Injection molding size deviation monitoring method and system based on machine vision

Abstract

The invention discloses an injection molding size deviation monitoring method based on machine vision, and the method comprises the steps: collecting geometric data of each region of an injection molding part through a sensor array, and generating original size distribution information; according to the geometric feature vector of each region, determining a dimensional deviation initial distribution state; analyzing the initial distribution state of the dimensional deviation by using a preset deviation feature library, and generating an abnormal region identifier set; determining a key process parameter combination causing deviation in combination with the production process parameter record; generating a parameter adjustment instruction for the abnormal area through the feedback control model, and updating the operation parameters of the injection molding equipment; and judging whether the correction effect reaches a preset standard or not, if the correction effect does not reach the preset standard, updating the deviation feature library, generating an optimized parameter adjustment instruction, and circularly executing the correction and monitoring process. Therefore, automatic identification, analysis and correction of the size deviation of the injection molding product can be realized, and the product quality stability and the production efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of injection molding dimensional deviation monitoring technology, and in particular to a machine vision-based injection molding dimensional deviation monitoring method, system, terminal, and storage medium. Background Technology

[0002] Injection molding, as a core area of ​​manufacturing, directly impacts product functionality and consistency through the control of dimensional deviations. Researching how to achieve precise monitoring and dynamic adjustment of dimensional deviations is a key direction for improving manufacturing efficiency. Current technologies rely heavily on manual sampling or fixed-pattern equipment inspection, which struggles to adapt to changing environmental factors and diverse product forms during production. Especially when dealing with complex-shaped products, existing technologies often fail to accurately distinguish subtle changes in different areas, leading to vague problem localization and a lack of targeted adjustment measures. Therefore, constructing a system capable of real-time dynamic monitoring of dimensional changes in multiple areas of injection-molded products and automatically adjusting production parameters based on the deviation characteristics of different parts has become a crucial problem to be solved. Summary of the Invention

[0003] The technical problem to be solved by this invention is that it is currently difficult to achieve accurate monitoring and dynamic adjustment of dimensional deviations in injection molding.

[0004] To address the aforementioned technical problems, the first aspect of this invention discloses a machine vision-based method for monitoring dimensional deviations in injection molding, the method comprising: Geometric data of multiple regions of the injection-molded product are acquired through a sensor array to generate original dimensional distribution information; Based on the original size distribution information, image segmentation technology is used to extract the geometric feature vectors of each region to determine the preliminary distribution state of size deviation. The initial distribution of the dimensional deviations is analyzed using a pre-defined deviation feature library, abnormal areas are marked, and a set of abnormal area identifiers is generated. Based on the set of abnormal area identifiers and the production process parameter records, the key process parameter combinations that cause the deviation are determined. The feedback control model generates parameter adjustment instructions for the abnormal region, and updates the operating parameters of the injection molding equipment. Collect corrected geometric data from multiple regions, analyze the trend of dimensional deviation changes, and determine whether the correction effect meets the preset standard. If the correction effect does not meet the preset standard, the deviation feature library is updated, an optimized parameter adjustment instruction is generated, and the correction and monitoring process is executed cyclically.

[0005] As an optional implementation, in the first aspect of the present invention, the step of acquiring geometric data of multiple regions of the injection-molded product through a sensor array to generate original size distribution information further includes: Geometric data streams from the surface of injection molded products are collected using a high-precision sensor array, and shape features of different parts are digitally mapped. The geometric data streams are integrated using data fusion technology to generate a unified size distribution mapping result; If the size distribution mapping result does not match the preset threshold range, secondary data collection is performed on the abnormal area to correct the original size distribution information.

[0006] As an optional implementation, in the first aspect of the present invention, the step of extracting geometric feature vectors of each region using image segmentation technology based on the original size distribution information to determine the preliminary distribution state of size deviation further includes: The original size distribution information is partitioned using image segmentation technology to generate a set of independent regions. For each region in the set of independent regions, geometric parameters are extracted, and feature vectors are constructed. Based on the feature vectors, a classification algorithm is used to determine the preliminary distribution of size deviations in each region.

[0007] As an optional implementation, in the first aspect of the present invention, the step of analyzing the preliminary distribution of the size deviation using a preset deviation feature library and marking abnormal areas further includes: Based on the initial distribution of the size deviations, a set of deviation feature vectors for each region is generated; The deviation feature library is used for comparative analysis. If the deviation feature vector exceeds a preset threshold range, it is marked as an abnormal region. Data cleaning processes generate a set of filtered abnormal area identifiers. As an optional implementation, in the first aspect of the present invention, the step of determining the key process parameter combination causing the deviation based on the abnormal region identifier set and the production process parameter records further includes: Retrieve the production process parameter records corresponding to the abnormal area identifier set from the database; A correlation analysis algorithm was used to determine the correlation between the production process parameter records and deviations; Based on the aforementioned correlation, a subset of high-impact process parameters is extracted to determine the combination of key process parameters.

