Method, device, medium and equipment for detecting enteral nutrition infusion set production

Abstract

This invention relates to a detection method, apparatus, medium, and equipment for the production of enteral nutrition infusion sets. The method includes acquiring an initial set of detection data during the manufacturing process, including continuous image sequences acquired by optical inspection equipment and dynamic pressure data sequences recorded by pressure sensors. Hierarchical feature analysis is then performed to generate a composite defect feature set. Based on this set, a defect classification model is invoked to generate defect type labeling results, indicating the failure mode category and defect spatial location information. Further, adaptive adjustment of production parameters is performed to generate a set of process optimization parameters, and the production line is reactivated for secondary detection and verification. This allows for comprehensive and accurate detection of various defects in the enteral nutrition infusion set production process, automatic adjustment of production parameters to optimize the production process, effectively reducing the defect rate and improving production quality.

Description

Technical Field

[0001] This disclosure relates to the field of medical device manufacturing technology, and in particular to a testing method, apparatus, medium and equipment for the production of enteral nutrition infusion sets. Background Technology

[0002] In the field of medical device manufacturing, enteral nutrition infusion sets are a key piece of equipment, and their production quality directly affects patient safety and treatment outcomes. Traditional methods for manufacturing and testing enteral nutrition infusion sets primarily rely on manual visual inspection and simple mechanical testing. These methods are not only inefficient but also easily affected by human factors and environmental conditions, making it difficult to guarantee the accuracy and consistency of test results. Furthermore, with the continuous advancement of medical technology and patients' increasing demands for medical quality, higher requirements are being placed on the production quality and testing precision of enteral nutrition infusion sets.

[0003] Currently, although some automated detection technologies have been introduced into the production process of enteral nutrition infusion sets, these technologies often can only detect single types of defects and lack the ability to comprehensively analyze and handle multiple defects. Furthermore, existing technologies often cannot directly guide the adjustment and optimization of production parameters after defects are detected, making it difficult to effectively reduce the defect rate during production. Summary of the Invention

[0004] The purpose of this invention is to provide a testing method, apparatus, medium, and equipment for the production of enteral nutrition infusion sets, aiming to accurately detect defects existing in the production of enteral nutrition infusion sets, thereby reducing the defect rate in the production process.

[0005] To achieve the above objectives, a first aspect of this disclosure provides a testing method for the production of enteral nutrition infusion sets, the method comprising:

[0006] During the manufacturing process of the enteral nutrition infusion set, a continuous image sequence of the surface area of ​​the enteral nutrition infusion set is acquired by an optical inspection device deployed at a preset first inspection station in a linear scanning manner, and a dynamic pressure data sequence of the enteral nutrition infusion set is acquired by a pressure sensor array deployed at a preset second inspection station in a periodic pressure loading test.

[0007] Based on the continuous image sequence and the dynamic pressure data sequence, an initial detection data set is constructed, and a hierarchical feature parsing operation is performed on the initial detection data set to generate a composite defect feature set containing morphological defect feature distribution information and sealing defect feature distribution information.

[0008] Based on the composite defect feature set, a preset defect classification model is invoked to generate defect type labeling results that indicate the failure mode category and corresponding defect spatial location information of the enteral nutrition infusion device;

[0009] Based on the defect type labeling results, an adaptive adjustment operation of production parameters is performed to generate a set of process optimization parameters that match the failure mode category. The production line is then reactivated based on the set of process optimization parameters, and a secondary detection and verification operation is performed to generate optimization verification results.

[0010] A second aspect of this disclosure provides a testing device for the production of enteral nutrition infusion sets, the device comprising:

[0011] The acquisition module is configured to acquire, during the manufacturing process of the enteral nutrition infusion set, a continuous image sequence of the surface area of ​​the enteral nutrition infusion set by means of an optical detection device deployed at a preset first detection station in a linear scanning manner, and to acquire the dynamic pressure data sequence of the enteral nutrition infusion set by means of a pressure sensor array deployed at a preset second detection station in a periodic pressure loading test.

[0012] The first generation module is configured to construct an initial detection data set based on the continuous image sequence and the dynamic pressure data sequence, and perform a hierarchical feature parsing operation on the initial detection data set to generate a composite defect feature set containing morphological defect feature distribution information and sealing defect feature distribution information.

[0013] The second generation module is configured to call a preset defect classification model based on the composite defect feature set to generate defect type labeling results that indicate the failure mode category and corresponding defect spatial location information of the enteral nutrition infusion device.

[0014] The third generation module is configured to perform adaptive adjustment of production parameters based on the defect type labeling results, generate a set of process optimization parameters that match the failure mode category, reactivate the production line operation according to the set of process optimization parameters, and perform secondary detection and verification operations to generate optimization verification results.

[0015] A third aspect of this disclosure provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method described in any of the first aspects.

[0016] A fourth aspect of this disclosure provides an electronic device, comprising:

[0017] A memory on which computer programs are stored;

[0018] A processor for executing the computer program in the memory to implement the steps of the method of any one of the first aspects.

[0019] This invention provides a testing method, apparatus, medium, and equipment for the production of enteral nutrition infusion sets. Compared with the prior art, it has the following advantages:

[0020] By acquiring the initial test data set associated with enteral nutrition infusion sets during the manufacturing process and performing hierarchical feature analysis, the morphological and sealing defects of the enteral nutrition infusion sets can be comprehensively and accurately identified. Based on these composite defect features, a preset defect classification model is invoked to generate defect type labeling results, which not only indicate the failure mode category of the enteral nutrition infusion set but also provide corresponding defect spatial location information. By performing adaptive adjustment of production parameters, a set of process optimization parameters matching the failure mode category can be generated, directly guiding the adjustment and optimization of the production line, thereby effectively reducing the defect rate in the production process. Finally, by reactivating the production line and performing secondary test verification, optimized verification results can be generated, ensuring continuous improvement in production quality and significantly improving the accuracy and efficiency of enteral nutrition infusion set production testing.

[0021] Other features and advantages of this disclosure will be described in detail in the following detailed description section. Attached Figure Description

[0022] The accompanying drawings are provided to further illustrate the present disclosure and form part of the specification. They are used together with the following detailed description to explain the present disclosure, but do not constitute a limitation thereof. In the drawings:

[0023] Figure 1 This is a flowchart illustrating a testing method for manufacturing an enteral nutrition infusion set, as shown in the embodiments of the instruction manual.

[0024] Figure 2 This is a block diagram of a testing device for the production of an enteral nutrition infusion set, as shown in the embodiments of the instruction manual.

[0025] Figure 3 This is a block diagram illustrating a testing device for the production of enteral nutrition infusion sets, according to an embodiment of the specification. Detailed Implementation

[0026] 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.

[0027] The specific embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit this disclosure.

[0028] This disclosure provides a testing method for the production of enteral nutrition infusion sets. Figure 1 This is a flowchart illustrating a testing method for manufacturing an enteral nutrition infusion set according to one embodiment. Specifically, the method includes:

[0029] In step S11, during the manufacturing process of the enteral nutrition infusion set, a continuous image sequence of the surface area of ​​the enteral nutrition infusion set is acquired by an optical detection device deployed at a preset first detection station in a linear scanning manner, and a dynamic pressure data sequence of the enteral nutrition infusion set is acquired by a pressure sensor array deployed at a preset second detection station in a periodic pressure loading test.

[0030] In this embodiment of the disclosure, the continuous image sequence covers the complete surface texture information of the tube connection area, sealing ring installation area and interface adaptation area of ​​the enteral nutrition infusion device, and the dynamic pressure data sequence includes a first pressure fluctuation feature during the pressure rise phase, a second pressure fluctuation feature during the pressure stabilization phase and a third pressure fluctuation feature during the pressure release phase.

[0031] In the production process of enteral nutrition infusion sets, to effectively control product quality, it is necessary to obtain an initial set of related test data. This set consists of two important data parts: a continuous image sequence acquired by optical inspection equipment and a dynamic pressure data sequence recorded by a pressure sensor. These two types of data reflect the production quality status of the enteral nutrition infusion set from different perspectives. The continuous image sequence acquired by the optical inspection equipment can visually present the product's appearance and morphological characteristics, while the dynamic pressure data sequence recorded by the pressure sensor helps to detect key indicators such as the product's sealing performance. By comprehensively analyzing these two types of data, it is possible to comprehensively and accurately assess whether the enteral nutrition infusion set meets production standards.

[0032] A high-precision optical inspection device is installed at the first pre-set inspection station on the production line. This device features advanced imaging technology and linear scanning capabilities. When image acquisition begins, the enteral nutrition infusion set is transported to this inspection station and fixed on a specific inspection platform to ensure stable positioning during the scanning process. After the optical inspection device is activated, it performs a linear scan along the axial direction of the enteral nutrition infusion set. During the scan, the device's lens captures images at regular intervals, thus forming a continuous image sequence.

[0033] For the tubing connection area of ​​enteral nutrition infusion sets, optical inspection equipment focuses on capturing the surface texture information of this area. The tubing connection area is typically composed of tubing of different diameters connected in a specific manner, and its surface texture may include features such as weld marks and connection gaps. The high-resolution lens of the optical inspection equipment can clearly record these texture details, providing a basis for subsequent analysis of the quality of the tubing connection.

