module strictly follows the rule of using sensor data as the X-axis dimension, process parameter data as the Y-axis dimension, and process execution time data as the Z-axis dimension. It normalizes the data in each dimension and extracts feature values. Normalization effectively eliminates dimensional differences between data dimensions, making data of different types such as pressure, torque, speed, and time comparable and fusionable. Feature value extraction condenses core information from massive amounts of multi-dimensional data, efficiently eliminating invalid and redundant interference data. Subsequently, it fuses and constructs a process-level real-time 3D data fingerprint for the bolt tightening process, ensuring that the generated real-time 3D data fingerprint comprehensively represents the actual operating status of the hydraulic pump bolt tightening process. Simultaneously, this module selects standard multi-dimensional assembly data for the bolt tightening process to construct a standard 3D data fingerprint, and then, based on the data distribution of the standard 3D data fingerprint library... The module calibrates the initial parameters of the three-axis weight coefficients and deviation correction coefficients of the 3D fingerprint comprehensive deviation algorithm based on the characteristics of the data distribution. The initial parameters calibrated according to the data distribution characteristics make the initial calculation logic of the 3D fingerprint comprehensive deviation algorithm more in line with the torque-sensitive characteristics of the hydraulic pump bolt tightening process. Subsequently, the module adopts a dual storage mode of cloud server and local server to store the real-time 3D data fingerprint, standard 3D data fingerprint library, and the initial calibration parameters of the 3D fingerprint comprehensive deviation algorithm synchronously on both ends. The dual storage mode of cloud server and local server not only realizes remote sharing and cloud security backup of various types of data, but also ensures the efficiency of data retrieval on the local assembly line, meeting the production needs of real-time and high-efficiency production. The synchronous storage on both ends ensures that various types of data always maintain a high degree of consistency between the cloud and local servers, effectively avoiding subsequent matching and verification errors caused by data loss, data disorder, or data inconsistency.

[0057] After receiving real-time 3D data fingerprints, standard 3D data fingerprint library data, and initial calibration parameters, the matching and early warning module retrieves the corresponding standard 3D data fingerprint from the standard 3D data fingerprint library based on the assembly process identifier of the bolt tightening process. This ensures a precise match between the reference object for deviation calculation and the currently executed bolt tightening process. Subsequently, the 3D fingerprint comprehensive deviation algorithm is invoked to calculate the comprehensive deviation value of the process. The application of the 3D fingerprint comprehensive deviation algorithm allows the deviation calculation between real-time and standard 3D data fingerprints to simultaneously consider the core features of the three dimensions of sensing, process, and time, ensuring that the calculated comprehensive deviation value fully and accurately reflects the bolt tightening. To assess actual deviations in the process and eliminate judgment errors caused by missing dimensions, this module compares the calculated comprehensive deviation value with a pre-defined threshold for the bolt tightening process. The one-to-one correspondence between the pre-defined threshold and the assembly process identifier ensures that the threshold judgment closely matches the high precision requirements of the hydraulic pump bolt tightening process. When the comprehensive deviation value does not exceed the pre-defined threshold, the module transmits the matching result and real-time 3D data fingerprint back to the fingerprint construction and storage module for archiving. Continuous archiving of qualified process data gradually forms a fully traceable hydraulic pump assembly quality database, providing accurate and effective on-site production data for subsequent optimization and iteration of the standard 3D data fingerprint library. Supported by data, when the overall deviation value exceeds a preset threshold, the module accurately classifies the risk level based on the ratio of the overall deviation value to the preset threshold and outputs a corresponding quality warning signal. A comprehensive deviation value between the preset threshold and 1.2 times the preset threshold triggers a Level 1 warning and outputs a yellow warning signal; a comprehensive deviation value between 1.2 and 1.5 times the preset threshold triggers a Level 2 warning and outputs an orange warning signal; and a comprehensive deviation value exceeding 1.5 times the preset threshold triggers a Level 3 warning and outputs a red warning signal. Simultaneously, audible and visual alarms and background system pop-up reminders are triggered. This tiered warning system makes the handling of quality risks more targeted, allowing on-site personnel to accurately assess the risk level based on the warning signal. The system enables rapid and accurate assessment of risk levels. The three-level early warning system, with its audible and visual alarms and background system pop-ups, promptly and effectively alerts staff to high-level quality risks, preventing major quality problems caused by negligence. Different risk levels correspond to different process parameter adjustment priorities, allowing subsequent process parameter adjustments to focus on core issues and proceed in an orderly manner according to priority, significantly improving the efficiency of process adjustments. Simultaneously, the module transmits the comprehensive deviation value, complete deviation data, and the assembly process identifier for the bolt tightening operation to the dynamic self-adjustment module. The unique process identifier and comprehensive deviation data enable the dynamic self-adjustment module to quickly and accurately locate the bolt tightening operation to be adjusted.

