A Product Quality Assessment Method and System Based on Vernier Caliper Measurement Big Data

Abstract

This application relates to the field of product quality assessment technology, specifically to a product quality assessment method and system based on vernier caliper measurement big data. It includes: collecting vernier caliper measurement data along with its local timestamp, environmental vibration data, network transmission status data, and measurement equipment calibration status data; obtaining network delay error characteristics, vibration error characteristics, and linearity deviation error characteristics; calculating the total deviation and decomposing it into vibration error components, network delay error components, linearity deviation error components, and other error components; correcting the vernier caliper measurement data based on the vibration error components and linearity deviation error components to obtain corrected measurement values; and comparing the corrected measurement values ​​with the product design tolerance range to output a quality acceptance result. This solves the technical dilemma of traditional quality control methods being unable to effectively identify and trace potential defects, leading to high misjudgment rates and impacting production efficiency and product quality.

Description

Technical Field

[0001] This application relates to the field of product quality assessment technology, specifically to a product quality assessment method and system based on big data measurement using vernier calipers. Background Technology

[0002] In the manufacturing of high-precision components, accurate quality assessment is crucial for ensuring product performance and reliability. Industries with extremely high dimensional accuracy requirements, such as advanced robotics, aerospace, and medical devices, commonly employ automated vernier calipers on production lines for real-time inspection, rapidly collecting large-scale data to support quality control. However, the real-world environment is complex: mechanical vibrations from heavy equipment, uncertainties in time recording due to wireless transmission, and systematic deviations caused by improper equipment calibration all contribute to weakening the accuracy of measurement data and the reliability of assessments, increasing misjudgments and impacting efficiency and quality.

[0003] In large-scale production facilities, automated vernier calipers are often used alongside heavy equipment. The low-frequency vibrations generated by the operation of heavy equipment are transmitted to the caliper mounting base, causing minute and non-random instantaneous fluctuations between the caliper jaws and the workpiece surface, resulting in measurement deviations. At the same time, the variable latency introduced by wireless networks causes random differences between the local timestamp and the server's received timestamp, making it difficult to accurately align the vernier caliper measurement data with the vibration sensor data, further masking the regular deviations caused by vibration. In addition, inadequate daily calibration can cause the calipers to deviate from linearity within a certain measuring range, forming a persistent systematic error.

[0004] The combination of these factors results in quality data exhibiting complex characteristics such as nonlinearity, multi-source heterogeneity, and ambiguous temporal correlation: periodic fluctuations caused by vibration, deviations in equipment linearity, and time recording errors are intertwined. Traditional one-dimensional statistics (such as mean and standard deviation) or simple regression are no longer sufficient to distinguish the sources of error and handle their interactions.

[0005] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0006] This application discloses a product quality assessment method and system based on vernier caliper measurement big data, which aims to solve the technical dilemma in the field of high-precision parts manufacturing where traditional quality control methods are difficult to effectively identify and trace potential defects, resulting in high misjudgment rates and affecting production efficiency and product quality.

[0007] The technical solution of this application is as follows: In a first aspect, this application discloses a product quality assessment method based on big data from vernier caliper measurements, comprising the following steps: Collect vernier caliper measurement data and its local timestamp, environmental vibration data, network transmission status data, and measurement equipment calibration status data. The network transmission status data includes the signal strength and channel occupancy before transmission, access end load and channel interference, and server receiving timestamp. The measurement equipment calibration status data includes the measurement results of the gold sample and environmental and equipment operating parameters. Based on the local timestamp of the vernier caliper measurement data, the server receiving timestamp, and the network transmission status data, timestamp calibration is performed to estimate the actual network transmission delay, obtain the true measurement timestamp, and the actual network transmission delay and its fluctuation relative to the historical baseline are determined as network delay error characteristics. Based on the actual measurement timestamp, the vernier caliper measurement data and environmental vibration data are time-aligned, and time-frequency analysis is performed to obtain the vibration phase at the moment of measurement. Based on this, the instantaneous deviation caused by vibration is calculated and determined as the vibration error characteristic. Based on the measurement results of the gold sample and combined with the environmental and equipment operating parameters, a linearity deviation model of the measuring equipment under different measurement ranges and operating conditions is established, and the systematic deviation caused by the linearity deviation of the equipment is calculated and determined as the linearity deviation error characteristic. Calculate the total deviation between the vernier caliper measurement data and the target size, and decompose it into vibration error components corresponding to vibration error characteristics, network delay error components corresponding to network delay error characteristics, linearity deviation error components corresponding to linearity deviation error characteristics, and other error components; The measurement data of the vernier caliper is corrected based on the vibration error component and the linearity deviation error component to obtain the corrected measurement value. The corrected measurement value is then compared with the product design tolerance range to output the quality qualification result.

[0008] Furthermore, when calculating the total deviation between the vernier caliper measurement data and the target size, and decomposing it into vibration error components corresponding to vibration error characteristics, network delay error components corresponding to network delay error characteristics, linearity deviation error components corresponding to linearity deviation error characteristics, and other error components, it includes: Based on the actual measurement timestamp, the vibration intensity index is calculated from the environmental vibration data within a preset time window, and a preset vibration intensity threshold is set. When the vibration intensity index exceeds the preset vibration intensity threshold, the decomposition priority of the error components is dynamically adjusted, and the vibration error components are estimated first. Based on the characteristics of network latency error, the network latency error component is extracted from the total deviation; The decomposition and iterative adjustment process includes: estimating the vibration error components based on the vibration phase and intensity indices corresponding to the actual measurement timestamps, and combining the statistical relationship between the vibration error components and the total deviation under similar vibration phase and intensity conditions in historical measurement data; after separating the estimated vibration error components and network delay error components from the total deviation, calculating the correlation between the remaining deviation and the linearity deviation error characteristics, until the absolute value of the correlation coefficient between the vibration error components and the remaining deviation relative to the linearity deviation error characteristics is lower than a preset low correlation threshold. This remaining deviation is the total deviation minus the vibration error components and the network delay error components. The initial estimated values ​​of the linearity deviation error components are obtained from the linearity deviation model based on the environmental and equipment operating parameters. The initial estimates of the linearity deviation error components are adjusted based on the residual bias to obtain updated linearity deviation error components. Randomness checks are performed on the remaining error components after stripping the vibration error component, linearity deviation error component, and network delay error component. When a preset non-random pattern is detected, an anomaly warning or model update is triggered.

[0009] Furthermore, when calculating the total deviation between the vernier caliper measurement data and the target size, and decomposing it into vibration error components corresponding to vibration error characteristics, network delay error components corresponding to network delay error characteristics, linearity deviation error components corresponding to linearity deviation error characteristics, and other error components, the method further includes: Continuously monitor auxiliary sensor data, including ambient temperature, humidity, air pressure, power supply voltage fluctuations, wireless channel interference intensity, heavy equipment operating power, tool wear status, and operator operation logs; Statistical analysis is performed on other error components to obtain statistical characteristics, and the statistical characteristics are compared with the preset random noise statistical characteristics. When the comparison results show significant differences and a non-random pattern, a multi-dimensional environment-device state vector is constructed based on the auxiliary sensor data at the current measurement time. The multi-dimensional environment-device state vector is matched with historical state vectors in a pre-set unknown error pattern library to identify whether there are similar combination patterns. When a similar combination pattern is matched, the feature signature of the non-random pattern is compared with the pattern signature recorded in the unknown error pattern library. If the comparison result shows that there is similarity, the non-random pattern is classified as a known unknown error source, and the potential system error source and suggested investigation direction are output based on the source tracing information recorded in the unknown error pattern library. When no similar combination pattern is matched, the non-random pattern and its corresponding multi-dimensional environment-device state vector are recorded as new unknown error patterns and the unknown error pattern library is updated. Based on the characteristics of the identified unknown error sources, the control parameters are dynamically adjusted, and the historical baseline of the linearity deviation model, the network delay error characteristics, and the functional relationship used to generate vibration error characteristics are updated. The control parameters include at least a preset time window, a preset vibration intensity threshold, a decomposition priority, and a preset low correlation threshold.

[0010] Furthermore, after comparing the corrected measurement value with the product design tolerance range to output a quality acceptance result, the method also includes: Determine whether the quality pass determination result meets the preset persistent deviation determination condition. The persistent deviation determination condition is set based on the persistent deviation index, which includes at least the rolling window misjudgment rate calculated based on the true value of the gold sample. When the quality pass determination result meets the preset persistent deviation determination condition, the quality pass determination result is statistically analyzed based on the preset sliding time window to identify the type and trend of persistent deviation. Based on the type and trend of persistent deviation, the estimation models of the error components that need to be checked first are determined. These estimation models include the vibration error component estimation model and the linearity deviation error component estimation model. Based on the corresponding environmental vibration data and corrected measurements, the functional relationship used to generate vibration error characteristics is re-evaluated; Based on the measurement results of the corresponding gold samples and the environmental and equipment operating parameters, the fitting accuracy of the measurement equipment under the linearity deviation model is re-evaluated, and the model parameters that need to be adjusted are determined. Based on the re-evaluation results of the functional relationship used to generate vibration error characteristics and the re-evaluation results of the linearity deviation model, feedback iterative adjustment is performed to adjust the parameters of the vibration error component estimation model and the linearity deviation error component estimation model. The parameters of the adjusted vibration error component estimation model and linearity deviation error component estimation model are used to back-calculate the historical data, regenerate the vibration error component and linearity deviation error component, update and correct the measured values, and re-output the quality qualification judgment result. When the persistent deviation index after backtracking calculation is lower than the preset deviation threshold, the maximum number of feedback iterations is reached, or the change in model parameters is lower than the preset convergence threshold, the parameters of the vibration error component estimation model and the linearity deviation error component estimation model are applied to the subsequent real-time evaluation; if the termination condition is not met, the feedback iteration adjustment continues.