[0034] The area where the sealing ring is installed is also a key area for scanning. In this area, the optical inspection equipment focuses on information such as the installation position of the sealing ring, the shape of the sealing lip, and the fit between the sealing ring and surrounding components. By collecting and analyzing the surface texture of this area, it can be determined whether the sealing ring is installed correctly and whether there are any problems such as deformation or damage.

[0035] The interface adapter area also requires detailed scanning. This area involves the connection between the enteral nutrition infusion set and other devices or components, and its surface texture reflects key information such as the interface's dimensional accuracy and surface flatness. Optical inspection equipment will meticulously capture images of the interface's edges and internal contours to ensure the accuracy and reliability of the interface adapter.

[0036] A pressure sensor array is deployed at the pre-designed second detection station. This pressure sensor array consists of multiple high-precision pressure sensors, which are rationally distributed in different parts of the enteral nutrition infusion set to ensure comprehensive and accurate measurement of pressure changes inside the enteral nutrition infusion set.

[0037] Once the enteral nutrition infusion set is delivered to the second testing station, a periodic pressure loading test begins. The test process consists of three main phases: the pressure rise phase, the pressure stabilization phase, and the pressure release phase.

[0038] During the pressure build-up phase, pressure is slowly applied to the enteral nutrition infusion set via a specialized pressure loading device, while a pressure sensor array records pressure changes in real time. As the pressure gradually increases, the pressure sensors capture fluctuations during this process, reflecting the response characteristics of the enteral nutrition infusion set during the pressure build-up phase. For example, in the initial stage of pressure build-up, small pressure fluctuations may occur due to the compression and flow of gas within the tubing; these fluctuations are recorded, forming the first pressure fluctuation characteristic.

[0039] Once the pressure reaches a predetermined stable value, the system enters a pressure stabilization phase. During this phase, the pressure loading device maintains a constant pressure, and the pressure sensor array continuously monitors minute pressure changes. Due to factors such as the material properties and sealing performance of the enteral nutrition infusion set, pressure fluctuations will still occur even during the stabilization phase. The pressure sensors record these fluctuations, forming a second pressure fluctuation characteristic. These fluctuation characteristics reflect the sealing performance and structural stability of the enteral nutrition infusion set under stable pressure.

[0040] Finally, when the test enters the pressure release phase, the pressure loading device gradually reduces the pressure, and the pressure sensor array again records the fluctuation characteristics during the pressure drop, forming the third pressure fluctuation characteristic. During the pressure release process, the pressure inside the enteral nutrition infusion set drops rapidly, but due to factors such as the expansion of internal gas and the elastic deformation of components, the pressure drop process is also accompanied by some fluctuations. By analyzing the third pressure fluctuation characteristic, we can understand the performance of the enteral nutrition infusion set during pressure release.

[0041] In step S12, an initial detection data set is constructed based on the continuous image sequence and the dynamic pressure data sequence, and a hierarchical feature parsing operation is performed on the initial detection data set to generate a composite defect feature set containing morphological defect feature distribution information and sealing defect feature distribution information.

[0042] In this embodiment of the disclosure, an initial detection data set is constructed based on the continuous image sequence and the dynamic pressure data sequence. This can be achieved by performing timestamp alignment processing on the continuous image sequence and the dynamic pressure data sequence, so that each frame of image data in the continuous image sequence and the corresponding pressure data segment in the dynamic pressure data sequence form a synchronous mapping relationship.

[0043] After acquiring the continuous image sequence and the dynamic stress data sequence, timestamp alignment is required to establish a synchronization mapping between them. First, a timestamp is added to each data point in both the continuous image sequence and the dynamic stress data sequence. These timestamps record the acquisition time of each data point and are crucial for data alignment.

[0044] A precise matching algorithm is used for timestamp alignment. This algorithm traverses each frame of image data in a continuous image sequence to find the dynamic stress data segment with the closest timestamp. Specifically, for a given frame of image data in a continuous image sequence, the algorithm searches for the stress data point with the closest timestamp in the dynamic stress data sequence, and then extracts a stress data segment of appropriate length centered on that stress data point as the stress data segment corresponding to that frame of image data. In this way, each frame of image data in the continuous image sequence can form a synchronous mapping relationship with the corresponding stress data segment in the dynamic stress data sequence.

[0045] For example, during pressure testing of an enteral nutrition infusion set, when the optical detection device acquires an image at a certain moment, the pressure sensor also records the pressure data at that moment and for a period of time before and after. By aligning the timestamps, the image data at that moment can be associated with the corresponding pressure data segment. In this way, in subsequent analysis, both image and pressure information can be considered simultaneously, thus providing a more comprehensive assessment of the quality of the enteral nutrition infusion set.

[0046] The continuous image sequence and dynamic pressure data sequence after synchronous mapping are standardized and encoded to generate an initial detection dataset with a unified data format, which is then merged and stored in the cache module.

[0047] After completing the timestamp alignment of the continuous image sequence and the dynamic stress data sequence, the next step is to perform standardized encoding on these two types of data after synchronization and mapping to generate an initial detection dataset with a unified data format.

[0048] For a continuous image sequence, each frame of image data is first normalized. The purpose of normalization is to unify the pixel values ​​of the image data to a specific range, for example, normalizing pixel values ​​from 0-255 to 0-1. This can eliminate differences between different images caused by factors such as lighting and contrast, improving data consistency and comparability.

[0049] Next, the normalized image data is encoded. The encoding method can employ an algorithm suitable for image data processing, such as converting the image data into vector form, where each vector represents a feature of the image. Through encoding, a continuous sequence of images is transformed into a series of image vectors with a uniform format.

[0050] For dynamic pressure data sequences, normalization is also performed. The pressure data values ​​are normalized to a suitable range to ensure comparability between different pressure data points. Then, the normalized pressure data is encoded. Based on the characteristics of the pressure data, it can be converted into a suitable encoding format for analysis; for example, the pressure data can be divided into several data segments according to time sequence, and each data segment can be converted into a vector.

[0051] After standardizing and encoding the continuous image sequences and dynamic stress data sequences, they are combined into an initial detection dataset with a unified data format. This dataset contains encoded image vectors and stress vectors, and a synchronization mapping is established between them through timestamps.

[0052] Finally, the generated initial detection dataset is stored in the caching module. The caching module can use high-speed storage devices, such as memory or solid-state drives, to ensure fast data retrieval and processing. The initial detection dataset stored in the caching module provides the foundational data support for subsequent operations such as feature parsing and defect classification.

[0053] After obtaining the initial detection dataset in a unified data format, a hierarchical feature parsing operation needs to be performed to extract feature information related to defects in the enteral nutrition infusion device. The hierarchical feature parsing operation is divided into multiple levels, progressively extracting morphological defect feature distribution information and sealing defect feature distribution information from the initial detection data, and finally fusing these two types of information into a composite defect feature set.

[0054] In step S13, a preset defect classification model is invoked based on the composite defect feature set to generate defect type labeling results that indicate the failure mode category and corresponding defect spatial location information of the enteral nutrition infusion device.

[0055] In this embodiment of the disclosure, after obtaining a composite defect feature set that simultaneously labels morphological defects and sealing defects, a preset defect classification model needs to be invoked based on this set to generate defect type labeling results. The preset defect classification model is a trained artificial intelligence model that can accurately classify the defect type of the enteral nutrition infusion device based on the input composite defect feature set, and label the corresponding failure mode category and defect spatial location information.

[0056] In one possible implementation, the step of generating a defect type labeling result indicating the failure mode category and corresponding spatial location information of the enteral nutrition infusion device by calling a preset defect classification model based on the composite defect feature set includes:

[0057] The dimensional deviation features in the morphological defect feature distribution information are input into the first classification sub-model of the defect classification model to generate a first defect category label associated with the pipe body connection tolerance exceeding the standard.

[0058] In this embodiment of the disclosure, after generating the composite defect feature set, the dimensional deviation feature within the morphological defect feature distribution information is input into the first classification sub-model. The first classification sub-model is a key part of the preset defect classification model, specifically designed to handle defect classification related to pipe connection tolerances.

[0059] Dimensional deviation features encompass the dimensional deviations of the pipe connection region from standard values. These features are multi-dimensional, with each dimension representing dimensional deviation information for different parts of the pipe connection region. For example, they may include deviations in the pipe's inner diameter, outer diameter, length, and other aspects. When these dimensional deviation features are input into the first classification sub-model, the model analyzes them based on its internal training rules and logic.

[0060] The first sub-model is an artificial intelligence model trained on a large amount of sample data, with a multi-layered neural network structure. The input layer receives dimensional deviation feature data, and the neurons in the intermediate layers perform complex nonlinear transformations and feature extraction on this input data. By continuously learning and adjusting the weights between neurons, the model can identify pipe connection tolerance deviations corresponding to different dimensional deviation features.

[0061] After the model analyzes the input dimensional deviation features, it outputs a first defect category label associated with the pipe connection tolerance exceeding the standard. This label is not just a simple identifier; it also contains information about the specific circumstances of the pipe connection tolerance exceeding the standard, such as the location and degree of the exceedance.

[0062] The shape distortion features and surface crack features in the morphological defect feature distribution information are input into the second classification sub-model of the defect classification model to generate a second defect category label associated with injection molding anomalies.