[0058] like Figure 3As shown, after receiving the transmitted deviation data and assembly process identifier, the parameter retrieval unit retrieves the process parameter list and the correlation rules between process parameters and assembly quality for the hydraulic pump bolt tightening process from a dedicated database based on the assembly process identifier. This dedicated process parameter list and correlation rules ensure that the process parameter adjustment is highly aligned with the quality requirements of the bolt tightening process, effectively avoiding adjustment deviations caused by general rules. After receiving the deviation data and assembly process identifier, the adjustment calculation unit uses a dynamic process parameter adjustment algorithm to accurately calculate the adjustment amounts of process parameters such as speed, torque, and holding time. The application of the dynamic process parameter adjustment algorithm makes the calculation of adjustment amounts for each process parameter more scientific, based on the degree of deviation. The system determines the degree of influence of each process parameter on assembly quality, matches the degree of deviation with the adjustment range of process parameters, and effectively avoids secondary quality deviations caused by blind adjustment. The equipment adjustment unit and the control system of the bolt tightening execution equipment achieve a stable and interference-free communication connection, allowing the adjustment commands of process parameters to be transmitted to the control system of the bolt tightening execution equipment without deviation and with precision. Subsequently, the equipment adjustment unit dynamically adjusts the process parameters of the tightening execution equipment, such as speed, torque value, and holding time, based on the calculated adjustment amount of each process parameter. This ensures that the process parameters of the bolt tightening execution equipment can be adapted to the quality requirements of the bolt tightening process in real time, effectively reducing quality risks such as hydraulic pump seal failure and bolt loosening caused by insufficient torque, excessive speed, or insufficient holding time.

[0059] After receiving the adjusted process parameters from the dynamic self-adjustment module, the feedback verification module immediately triggers the data acquisition module to re-acquire multi-dimensional assembly data of the hydraulic pump bolt tightening assembly process under the adjusted process parameters. This re-acquired multi-dimensional assembly data accurately reflects the actual operating status of the bolt tightening process under the adjusted process parameters, providing a true and effective data foundation for subsequent secondary matching verification. Then, using the 3D data fingerprint construction method of the fingerprint construction and storage module, a new process-level real-time 3D data fingerprint is generated based on the newly acquired multi-dimensional assembly data. The unified 3D data fingerprint construction method allows for effective comparison and analysis between the newly generated real-time 3D data fingerprint and the standard 3D data fingerprint, avoiding comparison errors caused by different construction methods. Subsequently, this new real-time 3D data fingerprint is sent back to the matching and early warning module. Following the complete process of the initial comparison verification, the matching and early warning module performs secondary matching verification on the new real-time 3D data fingerprint using the 3D fingerprint comprehensive deviation algorithm. Using the same process as the initial comparison verification makes the results of the secondary matching verification more consistent and objectively referential. The feedback verification module receives the data from the matching and early warning module in real time. The secondary matching verification result of the block and the corresponding comprehensive deviation value are used. When the comprehensive deviation value does not exceed the preset threshold, the feedback verification module immediately triggers the archiving operation of relevant information on the current process parameter adjustment and the corresponding assembly data. The archiving of effective process adjustment schemes and corresponding assembly data can form reusable and referable process optimization data, providing valuable on-site practical experience for the quality control of bolt tightening process of the same model of hydraulic pump. When the comprehensive deviation value still exceeds the preset threshold, the feedback verification module sends a secondary adjustment command to the dynamic self-adjustment module, allowing the incompletely eliminated deviation to be further precisely adjusted in process parameters. The dynamic self-adjustment module recalculates the adjustment amount of each process parameter according to the secondary adjustment command and the new deviation data and completes the adjustment of the process parameters of the tightening execution equipment. The feedback verification module then repeats the operation of collecting assembly data, generating new real-time three-dimensional data fingerprints, and sending them back to the matching and early warning module for matching verification. This closed-loop process is continuously cyclical until the calculated comprehensive deviation value meets the preset threshold requirements, ultimately forming a closed-loop linkage between quality early warning and process self-optimization for the bolt tightening assembly process of engineering machinery hydraulic pumps. Figure 2 As shown.