[0011] Furthermore, the corrected measured values ​​are compared with the product design tolerance range to output a quality acceptance result, including: Obtain the distance between the corrected measurement value and the nearest limit value within the product design tolerance range, where the nearest limit value of the product design tolerance range is the limit value closer to the corrected measurement value among the upper and lower limits of the product design tolerance range; Obtain the current environmental parameters and equipment operating parameters; When the distance is less than the preset critical distance, start the enhanced determination mode: Perform repeated measurements for a preset number of times within a preset short time window, continuously collect the corrected measurement values and their corresponding environmental parameters and equipment operating parameters; Based on the environmental parameters and equipment operating parameters corresponding to each measurement, combined with the historical statistical relationship under similar environments and equipment states, calculate the fluctuation range of the corresponding corrected measurement value; When the corrected measurement values and their fluctuation ranges for multiple consecutive times within the preset number of measurements are all within the product design tolerance range, it is determined as qualified; when the corrected measurement values and their fluctuation ranges for multiple consecutive times within the preset number of measurements are all outside the product design tolerance range, it is determined as unqualified; When the corrected measurement value and its fluctuation range cross the product design tolerance range, calculate the instantaneous change rate of the environmental parameters and equipment operating parameters; When the instantaneous change rate exceeds the preset instantaneous change rate threshold, extend the preset short time window and / or increase the preset number of measurements, and perform determination according to the enhanced determination mode based on the updated data.

[0012] Further, when comparing the corrected measurement value with the product design tolerance range to output a quality qualified determination result, it also includes: Continuously monitor the distance, and record the historical sequence of the distance and its change rate within a preset sliding time window; When it is determined based on the historical sequence that there is a fluctuation greater than the preset duration in the neighborhood of the preset critical distance for the distance, dynamically adjust the value of the preset critical distance according to the fluctuation frequency and amplitude of the distance; When the start condition or exit condition of the enhanced determination mode is satisfied, wait for the preset mode switching delay time, and only when the start condition or exit condition is still continuously satisfied after the preset mode switching delay time expires, perform the corresponding start or exit; Set the start setting priority of the enhanced determination mode according to the importance of the current production task.

[0013] Further, setting the start setting priority of the enhanced determination mode according to the importance of the current production task includes: Identify the type of the current part to be measured and its corresponding production task; Obtain the corresponding quality determination priority from the preset rule library according to the part type and production task, and set the start setting priority of the enhanced determination mode accordingly; Monitor system resource status, which includes at least computing resource usage and measurement equipment load; When the distance is less than a preset critical distance, the decision to activate the enhanced judgment mode is made based on the quality judgment priority and system resource status: when the quality judgment priority is high and the system resource status meets the preset resource threshold, the enhanced judgment mode is activated immediately; when the quality judgment priority is high but the system resource status does not meet the preset resource threshold, the enhanced judgment mode for parts with low activation priority is paused or switched to the normal judgment mode, and the enhanced judgment mode for parts with high activation priority is activated; when the quality judgment priority is low and the system resource status meets the preset resource threshold, the enhanced judgment mode is activated; when the quality judgment priority is low and the system resource status does not meet the preset resource threshold, the activation of the enhanced judgment mode is delayed or the normal judgment mode is maintained. After the enhanced judgment mode is activated, the preset number of measurements and the sampling frequency within the preset short time window are dynamically adjusted according to the quality judgment priority and system resource status.

[0014] Furthermore, after the enhanced judgment mode is activated, it also includes: Continuously monitor the data stream status of the measuring equipment and auxiliary sensors, including at least data packet integrity, transmission rate, and effective range of sensor output; When any interruption or abnormality is detected in the status of any data stream, a fault warning is triggered, and the data acquisition and judgment process in the enhanced judgment mode is suspended. During the pause, record the environmental parameters and equipment operating parameters at the time of the fault, and store the valid measurement data collected before the fault occurred; Continuously monitor the recovery status of faulty measuring equipment or auxiliary sensors; Once the faulty measuring equipment or auxiliary sensor returns to normal, based on the recorded environmental parameters and equipment operating parameters at the time of the fault, it is determined whether the suspended judgment task needs to be retrospectively processed. When retrospective processing is required, measurement data for the corresponding time period is re-collected after the fault is recovered, and merged with the stored valid measurement data to restart the judgment process in the enhanced judgment mode. When it is determined that no retrospective processing is required, the data acquisition and judgment process in the enhanced judgment mode is smoothly restored after the fault is recovered.

[0015] Furthermore, when the enhanced judgment mode is enabled and the data stream is normal, before comparing the corrected measurement value with the product design tolerance range, the method further includes: Based on vibration error characteristics, network delay error characteristics, and linearity deviation error characteristics, the corresponding confidence index is calculated and an error feature confidence vector is formed. Confidence gating coefficients are generated based on the error feature confidence vector, which are used to dynamically adjust the control parameters of the decomposition iteration, including adaptive adjustment of preset low correlation threshold, decomposition step size and decomposition priority; When the confidence index of any error feature is lower than the preset confidence threshold, phase stabilization resampling is triggered within a preset short time window, and additional measurements are performed in the phase stabilization interval selected based on the actual measurement timestamp and vibration phase, while the historical baseline of the network delay error feature is updated synchronously; when the phase stabilization resampling and the updated historical baseline of the network delay error feature still indicate high uncertainty, fast verification point measurement is performed on the gold sample to constrain the local extrapolation distance of the linearity deviation from the model. Based on the confidence gating coefficient, the vibration error component and the linearity deviation error component are correspondingly contracted or amplified to obtain the confidence gating correction measurement value, which is then used to replace the correction measurement value for subsequent quality acceptance judgment.

[0016] Secondly, this application also discloses a product quality assessment system based on vernier caliper measurement big data, including: The data acquisition module is used to collect vernier caliper measurement data and its local timestamp, environmental vibration data, network transmission status data, and measurement equipment calibration status data. The network transmission status data includes the signal strength and channel occupancy before transmission, access end load and channel interference, and server receiving timestamp. The measurement equipment calibration status data includes the measurement results of the gold sample and environmental and equipment operating parameters. The first feature extraction module is used to perform timestamp calibration based on the local timestamp of the vernier caliper measurement data, the server receiving timestamp, and the network transmission status data, estimate the actual network transmission delay, obtain the real measurement timestamp, and determine the actual network transmission delay and its fluctuation relative to the historical baseline as network delay error features. The second feature extraction module is used to time-align the vernier caliper measurement data with the environmental vibration data using the actual measurement timestamp, and to perform time-frequency analysis to obtain the vibration phase at the moment of measurement. Based on this, the instantaneous deviation caused by the vibration is calculated and determined as the vibration error feature. The third feature extraction module is used to establish a linearity deviation model of the measuring equipment under different measurement ranges and operating conditions based on the measurement results of the gold sample and combined with environmental and equipment operating parameters. Based on this, the system deviation caused by the linearity deviation of the equipment is calculated and determined as the linearity deviation error feature. The error component decomposition module is used to calculate the total deviation between the vernier caliper measurement data and the target size, and decompose it into vibration error components corresponding to vibration error characteristics, network delay error components corresponding to network delay error characteristics, linearity deviation error components corresponding to linearity deviation error characteristics, and other error components. The judgment module is used to correct the vernier caliper measurement data based on the vibration error component and the linearity deviation error component, obtain the corrected measurement value, and compare the corrected measurement value with the product design tolerance range to output the quality qualified judgment result.

[0017] Beneficial effects This application provides a product quality assessment method based on big data from vernier caliper measurements. By integrating multi-source heterogeneous data, including vernier caliper measurement data, environmental vibration data, network transmission status data, and measurement equipment calibration status data, and performing refined processing and analysis on these data, it effectively solves the problems of inaccurate measurement data and unreliable quality assessment caused by the complex interplay of factors such as mechanical vibration, network latency, and equipment linearity deviation in the prior art. Attached Figure Description

[0018] Figure 1 This application provides a flowchart illustrating a product quality assessment method based on big data measurement using vernier calipers.

[0019] Figure 2 This application provides a flowchart of a product quality assessment system based on big data measurement using vernier calipers.

[0020] In the diagram: 1. Data acquisition module; 2. First feature extraction module; 3. Second feature extraction module; 4. Third feature extraction module; 5. Error component decomposition module; 6. Judgment module. Detailed Implementation

[0021] The technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. In the description of this application, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0022] Reference Figure 1 This application proposes a product quality assessment method based on big data measurement using vernier calipers, including: S1000: Collects vernier caliper measurement data and its local timestamp, environmental vibration data, network transmission status data, and measurement equipment calibration status data. The network transmission status data includes the signal strength and channel occupancy before transmission, access end load and channel interference, and server receiving timestamp. The measurement equipment calibration status data includes the measurement results of the gold sample and environmental and equipment operating parameters. For example, sensors can be integrated or connected to vernier calipers to acquire measurement data and local timestamps; high-precision accelerometers can be deployed near the measurement area to acquire environmental vibration data; network transmission status data can be acquired through network monitoring tools or by embedding log recording functions in the data transmission module; and standard gauge blocks (i.e., gold samples) can be used regularly for measurement, and the measurement results and parameters such as ambient temperature, humidity, and air pressure can be recorded to acquire calibration status data of the measuring equipment.