[0063] In this embodiment, shape distortion features and surface crack features are extracted from the morphological defect feature distribution information and input into the second classification sub-model. Shape distortion features describe deviations from design standards in the enteral nutrition infusion set due to various factors during manufacturing. These features are also multi-dimensional and may involve different aspects such as bending, twisting, and deformation of the tube. Surface crack features refer to information related to cracks appearing on the surface of the enteral nutrition infusion set, including multiple dimensions such as crack location, length, and width.

[0064] The second classification sub-model is also part of the pre-defined defect classification model, focusing on identifying defects related to injection molding anomalies. This model also employs a multi-layer neural network structure, with the input layer receiving shape distortion features and surface crack features. In the intermediate layers, the model deeply mines and analyzes these features, adjusting the weights between neurons by learning from a large amount of injection molding anomaly sample data to accurately identify the injection molding anomalies corresponding to these features.

[0065] After analysis and processing by the model, a second defect category label associated with the injection molding anomaly will be generated. This label records detailed information about the injection molding anomaly, such as the anomaly type (whether it is shape distortion caused by improper temperature control or surface cracks caused by mold problems), the location of the anomaly, and other multi-dimensional content.

[0066] The pressure leak point location information in the sealing defect feature distribution information is input into the third classification sub-model of the defect classification model to generate a third defect category label associated with the sealing ring assembly failure.

[0067] In this embodiment, pressure leak location information is extracted from the sealing defect feature distribution information and input into the third classification sub-model. The pressure leak location information is a multi-dimensional dataset that accurately records the specific location information of pressure leaks on the enteral nutrition infusion set, which may include leak locations in different parts such as the tube connection area, the sealing ring installation area, and the interface adaptation area.

[0068] The third classification sub-model is specifically designed to handle classification defects related to seal assembly. It also features a multi-layer neural network structure, with the input layer receiving information about the location of pressure leaks. In the model's middle layers, through learning from a large amount of seal assembly failure sample data, neurons perform complex analysis and feature extraction on the input pressure leak location information. The model then identifies the correlation between pressure leaks at different locations and seal assembly failures.

[0069] Once the model completes its analysis, it generates a third defect category label associated with the seal assembly failure. This label contains detailed information about the seal assembly failure, such as the location of the failure (where the seal assembly went wrong), possible causes of failure (e.g., improper seal installation, seal damage), and other multi-dimensional information.

[0070] The leakage estimate from the sealing defect feature distribution information is input into the fourth classification sub-model of the defect classification model to generate a fourth defect category label associated with the wear of the interface sealing surface.

[0071] In this embodiment, the leakage estimate from the sealing defect feature distribution information is input into the fourth classification sub-model. The leakage estimate is a multi-dimensional numerical set that reflects the leakage at different parts of the enteral nutrition infusion set. This numerical set is obtained by performing pressure decay pattern recognition and analysis on the dynamic pressure data sequence, comprehensively considering the pressure changes during the pressure rise, pressure stabilization, and pressure release phases, thereby accurately estimating the leakage.

[0072] The fourth classification sub-model is a classification module for interface sealing surface wear defects within the pre-defined defect classification model. It features a multi-layer neural network structure, with the input layer receiving leakage estimate data. In the intermediate layers, the model performs in-depth analysis of this input data, adjusting the weights between neurons by learning from a large number of interface sealing surface wear sample data to identify the interface sealing surface wear conditions corresponding to different leakage estimates.

[0073] After processing by the model, a fourth defect category label associated with wear on the interface sealing surface will be generated. This label contains detailed information about the wear on the interface sealing surface, such as the location of the wear (which interface's sealing surface is worn), the degree of wear (the severity of the wear is judged based on the leakage estimate), and other multi-dimensional content.

[0074] Based on the first defect category label, the second defect category label, and the fourth defect category label, a first defect type labeling result is generated to indicate the failure mode category of the enteral nutrition infusion device, and based on the third defect category label, a second defect type labeling result is generated to indicate the spatial location information of the defect of the enteral nutrition infusion device.

[0075] In this embodiment of the disclosure, after obtaining the first defect category label, the second defect category label, and the fourth defect category label, these labels need to be logically integrated. Logical integration is a comprehensive analysis and evaluation process aimed at generating a defect type labeling result that includes priority ranking information and defect impact weight values.

[0076] First, for each defect category label, a defect impact weight value is assigned based on the degree to which the corresponding defect affects the performance and quality of the enteral nutrition infusion set. This weight value is determined based on extensive experimental data and empirical summaries, taking into account the potential consequences of different defects. For example, exceeding tolerances in the tubing connection may affect the product's installation and operational stability; abnormal injection molding may affect the product's appearance and structural strength; sealing ring assembly failure may lead to leakage problems; and wear on the interface sealing surface may reduce sealing performance.

[0077] Then, these defect category labels are prioritized based on their defect impact weight values. The priority ranking rule is that the higher the defect impact weight value, the higher the priority of the label. This ranking identifies which defects have the most critical impact on product quality and require priority handling.

[0078] The defect type labeling result is generated based on the first defect type labeling result and the second defect type labeling result.

[0079] In this embodiment, after logical integration, a first defect type labeling result is obtained. The detailed information of each defect category label (such as defect type, defect location, defect severity, etc.) is associated and integrated with the corresponding priority ranking information and defect impact weight value. Combining the first defect type labeling result and the second defect type labeling result, the final generated defect type labeling result is a comprehensive and detailed dataset. It clearly indicates the failure mode category of the enteral nutrition infusion device and the corresponding spatial location information of the defects, while also providing important information such as defect priority and impact weight.

[0080] In step S14, an adaptive adjustment operation of production parameters is performed based on the defect type labeling results to generate a set of process optimization parameters that match the failure mode category. The production line is then reactivated based on the set of process optimization parameters, and a secondary detection and verification operation is performed to generate optimization verification results.

[0081] In this embodiment of the disclosure, based on the previously generated defect type labeling results, an adaptive adjustment operation of production parameters needs to be performed to generate a set of process optimization parameters that match the failure mode category. The adaptive adjustment operation of production parameters is a process of automatically adjusting production parameters according to product defect conditions, with the aim of optimizing the production process and improving product quality.

[0082] In this embodiment of the disclosure, after generating a set of process optimization parameters that match the current production line configuration, it is necessary to reactivate the production line operation based on this set and perform secondary detection and verification operations to evaluate the effect of process optimization and generate optimization verification results.

[0083] In one possible implementation, the method further includes:

[0084] An adaptive mapping relationship is established between the defect type labeling results and the process optimization parameter set. When a new defect category is detected, the parameter optimization rule expansion operation is triggered.

[0085] To achieve continuous optimization of the production process, it is necessary to establish an adaptive mapping relationship between defect type labeling results and the set of process optimization parameters. This mapping relationship is based on a large amount of production data and experimental results, and it describes the correspondence between different types of defects and their corresponding process optimization parameters.

[0086] When establishing the mapping relationship, the defect type labeling results and process optimization parameter sets in the historical production data are first analyzed. For each defect type, the corresponding process optimization parameter adjustments are statistically analyzed. For example, when a defect occurs where the pipe connection tolerance exceeds the standard, the adjusted values ​​of the corresponding injection mold closing pressure parameter and cooling time parameter are recorded. When an injection molding abnormality defect occurs, the adjusted values ​​of the injection molding machine screw speed parameter and melt temperature parameter are recorded, etc.

[0087] By analyzing and statistically processing a large amount of historical data, the patterns and relationships between different defect types and process optimization parameters were summarized. These relationships were then stored in the system as mapping tables, forming an adaptive mapping relationship between defect type labeling results and the set of process optimization parameters.

[0088] During production, if a new type of defect is detected—that is, a defect type not seen in historical data—the system will trigger a parameter optimization rule expansion operation. First, a detailed analysis and feature extraction are performed on the new defect to determine its main characteristics and potential influencing factors. Then, based on these characteristics and influencing factors, combined with existing process knowledge and experience, preliminary predictions are made regarding process parameters that may need adjustment.

[0089] Next, the predicted process parameters were verified and adjusted through small-scale experimental production. During the experimental production, product quality was continuously monitored, and the process parameters were further optimized based on the experimental results. Finally, the new defect types and corresponding optimized process parameters were added to the adaptive mapping relationship, expanding the parameter optimization rules to enable timely responses to similar new defects in subsequent production.

[0090] The defect reduction rate in the optimization verification results is monitored in real time. When the defect reduction rate is lower than the preset threshold, the full parameter review process of the production line is initiated.

[0091] To ensure the continued effectiveness of process optimization, it is necessary to monitor the trend of defect reduction rate changes in the optimization verification results in real time. After each production batch completes the secondary inspection and verification operation, optimization verification results containing sets of morphological defect reduction rates and sealing defect reduction rates are generated. The system analyzes these defect reduction rate data in real time and observes their changing trends.

[0092] The preset threshold is a critical value set based on production quality requirements and historical production data. It represents the minimum acceptable level of defect reduction rate. If the defect reduction rate is lower than the preset threshold, it indicates that the process optimization is not effective, and there may be unreasonable settings of certain process parameters or new problems arising in the production process.