[0008] As an optional implementation, in the first aspect of the present invention, the step of generating parameter adjustment instructions for the abnormal region through a feedback control model and updating the operating parameters of the injection molding equipment further includes: Based on the combination of key process parameters, the feedback control model is invoked to generate preliminary parameter adjustment instructions. By verifying the initial parameter adjustment command in conjunction with historical adjustment records, an optimized parameter configuration scheme is generated. The operating parameters of the injection molding equipment are updated using the parameter configuration scheme to perform dynamic correction.

[0009] As an optional implementation, in the first aspect of the present invention, the step of collecting corrected multi-region geometric data, analyzing the trend of dimensional deviation changes, and determining whether the correction effect meets the preset standard further includes: Corrected multi-region geometric data is collected using a sensor array to generate deviation distribution data. The rate of change of the deviation distribution data is calculated using a time series analysis algorithm; Based on the comparison between the rate of change and the preset standard, a correction effect judgment result is generated; if the correction effect judgment result does not meet the standard, optimization parameter suggestions are generated.

[0010] A second aspect of this invention discloses a machine vision-based injection molding dimensional deviation monitoring system, the system comprising: The information acquisition module is used to acquire geometric data of multiple regions of the injection molded product through a sensor array and generate original size distribution information; The image segmentation module is used to extract the geometric feature vectors of each region based on the original size distribution information and to determine the preliminary distribution state of the size deviation. The analysis module is used to analyze the preliminary distribution of the size deviation using a preset deviation feature library, mark abnormal areas, and generate a set of abnormal area identifiers. The module is used to determine the combination of key process parameters that cause the deviation based on the set of abnormal area identifiers and the production process parameter records. The parameter update module is used to generate parameter adjustment instructions for the abnormal region through the feedback control model and update the operating parameters of the injection molding equipment. The judgment module is used to collect multi-region geometric data after correction, analyze the trend of dimensional deviation changes, and determine whether the correction effect meets the preset standard. The loop module is used to update the deviation feature library and generate optimized parameter adjustment instructions if the correction effect does not meet the preset standard, and to repeatedly execute the correction and monitoring process.

[0011] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute some or all of the steps in the machine vision-based injection molding dimensional deviation monitoring method disclosed in the first aspect of the present invention.

[0012] Compared with the prior art, the embodiments of the present invention have the following beneficial effects: In this embodiment of the invention, geometric data of multiple regions of the injection-molded product are acquired through a sensor array to generate original dimensional distribution information; geometric feature vectors of each region are extracted using image segmentation technology to determine the preliminary distribution state of dimensional deviations; the preliminary distribution state of dimensional deviations is analyzed using a preset deviation feature library, abnormal regions are marked, and an abnormal region identifier set is generated; the key process parameter combinations causing deviations are determined by combining production process parameter records; parameter adjustment instructions for abnormal regions are generated through a feedback control model to update the operating parameters of the injection molding equipment; the corrected multi-region geometric data are collected, the trend of dimensional deviation changes is analyzed, and it is determined whether the correction effect meets the preset standard; if the correction effect does not meet the preset standard, the deviation feature library is updated, optimized parameter adjustment instructions are generated, and the correction and monitoring process is executed cyclically.

[0013] As can be seen, the present invention can realize the automatic identification, analysis and correction of dimensional deviations in injection molded products, thereby improving product quality stability and production efficiency. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a schematic flowchart of a machine vision-based injection molding dimensional deviation monitoring method disclosed in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a machine vision-based injection molding dimensional deviation monitoring system disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of a machine vision-based injection molding dimensional deviation monitoring terminal disclosed in an embodiment of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0018] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0019] Example 1 Please see Figure 1 , Figure 1 This is a schematic flowchart of a machine vision-based injection molding dimensional deviation monitoring method disclosed in an embodiment of the present invention. Figure 1 As shown, the machine vision-based injection molding dimensional deviation monitoring method may include the following operations: S11. Obtain geometric data of multiple regions of the injection molded product through a sensor array to generate original size distribution information.