[0023] S2000: Based on the local timestamp of the vernier caliper measurement data, the server receiving timestamp, and the network transmission status data, timestamp calibration is performed to estimate the actual network transmission delay, obtain the true measurement timestamp, and the actual network transmission delay and its fluctuation relative to the historical baseline are determined as network delay error characteristics. For example, compare the difference between the local timestamp and the server's received timestamp, and combine this with factors such as signal strength and channel occupancy to estimate network transmission latency.

[0024] S3000: Based on the actual measurement timestamp, the vernier caliper measurement data and environmental vibration data are time-aligned, and time-frequency analysis is performed to obtain the vibration phase at the moment of measurement. Based on this, the instantaneous deviation caused by vibration is calculated and determined as the vibration error characteristic. For example, interpolation algorithms or synchronous sampling techniques can be used to achieve time alignment; after alignment, fast Fourier transform (FFT) or wavelet transform can be performed on the vibration data to extract the vibration frequency and phase information at the measurement time, and the instantaneous deviation can be calculated by combining the vibration amplitude and the preset model.

[0025] S4000: Based on the measurement results of the gold sample and combined with the environmental and equipment operating parameters, a linearity deviation model of the measuring equipment under different measurement ranges and operating conditions is established, and the systematic deviation caused by the linearity deviation of the equipment is calculated and determined as the linearity deviation error characteristic. For example, regularly measure gold samples of various sizes, record the difference between the measurement results and the true values, and record temperature, humidity, equipment operating time, etc., to construct a multinomial regression model or neural network model to describe the linearity deviation under different operating conditions.

[0026] S5000: Calculates the total deviation between the vernier caliper measurement data and the target size, and decomposes it into vibration error components corresponding to vibration error characteristics, network delay error components corresponding to network delay error characteristics, linearity deviation error components corresponding to linearity deviation error characteristics, and other error components. For example, total deviation = measured value − target size; decomposition can be achieved using methods such as statistical regression, Kalman filtering, or machine learning; for instance, using multiple linear regression, the total deviation is set as the dependent variable, and vibration error characteristics, network delay error characteristics, and linearity deviation error characteristics are used as independent variables, and the error components are estimated through regression coefficients.

[0027] S6000: Corrects the vernier caliper measurement data based on the vibration error component and the linearity deviation error component, obtains the corrected measurement value, and compares the corrected measurement value with the product design tolerance range to output the quality qualification judgment result.

[0028] For example, the original measured value is subtracted from the estimated vibration error component and linearity deviation error component to obtain the corrected measured value; it is then compared with the upper and lower limits of the product design tolerance range. If it falls within the range, it is considered qualified; otherwise, it is considered unqualified.

[0029] Specifically, vernier caliper measurement data refers to the dimensional values ​​directly obtained by the vernier caliper during the measurement process, which are usually output in digital form.

[0030] Local timestamps refer to the time information recorded by the vernier caliper when measurement data is generated.

[0031] Environmental vibration data refers to data collected by vibration sensors that reflects the mechanical vibration of the measured environment, such as vibration acceleration and frequency.

[0032] Network transmission status data includes signal strength and channel occupancy before transmission, access load and channel interference, and server reception timestamp. This data is used to assess network quality and latency during data transmission.

[0033] Measurement equipment calibration status data includes measurement results of gold samples as well as environmental and equipment operating parameters. This data is used to evaluate the performance and calibration status of the measurement equipment. Gold samples are standard pieces with known precise dimensions used to calibrate and verify the accuracy of the measurement equipment.

[0034] A true measurement timestamp refers to time information that, after calibration, accurately reflects the actual moment when the measurement event occurred.

[0035] Network latency error characteristics refer to the actual network transmission latency and its fluctuation relative to the historical baseline, reflecting the impact of network transmission on the accuracy of measurement timestamps.

[0036] Vibration error characteristics refer to the instantaneous measurement deviation caused by environmental vibration, which is obtained through time-frequency analysis of vibration data.

[0037] Linearity deviation error characteristics refer to the systematic deviations caused by linearity deviations in measuring equipment under different measurement ranges and operating conditions.

[0038] Total deviation refers to the overall difference between the vernier caliper measurement data and the target size.

[0039] Vibration error component, network delay error component, linearity deviation error component, and other error components are the different sources of error components obtained after decomposing the total deviation.

[0040] Corrected measurement refers to a measurement result that is closer to the true value after correcting the original vernier caliper measurement data by taking into account and deducting the main error components.

[0041] Product design tolerance range refers to the range of dimensional variations allowed in the product design specifications.

[0042] This method is typically deployed in the quality control system of an industrial production line. The system consists of data acquisition equipment (such as vernier calipers, vibration sensors, network monitoring equipment, etc.), a data processing server, and a quality assessment and judgment software module.

[0043] In summary, the product quality assessment method based on vernier caliper measurement big data proposed in this application significantly improves the accuracy and reliability of product quality assessment by systematically collecting multi-source data and refining the identification, quantification, and decomposition of various errors that may be introduced during the measurement process.

[0044] In another embodiment of this application, S5000 is further proposed to include: S5100: Calculates vibration intensity index from environmental vibration data within a preset time window based on the actual measurement timestamp, and sets a preset vibration intensity threshold. When the vibration intensity index exceeds the preset vibration intensity threshold, it dynamically adjusts the decomposition priority of error components and prioritizes the estimation of vibration error components. S5110: Based on the characteristics of network delay error, extract the network delay error component from the total deviation; S5120: Perform decomposition and iterative adjustment, including: estimating the vibration error component based on the vibration phase and vibration intensity index corresponding to the actual measurement timestamp, and combining the statistical relationship between the vibration error component and the total deviation under similar vibration phase and vibration intensity conditions in historical measurement data; after separating the estimated vibration error component and network delay error component from the total deviation, calculating the correlation between the remaining deviation and the linearity deviation error feature, until the absolute value of the correlation coefficient between the vibration error component and the remaining deviation relative to the linearity deviation error feature is lower than the preset low correlation threshold, and the remaining deviation is the total deviation minus the vibration error component and the network delay error component; S5130: Obtain the initial estimate of the linearity deviation error component from the linearity deviation model based on the environmental and equipment operating parameters; S5140: Adjust the initial estimate of the linearity deviation error component based on the residual deviation to obtain the updated linearity deviation error component; S5150: Performs randomness checks on other error components after stripping vibration error components, linearity deviation error components, and network delay error components. When a preset non-random pattern is detected, it triggers an anomaly warning or model update.

[0045] First, the vibration intensity index is calculated from environmental vibration data within a preset time window based on the actual measurement timestamp. Fourier transform / wavelet analysis can be used to extract specific frequency band energy or amplitude to quantify instantaneous vibration. A preset vibration intensity threshold is set; when the vibration intensity index exceeds the threshold, the decomposition priority of the error components is dynamically adjusted, prioritizing the estimation of vibration error components.

[0046] The removal of network delay error components is based on network delay error characteristics (obtained from timestamp calibration and actual network transmission delay estimation, reflecting systematic / random delays); for example, a mapping of "network delay error characteristics - network delay error components" is established based on the impact model of timestamp drift on the timing of the measurement process, separating the error introduced by network transmission instability from the total deviation.

[0047] A decomposition and iterative adjustment mechanism is introduced: the vibration error components are estimated by using the vibration phase and vibration intensity index corresponding to the actual measurement timestamp and combining the statistical relationship of similar historical working conditions (for example, by establishing a machine learning or statistical regression model to predict the magnitude of vibration error); after separating the estimated vibration error components and network delay error components from the total deviation, the correlation between the remaining deviation and the linearity deviation error characteristics is calculated and iterated until the absolute value of the correlation coefficient is lower than the preset low correlation threshold, so as to reduce the mutual influence of error components and improve the decomposition accuracy.

[0048] Linearity deviation error component: First, the initial estimate is given by the linearity deviation model (based on the golden sample and environmental and equipment operating parameters). Then, the remaining deviation generated by the iteration is adjusted to obtain the updated linearity deviation error component, so as to adapt to small changes in operating conditions and factors not fully covered by the model.

[0049] Randomness tests (such as operational tests and autocorrelation analysis) are performed on the remaining error components after the three types of errors are removed. If a preset non-random pattern (trend, periodicity, or abnormal peak) appears, an anomaly warning or model update is triggered, indicating that there may be a new error source or the existing model has failed, so as to facilitate timely intervention or adaptive adjustment.

[0050] In some preferred embodiments: In a vernier caliper measurement, the system calculated a total deviation of 0.1 mm from the target size. First, based on network delay error characteristics, a 0.01 mm network delay error component was extracted from the 0.1 mm total deviation. Then, based on the actual measurement timestamp, environmental vibration data was analyzed within a preset 1-second time window, yielding a vibration intensity index of 0.8 G, which is higher than the preset vibration intensity threshold of 0.5 G. Therefore, the decomposition priority of the error components was dynamically adjusted, prioritizing the estimation of the vibration error component. The decomposition iterative adjustment process began: In the first round, combining the current vibration phase and intensity, and utilizing a statistical model of historical similar working conditions, a vibration error component of 0.04 mm was estimated. At this point, the remaining deviation was 0.05 mm, and the absolute value of its correlation coefficient with the linearity deviation error characteristic was still higher than the preset low correlation threshold, so iteration continued. In the second round, based on environmental and equipment operating parameters, an initial estimate of 0.03 mm for the linearity deviation error component was obtained from the linearity deviation model, and adjusted to 0.035 mm based on the remaining deviation of 0.05 mm. The new remaining deviation was 0.015 mm, and the correlation was lower than the threshold, so the iteration terminated. Finally, after separating the vibration error component (0.04 mm), network delay error component (0.01 mm), and linearity deviation error component (0.035 mm), the remaining error component (0.015 mm) is obtained. A randomness test is then performed on it: if it exhibits random noise, the decomposition is complete; if it shows a periodic non-random pattern, an anomaly warning is triggered, and model updates or further investigation are recommended.