[0093] When the defect reduction rate is detected to be lower than a preset threshold, the system will initiate a full parameter review process for the production line. First, a comprehensive check and review of all process parameters of the production line will be conducted. This includes the mold closing pressure and cooling time parameters of the injection mold, the screw speed and melt temperature parameters of the injection molding machine, the positioning accuracy and clamping force parameters of the automatic assembly robot arm, and the grinding speed and feed rate parameters of the sealing surface polishing equipment.

[0094] For each process parameter, check whether its current setting value is consistent with the value in the process optimization parameter set and whether it meets the requirements of the production process. At the same time, analyze the interrelationships between process parameters to ensure that their combinations do not produce conflicts or unreasonable situations.

[0095] Then, each step of the production process is inspected to check for problems in equipment operation, raw material quality, and operator procedures. For example, it is checked whether the injection molding machine's heating system is working properly, whether the motion precision of the automatic assembly robot arm meets the requirements, and whether the physical and chemical properties of the raw materials are stable.

[0096] Based on the review results, any problems identified are promptly adjusted and rectified. If the process parameters are improperly set, they are readjusted. If there is equipment malfunction, timely repair and maintenance are performed. If there is a raw material quality issue, qualified raw materials are replaced. Through the full parameter review process of the production line, the reasonable setting of process parameters and the stable and reliable production process are ensured, thereby improving product quality and the effectiveness of process optimization.

[0097] Based on the improvement in pass rate shown in the optimization verification results report, the production batch size is dynamically adjusted, and a set of capacity optimization suggestion instructions is generated.

[0098] Based on the improvement in the pass rate shown in the optimization verification report, the production batch size can be dynamically adjusted to optimize production capacity. The improvement in the pass rate reflects the degree to which process optimization improves product quality. By reasonably adjusting the production batch size, production efficiency and capacity can be increased while ensuring product quality.

[0099] If the pass rate increases significantly, it indicates that process optimization has yielded remarkable results, and product quality has improved markedly. In this case, the production batch size can be appropriately increased. Increasing the production batch size can improve equipment utilization, reduce production preparation time and costs, thereby increasing capacity. The specific increase can be determined by comprehensively considering factors such as the magnitude of the pass rate improvement, equipment production capacity, and market demand. For example, if the pass rate improvement is substantial, the equipment still has significant production potential, and market demand is strong, the production batch size can be increased considerably.

[0100] If the improvement in the pass rate is small, it indicates that the effect of process optimization is limited and the improvement in product quality is not significant. In this case, it is necessary to carefully adjust the production batch size to avoid a decline in product quality due to blindly expanding production scale. It is advisable to maintain the current production batch size or appropriately increase it slightly, while further analyzing the problems existing in the process optimization process and seeking ways to improve product quality.

[0101] While dynamically adjusting production batch sizes, the system generates a set of capacity optimization suggestion instructions. This set includes a series of optimization suggestions and instructions for the production line, such as adjusting equipment operating parameters, optimizing production processes, and strengthening quality control. These suggestions and instructions are formulated based on the improvement in yield rate, adjustments to production batch sizes, and the actual operating conditions of the production line, aiming to further improve production line capacity and product quality.

[0102] Capacity optimization suggestion instructions can be presented to production managers in the form of reports. Based on these instructions, production managers can make corresponding adjustments and optimizations to the production line. By dynamically adjusting production batch sizes and implementing capacity optimization suggestion instructions, production line capacity can be optimized, thereby improving the company's production efficiency and market competitiveness.

[0103] By acquiring the initial test data set associated with enteral nutrition infusion sets during the manufacturing process and performing hierarchical feature analysis, the morphological and sealing defects of the enteral nutrition infusion sets can be comprehensively and accurately identified. Based on these composite defect features, a preset defect classification model is invoked to generate defect type labeling results, which not only indicate the failure mode category of the enteral nutrition infusion set but also provide corresponding defect spatial location information. By performing adaptive adjustment of production parameters, a set of process optimization parameters matching the failure mode category can be generated, directly guiding the adjustment and optimization of the production line, thereby effectively reducing the defect rate in the production process. Finally, by reactivating the production line and performing secondary test verification, optimized verification results can be generated, ensuring continuous improvement in production quality and significantly improving the accuracy and efficiency of enteral nutrition infusion set production testing.

[0104] In one possible implementation, step S14, which involves reactivating the production line operation based on the set of process optimization parameters and performing a secondary detection and verification operation to generate optimization verification results, includes:

[0105] In step S141, the set of process optimization parameters is sent to the production line control terminal, triggering the injection mold parameter adjustment module, injection molding machine parameter adjustment module, automatic assembly robotic arm control module, and sealing surface polishing equipment control module to synchronously perform parameter update operations.

[0106] In this embodiment, the generated set of process optimization parameters is sent to the production line control terminal via a data transmission system. The production line control terminal is the core control hub of the entire production line, responsible for receiving and processing various production parameter information and transmitting this information to the corresponding equipment control modules.

[0107] After receiving parameter information about the injection mold from the process optimization parameter set, the injection mold parameter adjustment module automatically adjusts the mold closing pressure and cooling time parameters. This module communicates with the injection mold's control system to write the new parameter values ​​into the mold's control program, ensuring that the mold operates according to the optimized parameters during subsequent production.

[0108] The injection molding machine parameter adjustment module adjusts the screw speed and melt temperature parameters of the injection molding machine based on the parameter information in the process optimization parameter set. This module connects to the injection molding machine's control system and modifies the parameter settings in the control program to enable the injection molding machine to operate with optimized parameters during production.

[0109] After receiving parameter information about the automated assembly robot arm from the process optimization parameter set, the automatic assembly robot arm control module adjusts the robot arm's positioning accuracy and clamping force parameters. This module communicates with the robot arm's motion control system to modify the robot arm's motion trajectory and force control program, ensuring that the robot arm operates according to the optimized parameters when assembling the sealing ring.

[0110] The control module of the sealing surface polishing equipment adjusts the grinding speed and feed rate parameters of the equipment based on the parameter information related to the sealing surface polishing equipment in the process optimization parameter set. This module is connected to the control system of the polishing equipment, and by modifying the parameter settings in the control program, the polishing equipment can operate with optimized parameters when polishing the interface sealing surface.

[0111] Through the coordination and control of the production line control terminal, the injection mold parameter adjustment module, injection molding machine parameter adjustment module, automatic assembly robotic arm control module, and sealing surface polishing equipment control module will synchronously perform parameter update operations to ensure that the entire production line can operate according to the optimized process parameters.

[0112] In step S142, after the parameter update operation is completed, the production line is restarted, and a full-coverage detection operation is performed on the subsequently produced enteral nutrition infusion sets to obtain optimized continuous image sequences and dynamic pressure data sequences.

[0113] In this embodiment, the production line is restarted after all equipment control modules have completed parameter update operations. Once started, the production line begins producing new enteral nutrition infusion sets. To evaluate the effectiveness of the process optimization, a full-coverage testing operation needs to be performed on subsequently produced enteral nutrition infusion sets.

[0114] The full-coverage inspection operation includes acquiring a continuous image sequence using optical inspection equipment and recording a dynamic pressure data sequence using pressure sensors. The optical inspection equipment performs a linear scan of the surface of the enteral nutrition infusion set at a preset first inspection station. During the scan, the optical inspection equipment moves along the axis of the enteral nutrition infusion set at a stable speed and accuracy, and its lens captures images at specific time intervals. Each captured image contains a portion of the enteral nutrition infusion set's surface. As the equipment moves, these images are gradually stitched together to form a continuous image sequence covering the complete surface texture information of the enteral nutrition infusion set's tube connection area, sealing ring installation area, and interface adapter area. Each frame of this continuous image sequence contains multi-dimensional information, such as pixel values ​​and color information, which can reflect the morphological characteristics of the enteral nutrition infusion set's surface in detail.

[0115] Simultaneously, at the pre-set second testing station, a pressure sensor array performs periodic pressure loading tests on the enteral nutrition infusion set. The pressure loading device applies pressure to the inside of the enteral nutrition infusion set according to a certain pattern, and the pressure sensors record the pressure changes inside the set in real time. During the pressure rise phase, the pressure loading device slowly increases the pressure, and the pressure sensors capture the pressure gradient characteristics during this process. These characteristics consist of a series of pressure values ​​and corresponding time points, reflecting the rate and fluctuation of pressure rise. Once the pressure reaches a predetermined stable value, the pressure stabilization phase begins. The pressure sensors continuously monitor minute changes in pressure, recording the steady-state pressure fluctuation amplitude characteristics, which reflect the range of pressure fluctuation during the stabilization phase. Finally, during the pressure release phase, the pressure loading device gradually reduces the pressure, and the pressure sensors record the pressure drop rate characteristics, which describe the speed and trend of pressure decrease. Through the recording of pressure data in these three phases, a dynamic pressure data sequence is formed. This sequence contains multi-dimensional pressure information, comprehensively reflecting the sealing performance of the enteral nutrition infusion set.

[0116] In step S143, real-time contour comparison processing is performed on the optimized continuous image sequence to generate optimized morphological defect feature distribution information.

[0117] In this embodiment of the disclosure, after obtaining the optimized continuous image sequence, real-time contour comparison processing needs to be performed on it to generate optimized morphological defect feature distribution information. Real-time contour comparison processing is an image analysis-based process designed to detect potential morphological defects in the enteral nutrition infusion device by comparing it with a standard contour template.