[0020] In step S11 above, a sensor array is formed by arranging multiple laser displacement sensors and industrial cameras at the injection molding machine exit. When the injection-molded product passes through the detection area, each sensor synchronously collects geometric data from different parts of the product, forming a multi-region data stream. During the acquisition process, corresponding scanning strategies are adopted for the structural characteristics of different parts of the product (such as planes, curved surfaces, edges, etc.) to obtain complete surface geometric information. The acquired surface geometric information is preprocessed, including noise filtering and coordinate unification, to obtain preliminary surface data information. Subsequently, digital processing technology is used to analyze the geometric features of each region through feature extraction algorithms. The extracted feature values ​​are compared with preset process standard thresholds. When a feature value of a certain region exceeds the threshold range, it is marked as a region to be confirmed, and the system automatically starts a secondary data acquisition process to obtain more accurate measurement data by adjusting the sensor parameters. The secondary acquisition data is compared and analyzed with the initial measurement results to confirm whether there is a real deviation. If a deviation is confirmed, the original size distribution data is corrected and updated, and finally, the verified size distribution information is output.

[0021] In an optional embodiment, the step of acquiring geometric data of multiple regions of the injection-molded product through a sensor array to generate original size distribution information includes: acquiring geometric data streams of the injection-molded product surface through a high-precision sensor array, and digitally mapping the shape features of different parts; integrating the geometric data streams using data fusion technology to generate a unified size distribution mapping result; if the size distribution mapping result does not match a preset threshold range, performing secondary data acquisition for abnormal areas to correct the original size distribution information.

[0022] A high-precision sensor array is used to acquire geometric data streams from the surface of injection-molded parts, digitally mapping the shape features of different parts of the part. The sensor array employs a combination of multiple laser displacement sensors and industrial cameras, synchronously triggering acquisition during ejection of the injection molded part to obtain complete surface geometric data. The acquired data stream undergoes coordinate unification and data preprocessing, followed by data fusion technology to integrate multi-source measurement data into a unified dimensional distribution mapping result. When the dimensional distribution results in certain areas exceed a preset threshold range, the system automatically performs secondary data acquisition on these abnormal areas. By comparing the two measurement results, the original dimensional distribution information is corrected to ensure the accuracy of the final output dimensional distribution mapping result. The entire acquisition and processing process is synchronized with the production line cycle time, providing a reliable data foundation for subsequent deviation analysis.

[0023] The entire process achieves closed-loop control from data acquisition to analysis and processing, ensuring the accuracy and reliability of dimensional inspection results. The system has an extremely short runtime, meeting the real-time inspection needs of the production line.

[0024] S12. Based on the original size distribution information, image segmentation technology is used to extract the geometric feature vectors of each region to determine the preliminary distribution state of size deviation.

[0025] In step S12 above, the raw dimensional data collected by the sensor is first preprocessed, including noise filtering and coordinate normalization, to provide a clean data foundation for subsequent analysis. Based on the preprocessed data, an image segmentation algorithm based on edge detection is used to divide the surface of the injection-molded product into multiple feature regions, forming a preliminary region set. For each region, key geometric features, including flatness, roundness, and contour deviation, are extracted to construct feature vectors with clear physical meaning. By statistically analyzing the numerical distribution of the feature vectors of each region and comparing them with preset process standard thresholds, the dimensional deviation status of each region is preliminarily determined. When the deviation features of some regions are found to exceed the allowable range, the system will initiate a secondary correction process, optimizing and adjusting the feature vectors of these regions by re-collecting data and performing feature weighted calculations. The corrected feature vectors are input into a pre-trained support vector machine classification model, which is built based on historical process data and can accurately identify typical deviation types such as shrinkage and warpage. Based on the classification results, the system automatically generates a dimensional compensation strategy for each region, including suggested process parameter adjustments and compensation values. Ultimately, by implementing these optimization strategies, precise adjustments to the size distribution of multiple regions are achieved, and optimized distribution results reflecting the actual quality status are output, providing a reliable basis for subsequent process improvements.

[0026] In an optional embodiment, the method based on the original size distribution information and using image segmentation... The segmentation technique extracts geometric feature vectors from each region to determine the preliminary distribution of size deviations. This includes: partitioning the original size distribution information using image segmentation technology to generate a set of independent regions; extracting geometric parameters for each region in the set of independent regions to construct a feature vector; and using a classification algorithm based on the feature vectors to determine the preliminary distribution of size deviations in each region.