[0051] In another embodiment of this application, S5000 further includes: S5200: Continuously monitors auxiliary sensor data, including ambient temperature, humidity, air pressure, power supply voltage fluctuations, wireless channel interference intensity, heavy equipment operating power, tool wear status, and operator operation logs. S5210: Perform statistical analysis on other error components and obtain statistical characteristics, and compare the statistical characteristics with the preset random noise statistical characteristics. When the comparison result shows a significant difference and a non-random pattern, construct a multi-dimensional environment-device state vector based on the auxiliary sensor data at the current measurement time. S5220: Matches the multi-dimensional environment-device state vector with the historical state vectors in the preset unknown error pattern library to identify whether there are similar combination patterns; S5230: When a similar combination pattern is matched, the feature signature of the non-random pattern is compared with the pattern signature recorded in the unknown error pattern library. If the comparison result shows that there is similarity, the non-random pattern is classified as a known unknown error source, and the potential system error source and suggested investigation direction are output based on the source tracing information recorded in the unknown error pattern library. S5240: When no similar combination pattern is matched, the non-random pattern and its corresponding multi-dimensional environment-device state vector are recorded as new unknown error patterns and the unknown error pattern library is updated. S5250: Based on the characteristics of the identified unknown error sources, dynamically adjust the control parameters and update the linearity deviation model, the historical baseline of the network delay error characteristics, and the functional relationship used to generate vibration error characteristics. The control parameters include at least a preset time window, a preset vibration intensity threshold, a decomposition priority, and a preset low correlation threshold.

[0052] Specifically, the system continuously monitors auxiliary sensor data: it continuously collects environmental factors such as ambient temperature, humidity, and air pressure, as well as equipment and operating parameters such as power supply voltage fluctuations, wireless channel interference intensity, heavy equipment operating power, tool wear status, and operator operation logs, in order to identify unknown error sources (e.g., changes in ambient temperature affect the physical characteristics of measuring equipment; power supply voltage fluctuations cause instability in the performance of electronic components; tool wear status is related to machining accuracy).

[0053] Statistical analysis and discrimination of other error components: After removing the vibration error component, network delay error component, and linearity deviation error component, the mean, variance, autocorrelation, power spectral density and other statistical characteristics of the remaining error components are calculated and compared with the preset random noise statistical characteristics; when significant differences occur and the pattern is non-random, a multi-dimensional environment-equipment state vector is constructed based on the current auxiliary sensor data.

[0054] Pattern matching and source tracing: Match the multi-dimensional environment-equipment state vector with historical state vectors in the unknown error pattern library (e.g., using Euclidean distance, cosine similarity, or machine learning classifiers), and compare the non-random pattern feature signatures; if similar, classify it as a known unknown error source and an unknown error source (archived type), and output potential systematic error sources (e.g., material problems in a specific batch, aging of a certain equipment component) and suggested investigation directions based on the source tracing information in the library; if no match is found, register the non-random pattern and its vector as a new unknown error pattern and update the unknown error pattern library.

[0055] Adaptive parameters and model updates: Based on the characteristics of the identified unknown error sources, the system dynamically adjusts control parameters such as preset time windows, preset vibration intensity thresholds, decomposition priorities, and preset low correlation thresholds; and updates the historical baselines of linearity deviation from the model, network delay error characteristics, and the functional relationships used to generate vibration error characteristics to improve the accuracy of the overall assessment.

[0056] In some preferred embodiments: During the measurement of a batch of products, the system detects that other error components, after stripping the vibration error component, network delay error component, and linearity deviation error component, exhibit a significant non-random pattern. Therefore, auxiliary sensor data monitoring is initiated, collecting and recording ambient temperature (25℃), humidity (60%), air pressure (101kPa), power supply voltage fluctuation (0.5V), wireless channel interference intensity (-70dBm), heavy equipment operating power (50kW), tool wear status (moderate wear), and operator operation logs to construct a multi-dimensional environment-equipment state vector. This vector is input for unknown error pattern matching: if it highly matches the historical state vector and its non-random pattern signature, the current non-random pattern is classified as a periodic error caused by equipment resonance. The potential system error source is output as resonance between the heavy equipment and the measuring equipment, suggesting the investigation direction be checking the installation foundation of the heavy equipment and the measuring equipment or adjusting the operating frequency of the heavy equipment. Based on this, control parameters are dynamically adjusted: the preset time window used for vibration error component estimation is shortened from 5 seconds to 2 seconds, the preset vibration intensity threshold is appropriately reduced, and the functional relationship used to generate vibration error characteristics is updated to adapt to this type of resonance pattern and improve subsequent measurement accuracy. If no similar combination pattern is found in the unknown error pattern library, the current non-random pattern and its multi-dimensional environment-device state vector are registered as new unknown error patterns and the unknown error pattern library is updated.

[0057] In another embodiment of this application, after comparing the corrected measurement value with the product design tolerance range to output a quality acceptance result, the method further includes: S6100: Determine whether the quality pass determination result meets the preset persistent deviation determination conditions. The persistent deviation determination conditions are set based on the persistent deviation index. The persistent deviation index includes at least the rolling window misjudgment rate calculated based on the true value of the gold sample. S6110: When the quality pass determination result meets the preset persistent deviation determination condition, the quality pass determination result is statistically analyzed based on the preset sliding time window to identify the type and trend of persistent deviation. S6120: Based on the type and trend of persistent deviation, determine the estimation model of the error component that needs to be checked first. The estimation model of the error component includes the vibration error component estimation model and the linearity deviation error component estimation model. S6130: Based on the corresponding environmental vibration data and corrected measurements, re-evaluate the functional relationship used to generate vibration error characteristics; S6140: Based on the measurement results of the corresponding gold sample and the environmental and equipment operating parameters, re-evaluate the fitting accuracy of the measurement equipment under the linearity deviation model, and determine the model parameters that need to be adjusted. S6150: Based on the re-evaluation results of the functional relationship used to generate vibration error characteristics and the re-evaluation results of the linearity deviation model, perform feedback iterative adjustment to adjust the parameters of the vibration error component estimation model and the linearity deviation error component estimation model. S6160: Use the parameters of the adjusted vibration error component estimation model and linearity deviation error component estimation model to perform back-calculation on historical data, regenerate the vibration error component and linearity deviation error component, update and correct the measured values, and re-output the quality qualification judgment result. S6170: When the persistent deviation index after backtracking calculation is lower than the preset deviation threshold, the maximum number of feedback iterations is reached, or the change in model parameters is lower than the preset convergence threshold, the parameters of the vibration error component estimation model and the linearity deviation error component estimation model are applied to the subsequent real-time evaluation; if the termination condition is not met, the feedback iteration adjustment continues.

[0058] The quality acceptance result refers to the conclusion of acceptance or non-acceptance obtained by comparing the corrected vernier caliper measurement data with the product design tolerance range. To capture non-random and trend-oriented or periodic deviations, a persistent deviation judgment condition is introduced: when more than a preset number of misjudgments occur consecutively within a preset time window, or when the rolling window misjudgment rate is consistently higher than a threshold (the rolling window misjudgment rate is the proportion of measurements that have been incorrectly judged as acceptable or unacceptable after comparison with the true value of the gold sample), a persistent deviation is considered to exist.

[0059] Upon detecting persistent deviations, the system statistically analyzes historical quality compliance results based on a preset sliding time window, identifying deviation types (e.g., systematically high / low) and trends (e.g., gradual deterioration / periodic occurrence). Based on this, it determines the priority error component estimation models to be checked: when deviations are generally high or low, the linearity deviation error component estimation model is prioritized; when highly correlated with environmental vibration events, the vibration error component estimation model is prioritized. Subsequently, using the latest environmental vibration data, corrected measurements, measurement results from the gold standard sample, and environmental and equipment operating parameters, the corresponding models are re-evaluated: for the vibration error component estimation model, the functional relationship used to calculate the instantaneous deviation caused by vibration is re-evaluated; for the linearity deviation error component estimation model, the fitting accuracy under the linearity deviation model is re-evaluated, and the model parameters that need adjustment (including slope, intercept, or higher-order term coefficients) are determined.

[0060] After the reassessment is completed, feedback iterative adjustments are performed: In each iteration, updated parameters are used to backcalculate historical data, regenerate vibration error components and linearity deviation error components, and update and correct measurement values ​​until the termination condition is met (the persistent deviation indicator after backtracking, such as the rolling window misjudgment rate, is lower than the preset deviation threshold, or the maximum number of feedback iterations is reached, or the change in model parameters is lower than the preset convergence threshold). After the termination condition is met, the adjusted model parameters are used for subsequent real-time quality assessments to improve the accuracy and reliability of the assessment.