[0118] In step S144, pressure decay mode verification processing is performed on the optimized dynamic pressure data sequence to generate optimized sealing defect feature distribution information.

[0119] In this embodiment of the disclosure, after obtaining the optimized dynamic pressure data sequence, it is necessary to perform pressure decay mode verification processing on it to generate optimized sealing defect feature distribution information. Pressure decay mode verification processing is a pressure data analysis-based process designed to detect potential defects in the sealing performance of the enteral nutrition infusion set by comparing it with a standard pressure mode.

[0120] In step S145, the optimized morphological defect feature distribution information and the optimized sealing defect feature distribution information are compared and analyzed for defect density to generate an optimization verification result that includes defect reduction rate and pass rate improvement indicators.

[0121] In this embodiment of the disclosure, after obtaining the optimized morphological defect feature distribution information and the optimized sealing defect feature distribution information, it is necessary to perform a defect density comparison analysis on these two types of information to generate optimization verification results that include defect reduction rate and pass rate improvement indicators. Defect density comparison analysis is a process of comprehensively evaluating the effect of process optimization. By comparing the defect density before and after optimization, the degree to which process optimization improves product quality is measured.

[0122] In one possible implementation, step S143, which involves performing real-time contour comparison processing on the optimized continuous image sequence to generate optimized morphological defect feature distribution information, includes:

[0123] In step S1431, a high-speed imaging device is deployed in a preset real-time detection station to capture a sequence of surface contour images of the enteral nutrition infusion device in motion at a millisecond-level sampling frequency.

[0124] In this embodiment, a high-speed imaging device is deployed at a pre-defined real-time detection station. This device possesses high resolution and rapid imaging capabilities, enabling it to capture images of the surface of the enteral nutrition infusion set in motion at millisecond-level sampling frequencies. As the enteral nutrition infusion set moves along the production line, the high-speed imaging device continuously captures images of its surface, forming a sequence of surface contour images. Due to the extremely high sampling frequency, different states of the enteral nutrition infusion set surface are captured every millisecond; therefore, this sequence of surface contour images can very accurately record the surface morphological changes of the enteral nutrition infusion set during its movement.

[0125] In step S1432, the edge computing node is invoked to perform online contour extraction on the surface contour image sequence, generating real-time pipe connection area contour data, real-time sealing ring installation area contour data, and real-time interface adaptation area contour data.

[0126] In this embodiment, the acquired surface contour image sequence is transmitted to an edge computing node. An edge computing node is a device with powerful computing capabilities, enabling real-time data processing locally and reducing data transmission latency. On the edge computing node, a specialized image analysis algorithm performs online contour extraction on the surface contour image sequence. This algorithm first preprocesses each frame of the image, including noise reduction and contrast enhancement, to improve image quality. Then, edge detection technology is used to identify the edge information of each region of the enteral nutrition infusion device in the image. For the tube connection region, its edge contour is extracted to form real-time tube connection region contour data. For the sealing ring installation region, the contour of this region is extracted to obtain real-time sealing ring installation region contour data. Similarly, the contour of the interface adaptation region is extracted to generate real-time interface adaptation region contour data. This real-time contour data contains multi-dimensional information, such as the coordinate position and curvature of the contour, and can accurately describe the real-time contour morphology of each region of the enteral nutrition infusion device.

[0127] In step S1433, the real-time pipe connection area contour data, real-time sealing ring installation area contour data, and real-time interface adaptation area contour data are dynamically compared with the preset standard contour template to generate optimized morphological defect feature distribution information including real-time dimensional deviation, real-time shape distortion, and real-time surface crack length.

[0128] In this embodiment, the generated real-time tube connection area contour data, real-time sealing ring installation area contour data, and real-time interface adapter area contour data are dynamically compared with a preset standard contour template. The preset standard contour template is formulated according to the design requirements and quality standards of the enteral nutrition infusion set and represents the contour structure under ideal conditions. The dynamic comparison process is a point-by-point comparison process. For each point in the real-time contour data, it is compared with the corresponding point in the standard contour template. By comparing information such as coordinate position and curvature, the difference between the two is calculated.

[0129] For real-time pipe connection area contour data, the real-time dimensional deviation can be obtained by comparison, which is the difference between the actual size of the pipe connection area and the standard size in each direction. Simultaneously, the real-time shape distortion can also be calculated, reflecting the degree of deviation of the shape of the pipe connection area from the standard shape. Similarly, for real-time sealing ring installation area contour data and real-time interface adapter area contour data, the corresponding real-time dimensional deviation and real-time shape distortion can be calculated.

[0130] Furthermore, during the comparison process, if abnormal edge features are found in the real-time contour data, such as discontinuous edges or sudden changes in curvature, it may indicate the presence of surface cracks. By analyzing and measuring these abnormal edge features, the real-time surface crack length can be determined. The real-time dimensional deviation, real-time shape distortion, and real-time surface crack length are integrated to generate optimized morphological defect feature distribution information containing this information.

[0131] In step S1434, the optimized morphological defect feature distribution information is transmitted to the central processing unit for defect density statistics to generate optimized morphological defect feature distribution information.

[0132] In this embodiment, the generated optimized morphological defect feature distribution information is transmitted to the central processing unit (CPU). The CPU is the core control and computing center of the entire detection system, possessing powerful data analysis and processing capabilities. Within the CPU, defect density statistics are performed on the optimized morphological defect feature distribution information. Defect density statistics is a process of quantitatively analyzing the distribution of defects on the surface of the enteral nutrition infusion device. First, the surface of the enteral nutrition infusion device is divided into multiple small regions, and then the number of defects in each small region is counted. Based on the number of defects, the defect density of each small region can be calculated.

[0133] Next, based on the calculated defect density, a real-time defect distribution heatmap is generated. This heatmap is a visual representation that uses different colors to indicate different defect densities. For example, darker colors indicate higher defect density, and lighter colors indicate lower defect density. The real-time defect distribution heatmap provides a clear view of the distribution of defects on the surface of the enteral nutrition infusion set, offering strong support for further analysis and defect processing.

[0134] In one possible implementation, step S144, which involves performing pressure decay mode verification processing on the optimized dynamic pressure data sequence to generate optimized sealing defect feature distribution information, includes:

[0135] In step S1441, a stepped pressure test protocol is loaded in the preset airtightness test station, and the enteral nutrition infusion device is subjected to increasing pressure according to the preset pressure gradient, and the dynamic response data of each pressure stage is recorded.

[0136] In this embodiment, a stepped pressure testing protocol is applied at a preset airtightness testing station. This protocol specifies the process of applying increasing pressure to the enteral nutrition infusion set according to a preset pressure gradient. The pressure loading device gradually increases the pressure applied inside the enteral nutrition infusion set according to the protocol requirements. Each pressure increase is the preset pressure gradient. At each pressure stage, the pressure loading device maintains a stable pressure for a period of time so that the pressure sensor can record the dynamic response data of the enteral nutrition infusion set under that pressure.

[0137] Dynamic response data includes pressure changes during the pressure rise phase, pressure fluctuations during the pressure stabilization phase, and pressure decreases during the pressure release phase. Pressure sensors record this data in real time, forming a dynamic response data sequence for each pressure phase. This data sequence contains multi-dimensional information, such as pressure values ​​and time points, and can reflect in detail the sealing performance of the enteral nutrition infusion set at different pressure phases.

[0138] In step S1442, the real-time pressure rise curve, real-time pressure steady-state fluctuation curve, and real-time pressure release curve are extracted from the dynamic response data.

[0139] In this embodiment of the disclosure, real-time pressure rise curve, real-time pressure steady-state fluctuation curve, and real-time pressure release curve are extracted from the recorded dynamic response data. The real-time pressure rise curve describes the change in internal pressure of the enteral nutrition infusion set over time during the pressure rise phase. By filtering and organizing the data from the pressure rise phase in the dynamic response data, the real-time pressure rise curve is plotted with time as the horizontal axis and pressure as the vertical axis.

[0140] The real-time pressure steady-state fluctuation curve reflects the minute pressure fluctuations inside the enteral nutrition infusion set during the pressure stabilization phase. During this phase, the pressure values ​​recorded by the pressure sensor fluctuate within a small range. These fluctuation data are extracted and analyzed to plot the real-time pressure steady-state fluctuation curve.

[0141] The real-time pressure release curve represents the decrease in internal pressure of the enteral nutrition infusion set over time during the pressure release phase. Similarly, real-time pressure release curves are plotted by processing the pressure release phase data from the dynamic response data. These three curves contain multiple data points, each representing a pressure value at a specific time point, comprehensively reflecting the pressure change characteristics of the enteral nutrition infusion set at different pressure phases.

[0142] In step S1443, the slope similarity of the real-time pressure rise curve and the preset standard pressure rise template is calculated to generate the real-time pressure gradient matching degree.

[0143] In this embodiment, the extracted real-time pressure rise curve is compared with a preset standard pressure rise template for slope similarity calculation. The preset standard pressure rise template is formulated based on the design requirements and normal sealing performance of the enteral nutrition infusion set, representing the pressure rise curve under ideal conditions. Slope similarity calculation is a process of measuring the degree of similarity between the slopes of two curves.