[0027] First, the collected raw dimensional distribution information undergoes standardized preprocessing, including noise removal and unified coordinate reference. Based on the preprocessed data, a region growing image segmentation algorithm is used to divide the product surface into multiple independent regions according to the continuity of geometric features, forming a preliminary set of partitions including key parts such as the gate area and parting surface. For each partition region, the system extracts basic geometric features such as length deviation, thickness deviation, and flatness, constructing corresponding feature vectors. By statistically analyzing the numerical distribution of feature vectors in each region and comparing them with preset process standard thresholds, the preliminary distribution of dimensional deviations is obtained. When deviation features in some regions are detected to exceed the allowable range, the system initiates a secondary correction process, optimizing and adjusting the feature vectors of these regions by re-collecting data and performing feature weighted calculations. The corrected feature descriptions are input into a support vector machine classification model, which is built based on historical process data and can identify typical deviation types. Based on the classification results, the system generates a dimensional compensation strategy for each region, ultimately optimizing and adjusting the dimensional distribution of multiple regions and outputting optimized distribution results that reflect the actual quality status. The entire process adopts a modular design with close integration of each link, ensuring both analytical accuracy and processing efficiency that meets the real-time requirements of the production line.

[0028] In an optional embodiment, the step of extracting geometric feature vectors of each region using image segmentation techniques based on the original size distribution information to determine the preliminary distribution state of size deviations further includes: For the same inspection area of ​​the injection-molded product, measurement data from three different sensors were simultaneously collected: a laser displacement sensor measured a thickness deviation ΔL = 0.12 mm (confidence weight 0.6), an industrial CCD camera measured a contour deviation ΔC = 0.15 mm (confidence weight 0.3), and a contact probe measured an actual value ΔP = 0.10 mm (confidence weight 0.1). Subsequently, a weighted fusion algorithm was used to calculate the comprehensive deviation value for this area, with the formula Δ = 0.12 × 0.6 + 0.15 × 0.3 + 0.10 × 0.1 = 0.127 mm. The weight allocation for each sensor was dynamically adjusted based on its measurement accuracy: the laser sensor had a weight of 1 / (0.05²) (accuracy ±0.05 mm), the vision sensor had a weight of 1 / (0.1²) (accuracy ±0.1 mm), and the contact probe had a weight of 1 / (0.2²) (accuracy ±0.2 mm). The final weight values ​​were normalized to ensure a sum of 1. Furthermore, the system incorporates an anomaly data filtering mechanism. If a sensor's measured value deviates from the median by more than three times the standard deviation (3σ), its weight is automatically reduced to 10% to prevent anomalies from interfering with the fusion results. This method effectively improves the fusion accuracy of multi-source heterogeneous data, providing a more reliable data foundation for subsequent deviation analysis and process optimization.

[0029] S13. Analyze the preliminary distribution of the size deviation using a preset deviation feature library, mark abnormal areas, and generate an abnormal area identifier set.

[0030] In step S13 above, the distribution of dimensional deviations is first analyzed. The system divides the detection area into several standard unit areas based on the product's structural characteristics. For each unit area, basic feature parameters, including the maximum deviation value, average deviation value, and standard deviation, are extracted to construct a comparable feature vector. The system calls a pre-established deviation feature library for comparison analysis. This feature library contains feature threshold ranges under typical normal operating conditions. When the feature vector of a certain area exceeds the corresponding threshold, the system marks it as an abnormal area to be confirmed, forming a preliminary identification set. To improve the accuracy of the judgment, the system performs data cleaning on the preliminary identification set: first, moving average filtering is used to eliminate random measurement errors; second, a neighboring area comparison method is used to exclude isolated anomalies; and finally, manual re-inspection confirms suspicious areas. After screening, the system outputs the final abnormal area identification set and records detailed feature data for each abnormal area, including deviation values, spatial location, and distribution patterns. Based on this feature data, the system classifies abnormal areas according to the severity and scope of the deviation, identifying typical deviation patterns such as shrinkage and warping. For the classified outlier regions, the system employs a priority evaluation model based on support vector machines, comprehensively considering factors such as deviation magnitude, location criticality, and process sensitivity to generate a processing priority sequence. The evaluation process incorporates weighted coefficients, with dimensional deviations accounting for 40%, location criticality for 30%, and process sensitivity for 30%. Based on the priority evaluation results, the system automatically matches a pre-defined adjustment strategy library, generating a specific solution for each outlier region that includes suggestions for adjusting process parameters and compensation values.

[0031] In one optional embodiment, the step of analyzing the preliminary distribution of the size deviation using a preset deviation feature library, marking abnormal areas, and generating an abnormal area identifier set includes: generating a set of deviation feature vectors for each area based on the preliminary distribution of the size deviation; performing comparative analysis using the deviation feature library, and marking the deviation feature vectors as abnormal areas if they exceed a preset threshold range; and generating a filtered set of abnormal area identifiers through data cleaning processing.