[0061] In another embodiment of this application, a step is further proposed to compare the corrected measurement value with the product design tolerance range to output a quality acceptance judgment result, specifically including: S6200: Obtain the distance between the corrected measurement value and the nearest limit of the product design tolerance range, where the nearest limit of the product design tolerance range is the limit that is closer to the corrected measurement value among the upper and lower limits of the product design tolerance range. S6210: Obtain current environmental parameters and equipment operating parameters; S6220: When the distance is less than the preset critical distance, the enhanced judgment mode is activated. S6230: Performs repeated measurements a preset number of times within a preset short time window, continuously collecting and correcting measurement values ​​and their corresponding environmental and equipment operating parameters; S6240: Based on the environmental parameters and equipment operating parameters corresponding to each measurement, and combined with the historical statistical relationship under similar environmental and equipment conditions, calculate the fluctuation range of the corresponding corrected measurement value; S6250: When the corrected measurement values ​​and their fluctuation range are all within the product design tolerance range for multiple consecutive measurements within the preset number of measurements, the product is deemed qualified; when the corrected measurement values ​​and their fluctuation range are all outside the product design tolerance range for multiple consecutive measurements within the preset number of measurements, the product is deemed unqualified. S6260: When the corrected measured value and its fluctuation range cross the product design tolerance range, calculate the instantaneous rate of change of environmental parameters and equipment operating parameters; S6270: When the instantaneous rate of change exceeds the preset instantaneous rate of change threshold, extend the preset short time window and / or increase the preset number of measurements, and make a judgment based on the updated data according to the enhanced judgment mode.

[0062] Specifically, the corrected measurement value refers to the value obtained after correcting the vernier caliper measurement data based on the vibration error component and the linearity deviation error component; the product design tolerance range refers to the upper and lower limits of the dimensions allowed by the design specifications; the nearest limit refers to the boundary closer to the corrected measurement value among the upper and lower limits of the product design tolerance range; the distance refers to the absolute difference between the corrected measurement value and the nearest limit; environmental parameters include ambient temperature, humidity, air pressure, etc., and equipment operating parameters include power supply voltage fluctuations, equipment load, sensor output effective range, etc., used to characterize the external conditions and equipment status during measurement; the preset critical distance is used to determine whether to enter a sensitive area; the enhanced judgment mode is used to make more refined judgments within sensitive areas; the preset short time window is the time period for repeated measurements, and the preset number of measurements is the number of measurements within this time window; the fluctuation range is the possible range of variation of the corrected measurement value under given environmental parameters and equipment operating parameters (which can be calculated based on historical statistics or real-time analysis); the instantaneous change rate is the rate of change of environmental parameters or equipment operating parameters in a short period of time, and the preset instantaneous change rate threshold is used to determine whether the judgment strategy needs to be adjusted.

[0063] When the distance is less than the preset critical distance, the enhanced judgment mode is activated: repeated measurements are performed a preset number of times within a preset short time window, continuously collecting each corrected measurement value and its corresponding environmental parameters and equipment operating parameters, and calculating the fluctuation range of each corrected measurement value based on historical statistical relationships; if the corrected measurement values ​​and their fluctuation ranges are all within the product design tolerance range within the preset number of measurements, the result is deemed qualified; if they are all outside the product design tolerance range, the result is deemed unqualified; when the corrected measurement values ​​and their fluctuation ranges cross the product design tolerance range, the instantaneous change rate of the environmental parameters and equipment operating parameters is calculated. If the instantaneous change rate exceeds the preset instantaneous change rate threshold, the preset short time window is extended and / or the preset number of measurements is increased, and the enhanced judgment mode is continued to make judgments based on the newly added data.

[0064] The above mechanism can significantly improve the accuracy and reliability of quality assessment when correcting measured values ​​that are close to the product design tolerance range.

[0065] In some preferred embodiments, assuming the target size of a part is 10.00 mm, the product design tolerance range is ±0.05 mm (acceptable range 9.95 mm – 10.05 mm), and the preset critical distance is 0.01 mm, when the corrected measurement value is 9.945 mm, the distance from the nearest limit of 9.95 mm is 0.005 mm (less than 0.01 mm), and the system activates an enhanced judgment mode: performing a preset number of measurements (e.g., 5 times) within a preset short time window (e.g., 30 seconds). The first measurement: corrected measurement value 9.945 mm, ambient temperature 25.1℃, equipment load 70%, calculated fluctuation range ±0.003 mm (9.942 mm – 9.948 mm), still outside the 9.95 mm – 10.05 mm range, and subsequent measurements continue. If five consecutive corrected measurements and their fluctuation ranges remain stable at around 9.945 mm and are always outside the tolerance range, the measurement is deemed unqualified. If a corrected measurement is 9.948 mm with a fluctuation range of ±0.003 mm (9.945 mm – 9.951 mm), exceeding the lower tolerance limit of 9.95 mm, the instantaneous change rate of environmental parameters and equipment operating parameters is immediately calculated. If the ambient temperature is detected to rise by 0.5℃ in a short period of time and the instantaneous change rate exceeds the preset instantaneous change rate threshold, the preset short time window is extended to 60 seconds and / or the preset number of measurements is increased to 10. The judgment is then completed using the enhanced judgment mode based on the new data to ensure accuracy and robustness under complex or unstable operating conditions.

[0066] In another embodiment of this application, after comparing the corrected measurement value with the product design tolerance range to output a quality acceptance result, the method further includes: S6300: Continuously monitors distance and records the historical sequence of distance and its rate of change within a preset sliding time window; S6310: When the distance is determined based on the historical sequence to have fluctuations in the neighborhood of the preset critical distance that are greater than the preset duration, the value of the preset critical distance is dynamically adjusted according to the fluctuation frequency and amplitude of the distance. S6320: When the start or exit conditions of the enhanced judgment mode are met, wait for the preset mode switching delay time, and only execute the corresponding start or exit if the start or exit conditions are still met after the preset mode switching delay time has expired. S6330: Set the priority for starting the enhanced judgment mode based on the importance of the current production task.

[0067] Specifically, the system continuously monitors distances, acquiring in real time the distance between the corrected measurement value and the nearest limit of the product design tolerance range, and storing and analyzing it as a time series. By using preset sliding time windows (such as 1 minute, 5 minutes, etc.), the system statistically analyzes the historical sequence of distances and their rate of change (which can be calculated using derivatives or differences) for short / medium-term trend assessment.

[0068] Specifically, when fluctuations exceeding a preset duration occur within the neighborhood of a preset critical distance based on historical sequence determination (e.g., repeated boundary crossings within the neighborhood for more than 10 consecutive seconds), the preset critical distance is dynamically adjusted according to the fluctuation frequency and amplitude: when the fluctuation frequency is high and the amplitude is large, the preset critical distance is appropriately increased to initiate the enhanced determination mode earlier; when the fluctuation is stable, the distance is appropriately reduced to decrease unnecessary enhanced determinations.

[0069] In practical applications, when the start-up condition (distance less than the preset critical distance) or exit condition (distance greater than the preset critical distance) of the enhanced judgment mode is met, the system first waits for the preset mode switching delay time (such as 2 seconds, 5 seconds, or 10 seconds). The system only executes the start-up or exit if the verification condition is still met after the delay expires, in order to avoid frequent switching caused by instantaneous fluctuations.

[0070] In addition, the activation priority of the enhanced judgment mode is set according to the importance of the current production task: critical components can be activated when there is a slight overstep, while ordinary components can be activated more leniently or with a delayed start, in order to match the importance of the business and the resources invested.

[0071] In some preferred embodiments: A production line measures high-precision aero-engine blades. The system continuously monitors the distance between the corrected measurement value and the nearest limit of the product design tolerance range. Due to the start-up of heavy equipment in the workshop, environmental vibration increases slightly, causing this distance to fluctuate continuously for 15 seconds within a preset critical distance of 0.01mm with an amplitude of 0.5Hz and 0.002mm (exceeding the preset duration of 10 seconds). Based on this, the system adjusts the preset critical distance from 0.01mm to 0.012mm to activate the enhanced judgment mode earlier and strengthen monitoring. Subsequently, when the measured distance is first less than 0.012mm, the system does not switch immediately but waits for a preset mode switching delay of 5 seconds. If the distance remains less than 0.012mm within 5 seconds, the enhanced judgment mode is officially activated after the delay expires, and repeated measurements and fluctuation range analysis are performed. If the distance briefly returns to above 0.012mm during this period, the switching is canceled to avoid mode oscillation. Given that aero-engine blade production is a highly important task, the system sets the highest priority for the enhanced decision mode activation, and will activate it even if system resources are slightly tight. For ordinary bolts (which have a lower priority) that are produced at the same time, the enhanced decision mode activation can be delayed or the regular decision mode can be maintained when resources are tight, so as to optimize the overall resource allocation.

[0072] In another embodiment of this application, S6330 further includes: S6331: Identify the type of part to be tested and its corresponding production task; S6332: Obtain the corresponding quality judgment priority from the preset rule base according to the part type and production task, and set the start setting priority of the enhanced judgment mode accordingly. S6333: Monitor system resource status, which includes at least computing resource usage and measurement device load; S6334: When the distance is less than a preset critical distance, the decision to activate the enhanced judgment mode is made based on the quality judgment priority and system resource status: when the quality judgment priority is high and the system resource status meets the preset resource threshold, the enhanced judgment mode is activated immediately; when the quality judgment priority is high but the system resource status does not meet the preset resource threshold, the enhanced judgment mode of the part type with low activation priority is paused or switched to the normal judgment mode, and the enhanced judgment mode of the part with high activation priority is activated; when the quality judgment priority is low and the system resource status meets the preset resource threshold, the enhanced judgment mode is activated; when the quality judgment priority is low and the system resource status does not meet the preset resource threshold, the activation of the enhanced judgment mode is delayed or the normal judgment mode is maintained. S6335: After the enhanced judgment mode is activated, the preset number of measurements and the sampling frequency within the preset short time window are dynamically adjusted according to the quality judgment priority and system resource status.