[0144] First, the real-time pressure rise curve and the standard pressure rise template are segmented into multiple small intervals. Within each small interval, the slopes of the real-time pressure rise curve and the standard pressure rise template are calculated. Then, the slopes of the two curves within each small interval are compared, and their similarity is calculated. The similarity can be calculated by taking the absolute value of the difference in slopes and then normalizing it. Finally, the similarities between each cell are combined to obtain the real-time pressure gradient matching degree. The real-time pressure gradient matching degree reflects the similarity between the real-time pressure rise curve and the standard pressure rise template in terms of slope; it is a multi-dimensional numerical set, with each dimension representing the degree of matching between different cells.

[0145] In step S1444, the amplitude difference analysis is performed between the real-time pressure steady-state fluctuation curve and the preset standard pressure steady-state template to generate the real-time pressure fluctuation deviation.

[0146] In this embodiment, an amplitude difference analysis is performed between the real-time pressure steady-state fluctuation curve and a preset standard pressure steady-state template. The preset standard pressure steady-state template is the pressure fluctuation curve during the stable phase under ideal conditions, which defines the normal fluctuation range of pressure during the stable phase. Amplitude difference analysis is a process of comparing the amplitude differences between the two curves.

[0147] First, the amplitude ranges of the real-time pressure steady-state fluctuation curve and the standard pressure steady-state template are determined. Amplitude refers to the difference between the maximum and minimum pressure values ​​within one cycle of the curve. Then, the amplitudes of the real-time pressure steady-state fluctuation curve and the standard pressure steady-state template are compared, and their differences are calculated. By statistically analyzing the amplitude differences over multiple cycles, the real-time pressure fluctuation deviation is obtained. The real-time pressure fluctuation deviation reflects the degree of difference between the real-time pressure steady-state fluctuation curve and the standard pressure steady-state template in terms of amplitude; it is also a multi-dimensional numerical set, with each dimension representing the degree of deviation for different cycles.

[0148] In step S1445, the real-time pressure release curve is compared with the preset standard pressure release template in terms of decay rate to generate a real-time pressure drop rate deviation value.

[0149] In this embodiment, the real-time pressure release curve is compared with a preset standard pressure release template in terms of decay rate. The preset standard pressure release template is designed based on the normal sealing performance of the enteral nutrition infusion set and represents the pressure release curve under ideal conditions. The decay rate comparison is a process of comparing the pressure drop rates of the two curves.

[0150] First, derivatives are calculated for the real-time pressure release curve and the standard pressure release template to obtain their pressure drop rates at each time point. Then, the pressure drop rates of the real-time pressure release curve and the standard pressure release template are compared at the same time points, and their differences are calculated. By statistically analyzing the differences at multiple time points, the real-time pressure drop rate deviation value is obtained. The real-time pressure drop rate deviation value reflects the degree of difference between the real-time pressure release curve and the standard pressure release template in terms of pressure drop rate; it is a multi-dimensional numerical set, with each dimension representing the degree of deviation at different time points.

[0151] In step S1446, a three-dimensional sealing performance evaluation matrix is ​​constructed based on the real-time pressure gradient matching degree, real-time pressure fluctuation deviation degree, and real-time pressure drop rate deviation value, and optimized sealing defect feature distribution information with marked potential leak point locations and leak risk levels is generated.

[0152] In this embodiment of the disclosure, a three-dimensional sealing performance evaluation matrix is ​​constructed based on the calculated real-time pressure gradient matching degree, real-time pressure fluctuation deviation degree, and real-time pressure drop rate deviation value. The three-dimensional sealing performance evaluation matrix is ​​a matrix composed of three dimensions: real-time pressure gradient matching degree, real-time pressure fluctuation deviation degree, and real-time pressure drop rate deviation value. The value of each dimension represents the degree of difference between the corresponding pressure characteristics and the standard template.

[0153] By analyzing the three-dimensional sealing performance evaluation matrix, the potential leakage points and leakage risk levels of the enteral nutrition infusion set can be determined. If the value at a certain position in the matrix exceeds a preset threshold, it indicates that there may be a sealing defect at that position, which is a potential leakage point. Based on the degree to which the value exceeds the threshold, different leakage risk levels can be classified, such as high risk, medium risk, and low risk.

[0154] By integrating the locations of potential leak points and their risk levels, optimized sealing defect feature distribution information is generated, annotated with this information. This distribution information records in detail the sealing performance defects of the enteral nutrition infusion set.

[0155] In one possible implementation, in step S145, the step of comparing the optimized morphological defect feature distribution information with the optimized sealing defect feature distribution information to generate an optimization verification result including defect reduction rate and pass rate improvement indicators includes:

[0156] In step S1451, the historical morphological defect feature distribution information and historical sealing defect feature distribution information before adjustment are extracted from the historical database.

[0157] In this embodiment, historical morphological defect feature distribution information and historical sealing defect feature distribution information before adjustment are extracted from a historical database. The historical database stores various test data from the production process of enteral nutrition infusion sets, recording product defects before process parameter adjustments. Through database query functions, historical morphological defect feature distribution information and historical sealing defect feature distribution information related to the current production batch are extracted. This historical information includes the defect distribution of the enteral nutrition infusion set in terms of morphology and sealing performance before adjustment, providing basic data for subsequent comparative analysis.

[0158] In step S1452, the ratio of the number of each type of defect in the optimized morphological defect feature distribution information to the corresponding number of defects in the historical morphological defect feature distribution information is calculated to generate a set of morphological defect reduction rates.

[0159] In this embodiment, the number of various types of defects in the optimized morphological defect feature distribution information and the historical morphological defect feature distribution information are statistically analyzed. The optimized morphological defect feature distribution information includes various defect information such as real-time dimensional deviation, real-time shape distortion, and real-time surface crack length. By classifying and statistically analyzing this information, the number of each type of defect can be obtained. Similarly, the number of each type of defect in the historical morphological defect feature distribution information is statistically analyzed.

[0160] Then, the ratio of the number of each type of defect in the optimized morphological defect feature distribution information to the corresponding number of defects in the historical morphological defect feature distribution information is calculated. For each type of defect, the optimized number of defects is divided by the historical number of defects to obtain the reduction rate of that type of defect. The reduction rates of all types of defects are integrated to generate a set of morphological defect reduction rates. This set contains information in multiple dimensions, each dimension representing the reduction rate of a type of defect, which intuitively reflects the degree of improvement of morphological defects by process optimization.

[0161] In step S1453, the ratio of the number of various leakage risk levels in the optimized sealing defect feature distribution information to the number of corresponding leakage risk levels in the historical sealing defect feature distribution information is calculated to generate a sealing defect reduction rate set.

[0162] In this embodiment of the disclosure, the number of various leakage risk levels in the optimized sealing defect feature distribution information and the historical sealing defect feature distribution information is statistically analyzed. The optimized sealing defect feature distribution information marks the locations of potential leakage points and their leakage risk levels. By statistically analyzing the number of potential leakage points at different leakage risk levels, the number of each type of leakage risk level is obtained. Similarly, the number of each type of leakage risk level in the historical sealing defect feature distribution information is statistically analyzed.

[0163] Next, the ratio of the number of each leakage risk level in the optimized sealing defect feature distribution information to the number of corresponding leakage risk levels in the historical sealing defect feature distribution information is calculated. For each leakage risk level, the optimized number is divided by the historical number to obtain the reduction rate for that leakage risk level. The reduction rates of all leakage risk levels are integrated to generate a set of sealing defect reduction rates. This set contains information in multiple dimensions, each representing the reduction rate of a leakage risk level, reflecting the degree of improvement in sealing defects due to process optimization.

[0164] In step S1454, the optimized morphological defect feature distribution information and the optimized sealing defect feature distribution information are checked for compliance item by item according to the preset qualification judgment rules, the number of qualified products is counted and the improvement rate of qualification rate is calculated.

[0165] In this embodiment, the optimized morphological defect feature distribution information and the optimized sealing defect feature distribution information are checked for compliance item by item according to preset compliance judgment rules. The compliance judgment rules are formulated based on the quality standards of enteral nutrition infusion sets, which specify the allowable range and compliance conditions for various defects. Information such as real-time dimensional deviation, real-time shape distortion, and real-time surface crack length in the optimized morphological defect feature distribution information, as well as information such as potential leak point locations and leak risk levels in the optimized sealing defect feature distribution information, are checked one by one according to the compliance judgment rules.

[0166] If all defect information for a product meets the requirements of the conformity assessment rules, the product is deemed a conformity product. Compliance verification is performed on all products, and the number of conformity products is counted. Then, the improvement rate is calculated. The improvement rate refers to the percentage increase in the number of conformity products after process optimization relative to the number of conformity products before process optimization. To calculate the improvement rate, the number of conformity products before process optimization needs to be obtained from the historical database. Assuming the number of conformity products before process optimization is A, and the number of conformity products after process optimization is B, the calculation process for the improvement rate is as follows: First, calculate the increase in the number of conformity products, i.e., BA. Then, divide the increase by the number of conformity products before process optimization, A, to obtain a ratio. Finally, convert this ratio into a percentage form to obtain the improvement rate. This percentage value directly reflects the effect of process optimization on improving the product conformity rate.

[0167] In step S1455, the set of morphological defect reduction rates, the set of sealing defect reduction rates, and the improvement in pass rate are visualized and packaged to generate a multi-dimensional optimization verification result report.