[0032] First, features are extracted from each detection area to generate a set of feature vectors containing key parameters such as maximum deviation, average deviation, and standard deviation. The system then calls a pre-set deviation feature library for intelligent comparison and analysis. This library stores normal threshold ranges for each feature parameter based on historical qualified data statistics, as well as feature templates for typical defect patterns. During the comparison process, the system employs a multi-level judgment strategy: first, it checks whether the basic deviation parameters exceed the allowable range; then, it analyzes the matching degree between the feature vectors and typical defect patterns; and finally, after comprehensive evaluation, areas exceeding the threshold are marked as anomalies to be confirmed. To improve the accuracy of the judgment, the system performs data cleaning on the initially marked anomaly areas, eliminating random errors through moving average filtering and excluding isolated anomalies using a neighboring region comparison method, ultimately generating a validated set of anomaly area identifiers. The entire analysis process adopts a standardized processing flow, ensuring both reliability and processing efficiency, and providing accurate anomaly area location information for subsequent process adjustments.

[0033] S14. Based on the set of abnormal area identifiers and the production process parameter records, determine the key process parameter combination that causes the deviation.

[0034] In step S14 above, the system first extracts complete process parameter records corresponding to the abnormal areas from the production database, including key parameters such as injection pressure, holding time, and mold temperature, forming an initial dataset. The system then performs standardization and cleaning on this dataset, including removing invalid records, filling in missing values, and unifying timestamps to ensure data quality. During the feature extraction stage, the system calculates the statistical characteristics (such as mean, range, and fluctuation coefficient) of each process parameter and their spatiotemporal correlation characteristics with the deviation area, constructing a multi-dimensional parameter feature set. By comparing against a preset process standard threshold library, feature parameters exceeding the allowable fluctuation range are selected, and priority is assigned based on the parameter's impact weight on product quality to determine a subset of high-impact parameters. Subsequently, the system uses a random forest algorithm to analyze the interaction relationships between these key parameters, constructing a parameter correlation matrix to quantify the contribution of each parameter combination to the formation of the deviation. By setting a reasonable correlation strength threshold (such as correlation coefficient > 0.7), the system identifies the core parameter combinations leading to the deviation, such as typical patterns like "interaction between injection speed and mold temperature," and stores these deterministic causal relationships in the process knowledge base, providing a basis for subsequent parameter optimization.

[0035] In one optional embodiment, determining the key process parameter combination causing the deviation based on the abnormal area identifier set and the production process parameter records includes: obtaining the production process parameter records corresponding to the abnormal area identifier set from the database; using a correlation analysis algorithm to determine the correlation between the production process parameter records and the deviation; and based on the correlation, extracting a subset of high-impact process parameters to determine the key process parameter combination.

[0036] First, process parameter records corresponding to abnormal areas, including key parameters such as injection pressure and holding time, are extracted from the production database to form an initial dataset. The data is then cleaned to remove outliers and missing data, ensuring data quality. Next, statistical characteristics of each process parameter, such as mean and range, are extracted using historical data to form a parameter feature set. These features are compared against preset process standard thresholds to filter out those exceeding allowable ranges. They are then prioritized based on their impact on quality, identifying a subset of high-impact parameters. Based on this, the interaction relationships between parameters are analyzed to identify the core parameter combinations causing deviations. Finally, parameter combinations with correlation strength exceeding preset thresholds are stored in the results database as a basis for process optimization.

[0037] In an optional embodiment, determining the key process parameter combination causing the deviation based on the abnormal region identifier set and the production process parameter records further includes: First, the dimensional deviation sequence of abnormal region A from the most recent 10 production cycles was extracted (e.g., [0.18mm, 0.22mm, 0.15mm, 0.25mm, 0.20mm...]). Simultaneously, the corresponding process parameter fluctuation data were acquired, including the injection pressure sequence ([85MPa, 88MPa, 82MPa, 90MPa, 86MPa...]) and the mold temperature sequence ([45℃, 47℃, 44℃, 48℃, 46℃...]). By calculating the Spearman rank correlation coefficient, it was found that the correlation coefficient between the deviation and injection pressure was ρ=0.82 (P<0.01), and the correlation coefficient with mold temperature was ρ=0.75 (P<0.05), while the correlation with holding time was weak (ρ=0.32, P>0.1). The system sets a dynamic judgment threshold. When |ρ|>0.7 and P<0.05, it is judged as a strongly correlated parameter, and a univariate linear regression model of injection pressure and deviation is established accordingly (deviation = 0.023 × pressure - 1.721, R² = 0.67). When a significant interaction between mold temperature and cooling time is detected (ANOVA test P<0.01), the system generates a two-dimensional process parameter adjustment matrix and recommends the optimal parameter combination (e.g., temperature ≤47℃ and holding time ≥10s). This implementation method, through rigorous statistical verification and dynamic modeling, significantly improves the scientificity and accuracy of process parameter adjustment, providing data support for the precise control of the injection molding process.