[0073] Specifically, the system identifies the type of part to be tested and its corresponding production task: the system automatically or manually inputs information such as part model and batch, and associates it with its production task (such as final inspection of key components and sampling inspection of ordinary components).

[0074] Specifically, the system retrieves the corresponding quality judgment priority from a preset rule base based on the part type and production task, and sets the activation priority of the enhanced judgment mode accordingly. The preset rule base maintains a mapping of "part type / production task ↔ quality judgment priority"; for example, critical aerospace structural components have the highest priority, while non-critical auxiliary components have a lower priority, and this priority directly affects the activation decision.

[0075] In practical applications, the monitoring system resource status (including at least computing resource usage and measurement equipment load) is used to obtain real-time indicators such as CPU, memory, storage I / O, as well as the workload, standby status, and maintenance requirements of measuring equipment such as vernier calipers, in order to assess whether the enhanced judgment mode can be started or maintained.

[0076] Furthermore, when the distance is less than a preset critical distance, an enhanced judgment mode activation decision is made based on the quality judgment priority and system resource status: If the quality assessment has a high priority and the resources are ≥ the preset resource threshold, the enhanced assessment mode will be activated immediately. If the quality judgment priority is high and the resources are less than the preset resource threshold, the enhanced judgment mode of the low-start-priority component is paused or switched to the normal judgment mode to ensure the start-up of the high-start-priority component. If the quality assessment has a low priority and the resources are greater than or equal to the preset resource threshold, the enhanced assessment mode will be activated. If the quality assessment has a low priority and the resources are less than the preset resource threshold, the startup will be delayed or the normal assessment mode will be maintained to avoid excessive resource consumption.

[0077] In addition, after the enhanced judgment mode is activated, the preset number of measurements and the sampling frequency within the preset short time window are dynamically adjusted according to the quality judgment priority and system resource status: when the priority is high, the number of measurements or the sampling frequency can be appropriately increased even if resources fluctuate; when the priority is low and resources are scarce, the frequency can be appropriately reduced or lowered to balance accuracy and resources.

[0078] The proposed solution avoids rigid priority settings and unreasonable resource allocation by introducing dynamic considerations of the importance of production tasks and the status of system resources, thereby enabling the on-demand activation and refined operation of the enhanced judgment mode.

[0079] In some preferred embodiments: A manufacturing company produces both Type A (critical aerospace components) and Type B (ordinary fasteners). In the preset rule base, Type A has a high priority for quality assessment, while Type B has a low priority. The preset resource thresholds are CPU utilization < 70% and measurement equipment load < 80%. When a Type A part is detected to be within a preset critical distance, the system identifies the part under test as Type A and assigns it high priority. At this point, CPU utilization is 60% and measurement equipment load is 70%, both meeting the thresholds. Therefore, the enhanced judgment mode is immediately activated, and due to the high priority of Type A, the preset number of measurements is adjusted to 10, and the sampling frequency within the preset short time window is increased to 5 times per second to ensure high-precision, high-density measurement.

[0080] In another scenario, the distance to the detected type B part is also less than the preset critical distance, and the system identifies type B as low priority. At this time, the CPU utilization rate is 85% and the measurement device load is 90%, neither of which meets the threshold. The system delays the start of the enhanced judgment mode or maintains the normal judgment mode to avoid unnecessary resource consumption of low priority tasks when resources are scarce.

[0081] In another embodiment of this application, it is further proposed that after the enhanced determination mode is activated, it also includes: A1: Continuously monitor the data stream status of the measuring equipment and auxiliary sensors. The data stream status includes at least data packet integrity, transmission rate, and effective range of sensor output. A2: When any interruption or abnormality is detected in the status of any data stream, a fault warning is triggered, and the data acquisition and judgment process in the enhanced judgment mode is suspended. A3: During the pause, record the environmental parameters and equipment operating parameters at the time of the fault, and store the valid measurement data collected before the fault occurred; A4: Continuously monitor the recovery status of faulty measuring equipment or auxiliary sensors; A5: After the faulty measuring equipment or auxiliary sensor returns to normal, determine whether it is necessary to perform retrospective processing on the suspended judgment task based on the recorded environmental parameters and equipment operating parameters at the time of the fault occurrence. A6: When retrospective processing is required, the measurement data for the corresponding time period is re-collected after the fault is recovered, and merged with the stored valid measurement data to restart the judgment process in the enhanced judgment mode. A7: When it is determined that no retrospective processing is required, the data acquisition and judgment process in the enhanced judgment mode will be smoothly restored after the fault is recovered.

[0082] Specifically, data stream status refers to the health status of data transmitted from measuring devices (e.g., vernier calipers) and auxiliary sensors (e.g., ambient temperature sensors, vibration sensors, etc.) to the data processing system. Data packet integrity is used to determine if transmission has been lost or damaged (checksums, serial numbers, etc., can be added to the data transmission protocol for monitoring); transmission rate is used to determine if the speed meets expectations (the amount of data received per unit time can be calculated in real time); and the effective range of sensor output is used to determine if the reading is within the normal operating range (e.g., a temperature sensor should not output -200°C or 1000°C; this can be determined based on the sensor's range or historical statistical range).

[0083] When any data stream status interruption or abnormality is detected, the system immediately triggers a fault warning (red warning on the control interface, recording of system logs, sending SMS or email to the operator), and suspends the data acquisition and judgment process in the enhanced judgment mode (stops receiving new measurement data and suspends the current quality judgment calculation).

[0084] During the pause, the system automatically records environmental parameters (ambient temperature, humidity, air pressure, etc.) and equipment operating parameters (power supply voltage fluctuations, equipment load, internal temperature, etc.) at the time of the fault, and saves the valid measurement data collected before the fault occurred for subsequent retrospective processing.

[0085] The system continuously monitors the recovery status of the measuring equipment or auxiliary sensors that caused the pause (whether the data stream status is stable, whether the sensor output has returned to the effective range of the sensor output, etc.) to determine whether the fault has been eliminated.

[0086] Once the faulty equipment or sensor returns to normal, the system determines whether retrospective processing is required based on the recorded environmental parameters and equipment operating parameters: if the fault occurred during a period of drastic fluctuations in environmental parameters or severe abnormalities in equipment operating parameters, retrospective processing is performed; otherwise, direct recovery is possible.

[0087] If retrospective processing is required, the system will re-collect measurement data for the corresponding time period after recovery (e.g., through re-measurement or self-backup), merge it with the stored valid measurement data, and then restart the judgment process in the enhanced judgment mode. If retrospective processing is not required, the data acquisition and judgment process in the enhanced judgment mode will be smoothly restored to maintain continuity and improve efficiency.

[0088] This application solves the problem of evaluation interruption or unreliable results caused by data flow anomalies in the enhanced judgment mode by continuously monitoring the data flow status, providing immediate warnings and pausing when anomalies occur, and performing retrospective processing or rapid recovery as needed after recovery.

[0089] In another embodiment of this application, when the enhanced determination mode is enabled and the data stream is normal, before comparing the corrected measurement value with the product design tolerance range, this application further includes the following steps: B1: Based on vibration error characteristics, network delay error characteristics, and linearity deviation error characteristics, calculate the corresponding confidence index and form an error feature confidence vector; B2: Generate confidence gating coefficients based on the error feature confidence vector, which are used to dynamically adjust the control parameters of the decomposition iteration, including adaptive adjustment of the preset low correlation threshold, decomposition step size and decomposition priority; B3: When the confidence index of any error feature is lower than the preset confidence threshold, phase stabilization resampling is triggered within a preset short time window, and additional measurements are performed in the phase stabilization interval selected based on the actual measurement timestamp and vibration phase, and the historical baseline of the network delay error feature is updated synchronously; when the phase stabilization resampling and the updated historical baseline of the network delay error feature still indicate high uncertainty, fast verification point measurement is performed on the gold sample to constrain the local extrapolation distance of the linearity deviation from the model; B4: Based on the confidence gating coefficient, the vibration error component and the linearity deviation error component are correspondingly contracted or amplified to obtain the confidence gating correction measurement value, and the confidence gating correction measurement value is used to replace the correction measurement value for subsequent quality acceptance judgment.

[0090] Specifically, the confidence index is used to quantify the accuracy of various error feature estimates. It can be calculated from model residuals, historical volatility, data source reliability, and correlation with auxiliary sensor data. The various indicators are integrated to form an error feature confidence vector, which reflects the reliability of the current error feature estimates.

[0091] Among them, the confidence gating coefficient calculated based on the confidence vector of error features is used as a dynamic adjustment factor to adaptively adjust key control parameters during the error component decomposition process: when the confidence of a certain error feature is low, the preset low correlation threshold can be adjusted accordingly to more carefully strip the error component, the decomposition step size can be adjusted to refine the iteration, or the decomposition priority can be adjusted to prioritize the processing of error components with higher confidence.

[0092] A pre-set confidence threshold is set to determine the reliability of the error feature estimation; values ​​below this threshold indicate high uncertainty, requiring additional measures. Specifically, phase stabilization resampling (adding measurements within a short period of time in an interval where the vibration phase is relatively stable) is triggered to improve the accuracy of vibration error feature estimation, and the historical baseline of network delay error features is updated simultaneously to reflect the latest network conditions and reduce uncertainty; if this is still insufficient, rapid verification point measurements are performed on the golden sample to constrain the local extrapolation distance of linearity deviation from the model, avoiding inaccurate extrapolation under high uncertainty conditions.