[0168] In this embodiment of the disclosure, after obtaining the set of morphological defect reduction rates, the set of sealing defect reduction rates, and the improvement in pass rate, these data need to be visualized and packaged to generate a multi-dimensional optimization verification result report. Visualization and packaging is a process of presenting data in an intuitive and easy-to-understand way, helping production managers quickly understand the effects of process optimization.

[0169] First, for the set of morphological defect reduction rates, the reduction rate data for each type of defect is organized. A bar chart can be used to represent the reduction rate of each type of defect using bars of different heights. The height of the bar represents the degree of reduction for that type of defect. By comparing the heights of different bars, the improvement effect of process optimization on various morphological defects can be clearly seen.

[0170] The set of reduction rates for sealing defects can also be visualized using bar charts or line graphs. If a bar chart is used, the reduction rate for different leakage risk levels is represented by bars. If a line graph is used, the reduction rates for different leakage risk levels are connected to form a line, which provides a clearer view of the changing trends in the reduction rates for different leakage risk levels.

[0171] The improvement in the pass rate is presented using a pie chart or a percentage. If a pie chart is used, the proportion of qualified products before and after process optimization can be represented by sectors of different sizes. By comparing the sizes of the sectors, the improvement in the pass rate can be visually reflected. If a percentage is used, the improvement in the pass rate is directly displayed as a percentage in the report.

[0172] By integrating these visualizations and data, and adding necessary titles, labels, and explanatory information, a multi-dimensional optimization verification results report is generated. This report not only includes key information such as the reduction rate of morphological defects, the reduction rate of sealing defects, and the improvement in pass rate, but also presents this information intuitively through visualization, providing strong support for production managers to evaluate the effectiveness of process optimization and formulate subsequent production strategies.

[0173] In one possible implementation, step S14, which involves performing an adaptive adjustment operation on production parameters based on the defect type labeling results to generate a set of process optimization parameters matching the failure mode category, includes:

[0174] In step S1401, when the defect type labeling result indicates that the pipe connection tolerance exceeds the standard, the mold closing pressure parameter and cooling time parameter of the injection mold are adjusted according to the distribution range of the dimensional deviation characteristics to generate a first process optimization parameter subset.

[0175] In this embodiment of the disclosure, when the defect type labeling result indicates that the pipe connection tolerance exceeds the standard, it is necessary to adjust the relevant parameters of the injection mold according to the distribution range of the dimensional deviation characteristics in the morphological defect feature distribution information. The distribution range of the dimensional deviation characteristics reflects the dimensional deviation of the pipe connection area in different parts. By analyzing this distribution range, it is possible to determine which parts of the injection mold need to be adjusted in the production process regarding pressure and cooling time.

[0176] The clamping pressure parameters of the injection mold have a significant impact on the molded dimensions of the tube. If the dimensional deviation characteristics indicate that certain parts of the tube are too large, it may be necessary to appropriately reduce the clamping pressure to decrease the pressure exerted by the mold on the plastic material, thereby bringing the tube dimensions closer to the standard value. If the dimensional deviation characteristics indicate that certain parts of the tube are too small, it may be necessary to appropriately increase the clamping pressure to allow the plastic material to more fully fill the mold cavity and achieve the standard dimensional requirements.

[0177] Cooling time also affects the molded dimensions of the tube. Excessive cooling time causes excessive shrinkage of the plastic material, potentially resulting in a smaller tube size. Insufficient cooling time prevents the plastic material from fully cooling and solidifying, potentially leading to dimensional instability in the tube. Therefore, the cooling time parameter should be adjusted appropriately based on the distribution range of dimensional deviation characteristics to ensure dimensional stability of the tube during the cooling process.

[0178] By adjusting the mold closing pressure and cooling time parameters of the injection mold, a first subset of optimized process parameters is generated. This first subset of optimized process parameters includes the injection mold parameters optimized to address the issue of excessive tolerances in the pipe connection.

[0179] In step S1402, when the defect type labeling result indicates that there is an injection molding abnormality, the screw speed parameters and melt temperature parameters of the injection molding machine are adjusted according to the distortion direction of the shape distortion characteristics to generate a second subset of process optimization parameters.

[0180] In this embodiment of the disclosure, if the defect type labeling result indicates an injection molding anomaly, the relevant parameters of the injection molding machine need to be adjusted according to the distortion direction of the shape distortion features in the morphological defect feature distribution information. The distortion direction of the shape distortion features reflects the specific direction in which the shape of the enteral nutrition infusion set deviates from the design standard during the injection molding process, such as the bending direction or twisting direction of the tube.

[0181] The screw speed of an injection molding machine affects the delivery speed and mixing uniformity of the plastic material. If shape distortion characteristics indicate that the tube is bent or twisted in a certain direction, it may be necessary to adjust the screw speed to make the flow rate and distribution of the plastic material in the mold more uniform, thereby reducing shape distortion. For example, if the tube is bent to one side, the screw speed may be appropriately reduced to allow the plastic material to fill the mold cavity more slowly, reducing shape distortion caused by uneven flow.

[0182] Melt temperature parameters have a significant impact on the flowability and molding properties of plastic materials. If shape distortion characteristics indicate deformation in certain areas of the tube, the melt temperature may need to be adjusted. Higher melt temperatures enhance the flowability of the plastic material but may lead to increased shrinkage after molding. Lower melt temperatures reduce the flowability of the plastic material but may result in more stable dimensions after molding. Therefore, adjusting the melt temperature parameters appropriately based on the direction of shape distortion characteristics can improve injection molding quality.

[0183] By adjusting the screw speed and melt temperature parameters of the injection molding machine, a second subset of process optimization parameters is generated. This second subset of process optimization parameters includes injection molding machine parameters optimized to address injection molding anomalies.

[0184] In step S1403, when the defect type labeling result indicates that there is a sealing ring assembly failure, the positioning accuracy parameters and clamping force parameters of the automatic assembly robot arm are adjusted according to the spatial coordinates of the pressure leakage point location information to generate a third process optimization parameter subset.

[0185] In this embodiment of the disclosure, when the defect type labeling result indicates a sealing ring assembly failure, the relevant parameters of the automated assembly robot arm need to be adjusted based on the spatial coordinates of the pressure leak point location information in the sealing defect feature distribution information. The spatial coordinates of the pressure leak point location information accurately record the specific location of the pressure leak point on the enteral nutrition infusion set. By analyzing these coordinates, it is possible to determine where the sealing ring encountered problems during the assembly process.

[0186] The positioning accuracy parameters of an automated assembly robot arm determine its ability to accurately install a sealing ring into a designated position. If the pressure leak location information indicates that a sealing ring is not properly installed at a certain location, it may be necessary to improve the positioning accuracy of the automated assembly robot arm at that location to ensure that the sealing ring can be accurately installed in the correct position.

[0187] The clamping force parameter affects the deformation of the sealing ring during assembly. Excessive clamping force may damage the sealing ring, while insufficient clamping force may result in an insecure installation. Therefore, the clamping force parameter of the automated assembly robot arm should be adjusted appropriately based on the spatial coordinates of the pressure leak location to ensure that the sealing ring is neither damaged nor securely installed during assembly.

[0188] By adjusting the positioning accuracy and clamping force parameters of the automated assembly robot arm, a third subset of optimized process parameters is generated. This third subset includes the automated assembly robot arm parameters optimized to address the seal assembly failure issue.

[0189] In step S1404, when the defect type labeling result indicates that there is wear on the interface sealing surface, the grinding speed parameters and feed parameters of the sealing surface polishing equipment are adjusted according to the numerical range of the leakage estimate to generate a fourth process optimization parameter subset.

[0190] In this embodiment of the disclosure, if the defect type labeling result indicates wear on the interface sealing surface, the relevant parameters of the sealing surface polishing equipment need to be adjusted based on the numerical range of the leakage estimate in the sealing defect feature distribution information. The numerical range of the leakage estimate reflects the severity of the interface sealing surface wear. By analyzing this range, the grinding speed and feed rate required by the sealing surface polishing equipment during the polishing process can be determined.

[0191] The grinding speed parameters of the sealing surface polishing equipment affect the efficiency and quality of polishing. If the leakage estimate indicates severe wear on the interface sealing surface, it may be necessary to appropriately increase the grinding speed to remove the worn surface layer more quickly. If the leakage estimate indicates mild wear on the interface sealing surface, it may be necessary to appropriately decrease the grinding speed to avoid over-grinding and a decrease in the precision of the sealing surface.

[0192] The feed rate parameter determines the distance the polishing equipment advances with each pass during the polishing process. If the estimated leakage indicates uneven wear on the interface sealing surface, the feed rate parameter may need to be adjusted to allow the polishing equipment to polish the worn areas more precisely.

[0193] By adjusting the grinding speed and feed parameters of the sealing surface polishing equipment, a fourth subset of optimized process parameters is generated. This fourth subset of optimized process parameters includes the parameters of the sealing surface polishing equipment optimized for the wear problem of the interface sealing surface.

[0194] In step S1405, the first process optimization parameter subset, the second process optimization parameter subset, the third process optimization parameter subset, and the fourth process optimization parameter subset are subjected to collaborative optimization processing to generate a set of process optimization parameters that matches the current production line configuration.