[0038] S16. Collect the corrected geometric data of multiple regions, analyze the trend of dimensional deviation changes, and determine whether the correction effect meets the preset standard.

[0039] In step S16 above, the acquired multi-region geometric data is first preprocessed. Random noise is eliminated by moving average filtering, and abnormal measurement values ​​are removed using the 3σ criterion to form a reliable geometric dataset. Based on this dataset, the system calculates the deviation between the actual size and the theoretical value of each detection area, generating complete deviation distribution data. Through a pre-established deviation calculation model, key features such as the maximum value, average value, and standard deviation of the deviation in each region are extracted to determine the preliminary deviation distribution state. The system analyzes the changing trend of deviations in each region over multiple consecutive production cycles, constructing feature data including trend slope and fluctuation amplitude. The support vector machine algorithm is used to classify the trend features and determine whether the deviation changes tend to stabilize. For stable trends, the system compares the current deviation features with the preset process standards and calculates the matching degree index; for unstable trends, a data re-acquisition process is triggered. Based on the matching degree analysis results, the system generates an evaluation report containing a correction effect score and optimization suggestions to guide the targeted adjustment of equipment parameters. The entire process adopts an iterative optimization mechanism until the correction effect reaches the preset standard requirements, ensuring continuous improvement in dimensional quality.

[0040] In one optional embodiment, the step of collecting and correcting multi-region geometric data, analyzing the trend of dimensional deviation changes, and determining whether the correction effect meets the preset standard includes: collecting and correcting multi-region geometric data through a sensor array to generate deviation distribution data; using a time series analysis algorithm to calculate the rate of change of the deviation distribution data; generating a correction effect judgment result based on the comparison of the rate of change with the preset standard; and generating optimization parameter suggestions if the correction effect judgment result does not meet the standard.

[0041] The corrected product is scanned from all sides to obtain the latest dimensional data of each detection area. After noise reduction and standardization, deviation distribution data for the current batch is generated. Based on data from multiple consecutive production batches, a moving average algorithm is used to calculate the trend line of deviation changes in each area, focusing on analyzing the deviation change rate of key quality feature points (such as parting surface and gate area). The calculated change rate is compared with the preset process standard threshold: if the change rate continues to decrease and enters a stable range (e.g., change rate ≤ 0.5% for 3 consecutive batches), the correction effect is considered satisfactory; if the change rate does not show a convergence trend or exceeds the allowable fluctuation range, the optimization process is triggered. For cases that do not meet the standards, the system combines the historical optimization case library to generate specific suggestions including the direction and magnitude of parameter adjustment (e.g., "increase mold temperature by 2-3℃"), and records the current judgment result for subsequent analysis.

[0042] The entire process employs automated data acquisition and analysis procedures to ensure the timeliness and accuracy of the calibration effect evaluation, providing a reliable basis for continuous process optimization.

[0043] S17. If the correction effect does not meet the preset standard, update the deviation feature library, generate optimized parameter adjustment instructions, and repeatedly execute the correction and monitoring process.

[0044] In step S17 above, the deviation data of the specified area is first analyzed for trend. By calculating the moving average and standard deviation, it is determined whether the deviation change tends to stabilize and meets the preset standard (e.g., deviation fluctuation ≤ 0.1mm for 5 consecutive batches). When a non-compliant area is found, the system extracts the latest deviation feature data (including deviation amount, distribution pattern, and change gradient) for that area and compares it with historical adjustment records to determine the feature update direction. By querying the feature database, matching criteria for similar cases are obtained, and a support vector machine model is used to classify the current deviation features and identify their optimization potential. For features with optimization potential, the system merges the current data with historical effective adjustment schemes and updates the matching rules in the feature database. Based on the optimized matching criteria, targeted parameter adjustment schemes are generated (e.g., holding time ±0.5s, mold temperature ±2℃, etc.). After the correction is performed, the system continuously monitors the deviation change trend. If it still does not meet the standard, the analysis and optimization cycle is restarted until the deviation enters a stable and controllable range.