[0093] Furthermore, the confidence gating coefficient directly adjusts the correction strength of the error component: a high confidence level amplifies the corresponding correction amount to fully eliminate the error; a low confidence level reduces the correction amount to avoid over-correction or introducing new errors. Through this mechanism, uncertainties can be identified and quantified in the enhanced judgment mode, adaptively adjusting decomposition and correction. Especially in critical situations approaching the product design tolerance range, combined with phase-stabilized resampling and rapid verification point measurement using gold samples, the uncertainty of key error characteristics is effectively reduced, improving the accuracy and reliability of judgments and reducing the risk of misjudgment or missed judgment.

[0094] In some preferred embodiments, assuming that in the enhanced judgment mode, the system repeatedly measures a key dimension: at a certain measurement moment, the system calculates the confidence index of the current vibration error feature as 0.6, the network delay error feature as 0.8, and the linearity deviation error feature as 0.7; the preset confidence threshold is 0.75. Since the confidence indices of vibration and linearity deviation (0.6, 0.7) are lower than 0.75, the system immediately triggers phase stabilization resampling; within the next preset short time window, additional measurements are performed in the phase stabilization interval based on the actual measurement timestamp and vibration phase, and the vibration error feature is re-estimated using the new data, while the historical baseline of the network delay error feature is updated. Assuming that after resampling and baseline update, the vibration error increases to 0.72 and the linearity deviation increases to 0.73, still lower than 0.75; the system further performs rapid verification point measurements on the gold sample and compares them with the linearity deviation model to locally correct and constrain the model accuracy. Subsequently, based on the updated error feature confidence vector (vibration 0.72, network 0.8, linearity 0.73), the system generates corresponding confidence gating coefficients. Since the confidence levels for vibration and linearity are still relatively low, the gating coefficients moderately reduce the corresponding correction values: if the original vibration correction is X and the linearity correction is Y, the actual applied correction values ​​might be 0.9X and 0.95Y. This yields the confidence-gated correction measurement value, which replaces the original correction measurement value for comparison with the product design tolerance range, outputting a more reliable quality acceptance result.

[0095] Reference Figure 2 The specific embodiments of this application also disclose a product quality assessment system based on vernier caliper measurement big data, which includes: Data acquisition module 1 is used to acquire vernier caliper measurement data and its local timestamp, environmental vibration data, network transmission status data and measurement equipment calibration status data. The network transmission status data includes the signal strength and channel occupancy before transmission, access end load and channel interference, and server receiving timestamp. The measurement equipment calibration status data includes the measurement results of the gold sample and environmental and equipment operating parameters. The first feature extraction module 2 is used to perform timestamp calibration based on the local timestamp of the vernier caliper measurement data, the server receiving timestamp, and the network transmission status data, estimate the actual network transmission delay, obtain the real measurement timestamp, and determine the actual network transmission delay and its fluctuation relative to the historical baseline as network delay error features. The second feature extraction module 3 is used to time-align the vernier caliper measurement data with the environmental vibration data using the real measurement timestamp, and to perform time-frequency analysis to obtain the vibration phase at the moment of measurement, and to calculate the instantaneous deviation caused by vibration and determine it as the vibration error feature. The third feature extraction module 4 is used to establish a linearity deviation model of the measuring equipment under different measurement ranges and operating conditions based on the measurement results of the gold sample and combined with the environmental and equipment operating parameters, and to calculate the system deviation caused by the linearity deviation of the equipment and determine it as the linearity deviation error feature. Error component decomposition module 5 is used to calculate the total deviation between the vernier caliper measurement data and the target size, and decompose it into vibration error component corresponding to vibration error characteristics, network delay error component corresponding to network delay error characteristics, linearity deviation error component corresponding to linearity deviation error characteristics, and other error components. The judgment module 6 is used to correct the vernier caliper measurement data based on the vibration error component and the linearity deviation error component, obtain the corrected measurement value, and compare the corrected measurement value with the product design tolerance range to output the quality qualified judgment result.

[0096] The product quality assessment system based on vernier caliper measurement big data proposed in this application aims to systematically solve the challenges faced in product quality assessment in the field of high-precision parts manufacturing through modular design and collaborative work.

[0097] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A product quality assessment method based on big data measurement using vernier calipers, characterized in that, include: The system collects vernier caliper measurement data and its local timestamp, environmental vibration data, network transmission status data, and measurement equipment calibration status data. The network transmission status data includes the signal strength and channel occupancy before transmission, access end load and channel interference, and server receiving timestamp. The measurement equipment calibration status data includes the measurement results of the gold sample and environmental and equipment operating parameters. Based on the local timestamp, server receiving timestamp, and network transmission status data of the vernier caliper measurement data, timestamp calibration is performed to estimate the actual network transmission delay, obtain the true measurement timestamp, and the actual network transmission delay and its fluctuation relative to the historical baseline are determined as network delay error characteristics. The vernier caliper measurement data and the environmental vibration data are time-aligned according to the actual measurement timestamp, and time-frequency analysis is performed to obtain the vibration phase at the moment of measurement. The instantaneous deviation caused by vibration is calculated based on this and determined as the vibration error characteristic. Based on the measurement results of the gold sample and combined with the environmental and equipment operating parameters, a linearity deviation model of the measuring equipment under different measurement ranges and operating conditions is established, and the system deviation caused by the linearity deviation of the equipment is calculated accordingly, which is determined as the linearity deviation error characteristic. Calculate the total deviation between the vernier caliper measurement data and the target size, and decompose it into the vibration error component corresponding to the vibration error characteristic, the network delay error component corresponding to the network delay error characteristic, the linearity deviation error component corresponding to the linearity deviation error characteristic, and other error components; The measurement data of the vernier caliper is corrected based on the vibration error component and the linearity deviation error component to obtain the corrected measurement value. The corrected measurement value is then compared with the product design tolerance range to output a quality qualification result.

2. The product quality assessment method based on big data measurement using vernier calipers as described in claim 1, characterized in that, Calculate the total deviation between the vernier caliper measurement data and the target size, and decompose it into the vibration error component corresponding to the vibration error characteristic, the network delay error component corresponding to the network delay error characteristic, the linearity deviation error component corresponding to the linearity deviation error characteristic, and other error components, including: Based on the actual measurement timestamp, the vibration intensity index is calculated from the environmental vibration data within a preset time window, and a preset vibration intensity threshold is set. When the vibration intensity index exceeds the preset vibration intensity threshold, the decomposition priority of the error components is dynamically adjusted, and the vibration error components are estimated first. Based on the network latency error characteristics, the network latency error component is extracted from the total deviation; The decomposition and iterative adjustment includes: estimating the vibration error component based on the vibration phase corresponding to the actual measurement timestamp and the vibration intensity index, and combining the statistical relationship between the vibration error component and the total deviation under similar vibration phase and intensity conditions in historical measurement data; after separating the estimated vibration error component and the network delay error component from the total deviation, calculating the correlation between the remaining deviation and the linearity deviation error feature, until the absolute value of the correlation coefficient between the vibration error component and the remaining deviation relative to the linearity deviation error feature is lower than a preset low correlation threshold, wherein the remaining deviation is the total deviation minus the vibration error component and the network delay error component; The initial estimated value of the linearity deviation error component is obtained from the linearity deviation model based on the environmental and equipment operating parameters. The initial estimate of the linearity deviation error component is adjusted based on the remaining deviation to obtain an updated linearity deviation error component. Randomness checks are performed on the remaining error components after stripping the vibration error component, the linearity deviation error component, and the network delay error component. When a preset non-random pattern is detected, an anomaly warning or model update is triggered.

3. The product quality assessment method based on vernier caliper measurement big data as described in claim 2, characterized in that, The total deviation between the vernier caliper measurement data and the target size is calculated and decomposed into the vibration error component corresponding to the vibration error characteristic, the network delay error component corresponding to the network delay error characteristic, the linearity deviation error component corresponding to the linearity deviation error characteristic, and other error components, including: Continuously monitor auxiliary sensor data, including ambient temperature, humidity, air pressure, power supply voltage fluctuations, wireless channel interference intensity, heavy equipment operating power, tool wear status, and operator operation logs; Statistical analysis is performed on the other error components to obtain statistical characteristics, and the statistical characteristics are compared with preset random noise statistical characteristics. When the comparison result shows a significant difference and a non-random pattern, a multi-dimensional environment-device state vector is constructed based on the auxiliary sensor data at the current measurement time. The multi-dimensional environment-device state vector is matched with historical state vectors in a preset unknown error pattern library to identify whether there are similar combination patterns. When a similar combination pattern is matched, the feature signature of the non-random pattern is compared with the pattern signature recorded in the unknown error pattern library. If the comparison result shows that there is similarity, the non-random pattern is classified as a known unknown error source, and the potential system error source and suggested investigation direction are output based on the source tracing information recorded in the unknown error pattern library. When no similar combination pattern is matched, the non-random pattern and its corresponding multi-dimensional environment-device state vector are recorded as new unknown error patterns, and the unknown error pattern library is updated. Based on the characteristics of the identified unknown error sources, the control parameters are dynamically adjusted, and the linearity deviation model, the historical baseline of the network delay error characteristics, and the functional relationship used to generate the vibration error characteristics are updated. The control parameters include at least the preset time window, the preset vibration intensity threshold, the decomposition priority, and the preset low correlation threshold.