[0060] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An assembly quality early warning and process self-optimization system based on data fingerprint matching, characterized in that, The system includes: a data acquisition module, a fingerprint construction and storage module, a matching and early warning module, a dynamic self-adjustment module, and a feedback verification module; The data acquisition module collects various sensor data, process parameter data, and process execution time data throughout the entire assembly process, and binds all collected data with assembly process identifiers to form multi-dimensional assembly data. The fingerprint construction and storage module receives multi-dimensional assembly data, constructs real-time three-dimensional data fingerprints at the process level, builds and stores a standard three-dimensional data fingerprint library corresponding to each assembly process, and completes the calibration and storage of the initial parameters of the three-dimensional fingerprint comprehensive deviation algorithm. The matching and early warning module receives real-time 3D data fingerprints, standard 3D data fingerprint database data, and initial calibration parameters; calculates a comprehensive deviation value using a 3D fingerprint comprehensive deviation algorithm; compares the comprehensive deviation value with a preset threshold; archives the matching result if the deviation value does not exceed the threshold; and outputs a corresponding quality early warning signal if the deviation value exceeds the threshold. Simultaneously, the deviation data and the corresponding assembly process identifier are transmitted to the dynamic self-adjustment module. The dynamic self-adjustment module: Based on the transmitted deviation data and assembly process identifier, it uses a process parameter dynamic adjustment algorithm to calculate the adjustment amount of each process parameter, and dynamically adjusts the process parameters of the assembly execution equipment according to the adjustment amount. The feedback verification module receives the adjusted process parameters, collects the corresponding assembly data and generates a new real-time three-dimensional data fingerprint, and sends it back to the matching and early warning module for secondary matching verification; based on the verification result, it triggers data archiving or secondary adjustment to form a closed-loop linkage.

2. The assembly quality early warning and process self-optimization system based on data fingerprint matching according to claim 1, characterized in that, The data acquisition module includes a sensor acquisition unit, a process parameter acquisition unit, a time acquisition unit, and a data binding unit; The sensing and acquisition unit collects pressure sensing data, torque sensing data, displacement sensing data, and temperature sensing data throughout the entire assembly process, and generates a sensing and acquisition dataset with millisecond-level timestamps. The process parameter acquisition unit collects process parameter data such as rotational speed, feed rate, pressing force, and holding time during equipment operation, and generates a process parameter acquisition dataset with millisecond-level timestamps. The time acquisition unit captures and records the start time of the assembly process, the start and end times of each sub-step, and the overall completion time of the process, generating a process execution time dataset with a unique time identifier. The data binding unit assigns a unique assembly process identifier to each executed assembly process, and associates and binds the sensor acquisition dataset, process parameter acquisition dataset, and process execution time dataset with the assembly process identifier one by one according to a unified millisecond-level timestamp. The bound multi-dimensional data is then standardized and structured to form multi-dimensional assembly data.

3. The assembly quality early warning and process self-optimization system based on data fingerprint matching according to claim 1, characterized in that, The fingerprint construction and storage module includes a fingerprint construction unit, a fingerprint database construction unit, a parameter calibration unit, and a data storage unit. The fingerprint construction unit uses sensor data as the X-axis dimension, process parameter data as the Y-axis dimension, and process execution time data as the Z-axis dimension. After normalizing the data in each dimension, feature values ​​are extracted and fused to form a process-level real-time three-dimensional data fingerprint. The fingerprint database construction unit selects multi-dimensional assembly data from the assembly process to construct a three-dimensional data fingerprint, which is then processed to obtain a standard three-dimensional data fingerprint. The standard three-dimensional data fingerprint database is then constructed by classifying the product model and process number. The parameter calibration unit calibrates the initial parameters of the three-axis weight coefficients and deviation correction coefficients of the three-dimensional fingerprint comprehensive deviation algorithm based on the data distribution characteristics of the standard three-dimensional fingerprint database. The data storage unit adopts a dual storage mode of cloud server and local server to synchronously store real-time 3D data fingerprint, standard 3D data fingerprint library, and initial calibration parameters of 3D fingerprint comprehensive deviation algorithm on both ends.

4. The assembly quality early warning and process self-optimization system based on data fingerprint matching according to claim 1, characterized in that, The specific steps for the matching and early warning module to calculate the comprehensive deviation value using the three-dimensional fingerprint comprehensive deviation algorithm are as follows: Based on the assembly process identifier, retrieve the corresponding standard 3D data fingerprint from the standard 3D data fingerprint database. Let the triaxial feature value of the standard 3D data fingerprint be... The triaxial feature values ​​of real-time 3D data fingerprints are The initial calibration parameters of the three-dimensional fingerprint comprehensive deviation algorithm include the three-axis weight coefficients. , , and deviation correction factor ,and ; Calculate the three-axis single-dimensional deviation value using the formula: , , ,in, This represents the single-dimensional deviation value along the X-axis. This represents the single-dimensional deviation value along the Y-axis. This represents the single-dimensional deviation value along the Z-axis. The comprehensive deviation value is calculated using the formula of the three-dimensional fingerprint comprehensive deviation algorithm. , ,in, The deviation correction coefficient is determined based on the data distribution characteristics of the standard three-dimensional data fingerprint database, and its value range is (0,1]. The calculated comprehensive deviation value Compare with a preset threshold, when If the value is less than or equal to a preset threshold, the matching results and real-time 3D data fingerprints will be sent back to the fingerprint construction and storage module for archiving; if the value is less than or equal to a preset threshold, the matching results and real-time 3D data fingerprints will be sent back to the fingerprint construction and storage module for archiving. If a preset threshold is set, a corresponding quality warning signal will be output, and... Deviation data and corresponding assembly process identifiers are synchronously transmitted to the dynamic self-adjustment module.