[0045] In one optional embodiment, updating the deviation feature library and generating optimized parameter adjustment instructions includes: extracting the latest deviation feature data of areas that do not meet the preset standard; updating the deviation feature library in combination with historical adjustment records to generate optimized feature matching criteria; predicting deviation trends using machine learning algorithms based on the feature matching criteria to generate a new round of parameter adjustment instructions; and storing the parameter adjustment instructions in association with area identifiers to generate a process adjustment log.

[0046] The system extracts the latest deviation characteristics of non-compliant areas in the current production batch, including core parameters such as dimensional deviations and distribution patterns. These characteristic data are compared and analyzed with historical adjustment records to identify common features in effective adjustment schemes, updating the deviation feature library to form new matching rules. Based on the optimized feature library, the system uses a trained machine learning model (such as the random forest method) to predict deviation trends and comprehensively considers the constraints of the equipment's process window to generate a new round of adjustment instructions containing specific parameter adjustment amounts and directions (e.g., "increase injection pressure by 5 bar, reduce holding time by 0.3 seconds"). All adjustment instructions are associated with corresponding area identification information (such as the gate area, parting line, etc.) to form a complete process adjustment log, recording information such as adjustment time, parameter changes, and personnel involved. This log also serves as the data foundation for subsequent optimization analysis, enabling continuous self-improvement of the feature library.

[0047] It is evident that implementation Figure 1The machine vision-based injection molding dimensional deviation monitoring method described herein can automatically identify, analyze and correct dimensional deviations in injection molded products, thereby improving product quality stability and production efficiency.

[0048] Example 2 Please see Figure 2 , Figure 2 This is a schematic diagram of a machine vision-based device for monitoring dimensional deviations in injection molding, as disclosed in an embodiment of the present invention. Figure 2 As shown, the machine vision-based injection molding dimensional deviation monitoring device may include the following modules: Information acquisition module 201 is used to acquire geometric data of multiple regions of injection molded products through sensor array and generate original size distribution information; Image segmentation module 202 is used to extract geometric feature vectors of each region based on the original size distribution information and to determine the preliminary distribution state of size deviation; Analysis module 203 is used to analyze the preliminary distribution of the size deviation using a preset deviation feature library, mark abnormal areas, and generate an abnormal area identifier set; Combined with module 204, it is used to determine the combination of key process parameters that cause the deviation based on the set of abnormal area identifiers and the production process parameter records. The parameter update module 205 is used to generate parameter adjustment instructions for the abnormal area through the feedback control model and update the operating parameters of the injection molding equipment. The judgment module 206 is used to collect the corrected multi-region geometric data, analyze the trend of dimensional deviation changes, and determine whether the correction effect meets the preset standard. The loop module 207 is used to update the deviation feature library and generate optimized parameter adjustment instructions if the correction effect does not meet the preset standard, and to repeatedly execute the correction and monitoring process.

[0049] For a detailed explanation of the machine vision-based injection molding dimensional deviation monitoring device, please refer to the detailed explanation of the machine vision-based injection molding dimensional deviation monitoring method. To avoid repetition, it will not be repeated here.

[0050] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of a machine vision-based injection molding dimensional deviation monitoring terminal disclosed in an embodiment of the present invention. Figure 3 As shown, the machine vision-based injection molding dimensional deviation monitoring terminal may include: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute some or all of the steps in the machine vision-based injection molding dimensional deviation monitoring method disclosed in Embodiment 1 of the present invention.

[0051] This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute the steps in the machine vision-based injection molding dimensional deviation monitoring method disclosed in Embodiment 1 of this invention.

[0052] The system embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0053] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0054] Finally, it should be noted that the machine vision-based injection molding dimensional deviation monitoring method, system, terminal, and storage medium disclosed in the embodiments of this invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this invention.

Claims

1. A machine vision-based method for monitoring dimensional deviations in injection molding, characterized in that, include: Geometric data of multiple regions of the injection-molded product are acquired through a sensor array to generate original dimensional distribution information; Based on the original size distribution information, image segmentation technology is used to extract the geometric feature vectors of each region to determine the preliminary distribution state of size deviation. The initial distribution of the dimensional deviations is analyzed using a pre-defined deviation feature library, abnormal areas are marked, and a set of abnormal area identifiers is generated. Based on the set of abnormal area identifiers and the production process parameter records, the key process parameter combinations that cause the deviation are determined. The feedback control model generates parameter adjustment instructions for the abnormal region, and updates the operating parameters of the injection molding equipment. Collect corrected geometric data from multiple regions, analyze the trend of dimensional deviation changes, and determine whether the correction effect meets the preset standard. If the correction effect does not meet the preset standard, the deviation feature library is updated, an optimized parameter adjustment instruction is generated, and the correction and monitoring process is executed cyclically.