4. The product quality assessment method based on vernier caliper measurement big data as described in claim 1, characterized in that, After comparing the corrected measurement value with the product design tolerance range to output the quality acceptance result, the method further includes: Determine whether the quality pass determination result meets the preset persistent deviation determination condition. The persistent deviation determination condition is set based on the persistent deviation index. The persistent deviation index includes at least the rolling window misjudgment rate calculated based on the true value of the gold sample. When the quality pass determination result meets the preset persistent deviation determination condition, the quality pass determination result is statistically analyzed based on the preset sliding time window to identify the type and trend of persistent deviation; Based on the type and trend of the persistent deviation, the estimation models of the error components that need to be checked first are determined. The estimation models of the error components include the vibration error component estimation model and the linearity deviation error component estimation model. Based on the corresponding environmental vibration data and corrected measurements, the functional relationship used to calculate the instantaneous deviation caused by vibration based on the vibration phase and vibration intensity aligned with the actual measurement timestamp is re-evaluated to generate the vibration error characteristics. Based on the measurement results of the corresponding gold samples and the environmental and equipment operating parameters, the fitting accuracy of the measuring equipment under the linearity deviation model is re-evaluated, and the model parameters that need to be adjusted are determined. Based on the functional relationship used to calculate the instantaneous deviation caused by vibration based on the vibration phase and vibration intensity aligned with the actual measurement timestamp to generate the vibration error characteristics, the re-evaluation results of the functional relationship used to generate the vibration error characteristics and the re-evaluation results of the linearity deviation model are used to perform feedback iterative adjustment to adjust the parameters of the vibration error component estimation model and the linearity deviation error component estimation model. The historical data is back-calculated using the parameters of the adjusted vibration error component estimation model and the linearity deviation error component estimation model. The vibration error component and the linearity deviation error component are regenerated, the measured values ​​are updated and corrected, and the quality qualification judgment result is re-output. When the persistent deviation index after backtracking calculation is lower than the preset deviation threshold, the maximum number of feedback iterations is reached, or the change in model parameters is lower than the preset convergence threshold, the parameters of the vibration error component estimation model and the linearity deviation error component estimation model are applied to the subsequent real-time evaluation; if the termination condition is not met, the feedback iteration adjustment continues.

5. The product quality assessment method based on vernier caliper measurement big data according to claim 1, characterized in that, The corrected measurement value is compared with the product design tolerance range to output a quality acceptance result, including: Obtain the distance between the corrected measurement value and the nearest limit of the product design tolerance range, wherein the nearest limit of the product design tolerance range is the limit that is closer to the corrected measurement value among the upper and lower limits of the product design tolerance range; Obtain current environmental parameters and equipment operating parameters; When the distance is less than a preset critical distance, the enhanced judgment mode is activated: Repeated measurements are performed a preset number of times within a preset short time window, and the corrected measurement values ​​and their corresponding environmental parameters and equipment operating parameters are continuously collected; Based on the environmental parameters and equipment operating parameters corresponding to each measurement, and combined with the historical statistical relationship under similar environments and equipment conditions, the fluctuation range of the corresponding corrected measurement value is calculated. When the corrected measurement value and its fluctuation range are within the product design tolerance range for multiple consecutive measurements within the preset number of measurements, the product is deemed qualified; when the corrected measurement value and its fluctuation range are outside the product design tolerance range for multiple consecutive measurements within the preset number of measurements, the product is deemed unqualified. When the corrected measurement value and its fluctuation range cross the product design tolerance range, calculate the instantaneous rate of change of the environmental parameters and the equipment operating parameters; When the instantaneous rate of change exceeds a preset instantaneous rate of change threshold, the preset short time window is extended and / or the preset number of measurements is increased, and a determination is made based on the updated data according to the enhanced determination mode.

6. The product quality assessment method based on big data measurement using vernier calipers according to claim 5, characterized in that, The process of comparing the corrected measurement value with the product design tolerance range to output a quality acceptance result also includes: The distance is continuously monitored, and its historical sequence and rate of change are recorded within a preset sliding time window; When it is determined from the historical sequence that the distance fluctuates for a duration greater than a preset duration in the neighborhood of the preset critical distance, the value of the preset critical distance is dynamically adjusted according to the fluctuation frequency and amplitude of the distance. When the start or exit conditions of the enhanced judgment mode are met, wait for a preset mode switching delay time, and only execute the corresponding start or exit if the start or exit conditions are still met after the preset mode switching delay time has expired. The activation priority of the enhanced judgment mode is set according to the importance of the current production task.

7. The product quality assessment method based on vernier caliper measurement big data as described in claim 6, characterized in that, The activation priority of the enhanced judgment mode is set according to the importance of the current production task, including: Identify the type of part to be tested and its corresponding production task; Based on the part type and the production task, the corresponding quality judgment priority is obtained from the preset rule base, and the activation priority of the enhanced judgment mode is set accordingly. Monitor system resource status, which includes at least computing resource usage and measurement device load; When the distance is less than the preset critical distance, the activation decision of the enhanced judgment mode is made based on the quality judgment priority and the system resource status: when the quality judgment priority is high and the system resource status meets the preset resource threshold, the enhanced judgment mode is activated immediately; when the quality judgment priority is high but the system resource status does not meet the preset resource threshold, the enhanced judgment mode of the part type with low activation priority is paused or switched to the normal judgment mode, and the enhanced judgment mode of the part with high activation priority is activated; when the quality judgment priority is low and the system resource status meets the preset resource threshold, the enhanced judgment mode is activated; when the quality judgment priority is low and the system resource status does not meet the preset resource threshold, the activation of the enhanced judgment mode is delayed or the normal judgment mode is maintained. After the enhanced judgment mode is activated, the preset number of measurements and the sampling frequency within the preset short time window are dynamically adjusted according to the quality judgment priority and the system resource status.

8. The product quality assessment method based on vernier caliper measurement big data as described in claim 6, characterized in that, After the enhanced determination mode is activated, it also includes: Continuously monitor the data stream status of the measuring equipment and auxiliary sensors, wherein the data stream status includes at least data packet integrity, transmission rate, and effective range of sensor output; When any interruption or abnormality is detected in any of the data stream states, a fault warning is triggered, and the data acquisition and judgment process in the enhanced judgment mode is suspended. During the pause, the environmental parameters and equipment operating parameters at the time of the fault occurrence are recorded, and the valid measurement data collected before the fault occurred are stored. Continuously monitor the recovery status of faulty measuring equipment or auxiliary sensors; Once the faulty measuring device or auxiliary sensor returns to normal, based on the recorded environmental parameters and device operating parameters at the time of the fault occurrence, it is determined whether the suspended determination task needs to be retrospectively processed. When it is determined that retrospective processing is required, the measurement data for the corresponding time period is re-collected after the fault is recovered, and merged with the stored valid measurement data to restart the determination process in the enhanced determination mode. When it is determined that no retrospective processing is required, the data acquisition and determination process under the enhanced determination mode is smoothly restored after the fault is recovered.

9. The product quality assessment method based on vernier caliper measurement big data according to claim 8, characterized in that, When the enhanced judgment mode is activated and the data stream is normal, before comparing the corrected measurement value with the product design tolerance range, the method further includes: Based on the vibration error characteristics, the network delay error characteristics, and the linearity deviation error characteristics, the corresponding confidence index is calculated and an error feature confidence vector is formed. Confidence gating coefficients are generated based on the error feature confidence vector to dynamically adjust the control parameters of the decomposition iteration, including adaptive adjustment of the preset low correlation threshold, decomposition step size and decomposition priority. When the confidence index of any error feature is lower than the preset confidence threshold, phase stabilization resampling is triggered within a preset short time window, and additional measurements are performed in the phase stabilization interval selected based on the actual measurement timestamp and vibration phase, while the historical baseline of the network delay error feature is updated synchronously; when the phase stabilization resampling and the updated historical baseline of the network delay error feature still indicate high uncertainty, fast verification point measurement is performed on the gold sample to constrain the local extrapolation distance of the linearity deviation from the model; Based on the confidence gating coefficient, the correction amount of the vibration error component and the linearity deviation error component is reduced or amplified accordingly to obtain the confidence gating correction measurement value, and the confidence gating correction measurement value is used to replace the correction measurement value for subsequent quality acceptance judgment.

10. A product quality assessment system based on big data measurement using vernier calipers, characterized in that, include: The data acquisition module is used to collect vernier caliper measurement data and its local timestamp, environmental vibration data, network transmission status data, and measurement equipment calibration status data. The network transmission status data includes the signal strength and channel occupancy before transmission, access end load and channel interference, and server receiving timestamp. The measurement equipment calibration status data includes the measurement results of the gold sample and environmental and equipment operating parameters. The first feature extraction module is used to perform timestamp calibration based on the local timestamp, server receiving timestamp, and network transmission status data of the vernier caliper measurement data, estimate the actual network transmission delay, obtain the real measurement timestamp, and determine the actual network transmission delay and its fluctuation relative to the historical baseline as network delay error features. The second feature extraction module is used to time-align the vernier caliper measurement data with the environmental vibration data using the actual measurement timestamp, perform time-frequency analysis to obtain the vibration phase at the moment of measurement, and calculate the instantaneous deviation caused by vibration accordingly to determine the vibration error feature. The third feature extraction module is used to establish a linearity deviation model of the measuring equipment under different measurement ranges and operating conditions based on the measurement results of the gold sample and combined with the environmental and equipment operating parameters, and to calculate the system deviation caused by the linearity deviation of the equipment and determine it as the linearity deviation error feature. The error component decomposition module is used to calculate the total deviation between the vernier caliper measurement data and the target size, and decompose it into the vibration error component corresponding to the vibration error feature, the network delay error component corresponding to the network delay error feature, the linearity deviation error component corresponding to the linearity deviation error feature, and other error components. The judgment module is used to correct the vernier caliper measurement data based on the vibration error component and the linearity deviation error component, obtain the corrected measurement value, and compare the corrected measurement value with the product design tolerance range to output a quality qualification judgment result.

Images