5. The assembly quality early warning and process self-optimization system based on data fingerprint matching according to claim 1, characterized in that, In the matching and early warning module, preset thresholds correspond one-to-one with assembly process identifiers, and are synchronously calibrated by the fingerprint construction and storage module when building a standard three-dimensional data fingerprint database. The quality early warning signal includes assembly process identifiers, deviation information, and risk levels, with the risk level based on the comprehensive deviation value. Size division, A level one warning is issued when the threshold value is between 1.2 times the preset threshold value, and a yellow warning signal is output. A level 2 warning is issued when the value is between 1.2 and 1.5 times the preset threshold, and an orange warning signal is output. If the risk exceeds 1.5 times the preset threshold, a Level 3 warning is issued, a red warning signal is output, and an audible and visual alarm and a pop-up reminder from the background system are triggered simultaneously. Different risk levels correspond to different priorities for adjusting process parameters.

6. The assembly quality early warning and process self-optimization system based on data fingerprint matching according to claim 1, characterized in that, The dynamic self-adjustment module includes a parameter retrieval unit, an adjustment amount calculation unit, and a device adjustment unit; The parameter retrieval unit is used to retrieve the list of process parameters for the corresponding assembly process and the association rules between process parameters and assembly quality according to the assembly process identifier. The adjustment calculation unit is used to receive deviation data and assembly process identifiers, and to calculate the adjustment amount of each process parameter using a dynamic adjustment algorithm for process parameters. The equipment adjustment unit is used to communicate with the control system of the assembly execution equipment and adjust the process parameters of the assembly execution equipment according to the adjustment amount.

7. The assembly quality early warning and process self-optimization system based on data fingerprint matching according to claim 1 or 6, characterized in that, The specific steps for calculating the adjustment amount of each process parameter using the dynamic adjustment algorithm for process parameters are as follows: Let the overall deviation value transmitted by the matching and early warning modules be... ; The preset process parameter quality qualification threshold is ; The basic coefficient for adjusting process parameters is The value is determined by the process characteristics of the corresponding assembly process and ranges from (0,2]. Let the process parameters to be adjusted be: , The number of process parameters to be adjusted; the correlation coefficient between the process parameters and the overall deviation value is... ,and The correlation coefficient is determined by the degree of influence of process parameters on assembly quality; the higher the degree of influence, the higher the correlation coefficient. The larger; The adjustment amount of each process parameter is calculated using a dynamic adjustment algorithm. , , among which, when ,but A negative value indicates that the process parameters need to be adjusted downwards; when ,but A positive value indicates that the process parameters need to be adjusted upwards; Calculate the adjusted process parameter values and will As the target value for adjusting the process parameters of the assembly execution equipment.

8. The assembly quality early warning and process self-optimization system based on data fingerprint matching according to claim 1, characterized in that, The process of secondary matching verification in the feedback verification module is as follows: after receiving the adjusted process parameters, the data acquisition module re-acquires the multi-dimensional assembly data of the corresponding assembly process under the adjusted process parameters. The three-dimensional data fingerprint construction method of the fingerprint construction and storage module generates a new process-level real-time three-dimensional data fingerprint based on the collected multi-dimensional assembly data; the new process-level real-time three-dimensional data fingerprint is sent back to the matching and early warning module, which performs secondary matching and verification on the new real-time three-dimensional data fingerprint according to the initial comparison process.

9. The assembly quality early warning and process self-optimization system based on data fingerprint matching according to claim 1, characterized in that, The closed-loop linkage process in the feedback verification module is as follows: receiving the secondary matching verification result and comprehensive deviation value from the matching and early warning module; when the comprehensive deviation value does not exceed the preset threshold, triggering process adjustment and archiving of corresponding assembly data; when the comprehensive deviation value still exceeds the threshold, sending a secondary adjustment command to the dynamic self-adjustment module; after the dynamic self-adjustment module completes the command adjustment, the feedback verification module repeatedly collects assembly data, generates new real-time three-dimensional data fingerprints and sends them back for verification until the deviation value meets the standard, thus forming a closed-loop linkage.

Images