2. The method as described in claim 1, characterized in that, The step of acquiring geometric data of multiple regions of the injection-molded product through a sensor array and generating original size distribution information includes: Geometric data streams from the surface of injection molded products are collected using a high-precision sensor array, and shape features of different parts are digitally mapped. The geometric data streams are integrated using data fusion technology to generate a unified size distribution mapping result; If the size distribution mapping result does not match the preset threshold range, secondary data collection is performed on the abnormal area to correct the original size distribution information.

3. The method as described in claim 1, characterized in that, Based on the original size distribution information, the geometric feature vectors of each region are extracted using image segmentation technology to determine the preliminary distribution state of size deviations, including: The original size distribution information is partitioned using image segmentation technology to generate a set of independent regions. For each region in the set of independent regions, geometric parameters are extracted, and feature vectors are constructed. Based on the feature vectors, a classification algorithm is used to determine the preliminary distribution of size deviations in each region.

4. The method as described in claim 1, characterized in that, The step of analyzing the preliminary distribution of the dimensional deviations using a preset deviation feature library and marking abnormal areas includes: Based on the initial distribution of the size deviations, a set of deviation feature vectors for each region is generated; The deviation feature library is used for comparative analysis. If the deviation feature vector exceeds a preset threshold range, it is marked as an abnormal region. Data cleaning processes generate a set of filtered abnormal area identifiers.

5. The method as described in claim 1, characterized in that, The step of determining the key process parameter combination causing the deviation based on the abnormal area identifier set and the production process parameter records includes: Retrieve the production process parameter records corresponding to the abnormal area identifier set from the database; A correlation analysis algorithm was used to determine the correlation between the production process parameter records and deviations; Based on the aforementioned correlation, a subset of high-impact process parameters is extracted to determine the combination of key process parameters.

6. The method as described in claim 1, characterized in that, The step of generating parameter adjustment instructions for the abnormal region through a feedback control model and updating the operating parameters of the injection molding equipment includes: Based on the combination of key process parameters, the feedback control model is invoked to generate preliminary parameter adjustment instructions. By verifying the initial parameter adjustment command in conjunction with historical adjustment records, an optimized parameter configuration scheme is generated. The operating parameters of the injection molding equipment are updated using the parameter configuration scheme to perform dynamic correction.

7. The method as described in claim 1, characterized in that, The process of collecting and correcting multi-region geometric data, analyzing the trend of dimensional deviation changes, and determining whether the correction effect meets the preset standard includes: Corrected multi-region geometric data is collected using a sensor array to generate deviation distribution data. The rate of change of the deviation distribution data is calculated using a time series analysis algorithm; Based on the comparison between the rate of change and the preset standard, a correction effect judgment result is generated; if the correction effect judgment result does not meet the standard, optimization parameter suggestions are generated.

8. The method as described in claim 1, characterized in that, The step of updating the deviation feature library and generating optimized parameter adjustment instructions includes: Extract the latest deviation feature data of areas that do not meet the preset standards; The deviation feature library is updated by combining historical adjustment records to generate optimized feature matching criteria; Based on the aforementioned feature matching criteria, a machine learning algorithm is used to predict the deviation trend and generate a new round of parameter adjustment instructions. The parameter adjustment instructions are associated with and stored with the region identifier to generate a process adjustment log.

9. A machine vision-based injection molding dimensional deviation monitoring system, characterized in that, include: The information acquisition module is used to acquire geometric data of multiple regions of the injection molded product through a sensor array and generate original size distribution information; The image segmentation module is used to extract the geometric feature vectors of each region based on the original size distribution information and to determine the preliminary distribution state of the size deviation. The analysis module is used to analyze the preliminary distribution of the size deviation using a preset deviation feature library, mark abnormal areas, and generate a set of abnormal area identifiers. The module is used to determine the combination of key process parameters that cause the deviation based on the set of abnormal area identifiers and the production process parameter records. The parameter update module is used to generate parameter adjustment instructions for the abnormal area through the feedback control model and update the operating parameters of the injection molding equipment. The judgment module is used to collect multi-region geometric data after correction, analyze the trend of dimensional deviation changes, and determine whether the correction effect meets the preset standard. The loop module is used to update the deviation feature library and generate optimized parameter adjustment instructions if the correction effect does not meet the preset standard, and to repeatedly execute the correction and monitoring process.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1 to 7.

Images