[0195] In this embodiment of the disclosure, after obtaining the first subset of process optimization parameters, the second subset of process optimization parameters, the third subset of process optimization parameters, and the fourth subset of process optimization parameters, it is necessary to perform collaborative optimization processing on these subsets. Collaborative optimization processing is a process that comprehensively considers the interrelationships and influences between the various subsets, with the aim of generating a set of process optimization parameters that matches the current production line configuration.

[0196] First, the interrelationships between the various subsets of process optimization parameters are analyzed. For example, the clamping pressure and cooling time parameters of an injection mold may affect the melt temperature and screw speed of the injection molding machine, because different clamping pressures and cooling times result in different states of the plastic material within the mold, thus affecting the injection molding machine's handling of the plastic material. Similarly, the positioning accuracy and clamping force parameters of an automated assembly robot arm may be interrelated with the parameters of the injection mold and the injection molding machine, as the quality of the injection molding affects the assembly difficulty and effectiveness of the sealing ring.

[0197] Then, based on the current production line configuration, the subset of process optimization parameters are adjusted and optimized. The current production line configuration includes factors such as equipment performance and production flow arrangement. For example, if the injection molding machine on the production line is relatively old, it may be necessary to make more conservative adjustments to the screw speed and melt temperature parameters of the injection molding machine to adapt to the performance of the equipment.

[0198] During collaborative optimization, the optimization objectives and constraints of each subset of process optimization parameters are comprehensively considered. By continuously adjusting and balancing these parameters, a set of process optimization parameters that matches the current production line configuration is ultimately generated. This set of process optimization parameters includes all production parameters optimized for different failure modes.

[0199] In one possible implementation, step S12, performing a hierarchical feature parsing operation on the initial detection data set to generate a composite defect feature set containing morphological defect feature distribution information and sealing defect feature distribution information, includes:

[0200] In step S121, a multi-scale contour extraction operation is performed on the continuous image sequence to generate a first contour topology of the tube connection area of ​​the enteral nutrition infusion device, a second contour topology of the sealing ring installation area, and a third contour topology of the interface adaptation area.

[0201] In this embodiment of the disclosure, a multi-scale contour extraction operation is performed on a continuous image sequence stored in a cache module. Multi-scale contour extraction is an image analysis-based technique that can extract the contours of target objects in an image at different scales.

[0202] First, each frame in the continuous image sequence is preprocessed, including noise reduction and contrast enhancement, to improve image quality and clarity. Then, a multi-scale analysis method is used to decompose the preprocessed image at different scales. At each scale, an edge detection algorithm is used to extract edge information from the image; this edge information constitutes the contours of objects.

[0203] For the tube connection region of the enteral nutrition infusion set, the edge features of this region are the primary focus during multi-scale contour extraction. By analyzing contour information at different scales, the first contour topology of the tube connection region can be obtained. This first contour topology reflects the shape, size, and connection method of the tube connection region.

[0204] For the sealing ring mounting area, a multi-scale contour extraction operation is also performed. Since the shape and size of the sealing ring mounting area are crucial to sealing performance, it is necessary to accurately extract the contour information of this area. Through multi-scale analysis, subtle features of the sealing ring mounting area can be captured, generating a second contour topology.

[0205] For interface adaptation regions, multi-scale contour extraction can help detect whether the shape and size of the interface meet design requirements. By analyzing contour information at different scales, a third contour topology is generated, which reflects the contour features and connection accuracy of the interface adaptation region.

[0206] In step S122, the first contour topology, the second contour topology, and the third contour topology are compared pixel by pixel with a preset standard contour template to generate morphological defect feature distribution information including size deviation features, shape distortion features, and surface crack features.

[0207] After obtaining the first contour topology of the tube connection area, the second contour topology of the sealing ring installation area, and the third contour topology of the interface adaptation area, they are compared pixel by pixel with a preset standard contour template. The preset standard contour template is formulated according to the design requirements and quality standards of the enteral nutrition infusion set and represents the contour structure under ideal conditions.

[0208] Pixel-by-pixel alignment is a precise image comparison method that compares each pixel in the first, second, and third contour topologies with its corresponding pixel in a standard contour template. By comparing information such as pixel position and grayscale value, differences between the two can be identified.

[0209] If, during the comparison process, a deviation is found between the pixel position of a certain area and the standard contour template, then it can be determined that there is a dimensional deviation feature in that area. Dimensional deviation features reflect the difference between the size of the enteral nutrition infusion set and the design requirements, which may affect the installation and performance of the product.

[0210] If the outline shape of a certain area does not conform to the standard outline template, exhibiting distortion or deformation, then it can be determined that the area has shape distortion characteristics. Shape distortion characteristics may be caused by process problems or material properties during production, and will adversely affect the appearance and performance of the product.

[0211] Furthermore, during pixel-by-pixel comparison, if abnormal changes in pixel grayscale values ​​are detected in a certain area, exhibiting crack-like characteristics, then the presence of surface cracks in that area can be identified. Surface cracks are a serious morphological defect that can reduce the structural strength of enteral nutrition infusion sets and even lead to problems such as leakage.

[0212] By performing pixel-by-pixel comparison processing, the detected dimensional deviation features, shape distortion features, and surface crack features are integrated to generate morphological defect feature distribution information containing these features. This distribution information records in detail the morphological defects of each region of the enteral nutrition infusion set.

[0213] In step S123, a pressure decay pattern recognition operation is performed on the dynamic pressure data sequence to extract the pressure gradient features during the pressure rise phase, the pressure steady-state fluctuation amplitude features during the pressure stabilization phase, and the pressure decrease rate features during the pressure release phase.

[0214] Pressure decay pattern recognition is performed on dynamic pressure data sequences to extract key features related to pressure changes. This operation, based on the analysis and processing of pressure data, uses specific algorithms and models to identify different stages and characteristics within the pressure data.

[0215] First, the dynamic pressure data sequence is segmented into three phases: pressure rise, pressure stabilization, and pressure release. During the pressure rise phase, a slope calculation method is used to extract pressure gradient features. The pressure gradient features reflect the rate of pressure change during the rise and can be obtained by calculating the ratio of the difference between adjacent pressure data points to the time interval.

[0216] During the pressure stabilization phase, the focus is on extracting the steady-state pressure fluctuation amplitude characteristic. This characteristic describes the pressure fluctuation during the steady-state phase and can be obtained by calculating the difference between the maximum and minimum pressure values ​​during this phase. This feature reflects the sealing performance and structural stability of the enteral nutrition infusion set under stable pressure.

[0217] During the pressure release phase, a differential calculation method is used to extract the pressure drop rate characteristic. The pressure drop rate characteristic represents the speed at which pressure decreases during the release process, and it can be obtained by calculating the ratio of the difference between adjacent pressure data points to the time interval. The pressure drop rate characteristic reflects the performance of the enteral nutrition infusion set during pressure release and is of great significance for detecting sealing defects.

[0218] By identifying pressure decay patterns in dynamic pressure data sequences, the pressure gradient characteristics during the pressure rise phase, the steady-state fluctuation amplitude characteristics during the pressure stabilization phase, and the pressure decrease rate characteristics during the pressure release phase were successfully extracted.

[0219] In step S124, the pressure gradient characteristics, pressure steady-state fluctuation amplitude characteristics, and pressure drop rate characteristics are matched with preset sealing performance benchmark parameters to generate sealing defect feature distribution information that includes pressure leak point location information and leakage amount estimation.

[0220] After extracting the pressure gradient characteristics, steady-state pressure fluctuation amplitude characteristics, and pressure drop rate characteristics, these characteristics were compared with preset sealing performance benchmark parameters for trend matching analysis. The preset sealing performance benchmark parameters were formulated based on the design requirements and quality standards of the enteral nutrition infusion set, representing the pressure change trend and characteristics under normal conditions.

[0221] Trend matching analysis is a method based on data comparison and pattern recognition. It compares the pressure gradient characteristics, pressure steady-state fluctuation amplitude characteristics, and pressure drop rate characteristics with the sealing performance benchmark parameters one by one to analyze the similarities and differences between them.

[0222] If a significant deviation is found between the pressure characteristics at a certain stage and the baseline sealing performance parameters during the comparison process, it can be determined that a sealing defect may exist at that stage. For example, if the pressure gradient characteristics during the pressure rise stage are significantly higher than the baseline parameters, it may indicate a gas leak during the pressure rise. If the steady-state pressure fluctuation amplitude characteristics during the pressure stabilization stage exceed the baseline range, it may indicate poor sealing performance of the enteral nutrition infusion set. If the pressure drop rate characteristics during the pressure release stage differ significantly from the baseline parameters, it may suggest the presence of a leak.

[0223] To further determine the location of the pressure leak and estimate the leakage amount, an analysis method based on a pressure distribution model was employed. This method establishes a pressure distribution model based on pressure data collected by a pressure sensor array and a structural model of the enteral nutrition infusion set. By analyzing the pressure changes within the pressure distribution model, the possible location of the pressure leak can be inferred.

[0224] Simultaneously, based on the degree of deviation between the pressure characteristics and the baseline parameters, and combined with the pressure distribution model and related physical principles, the magnitude of the leakage can be estimated. The location information of the pressure leak point and the estimated leakage value are integrated to generate sealing defect feature distribution information containing this information. This distribution information records in detail the sealing defects of the enteral nutrition infusion set.

[0225] In step S125, the spatial coordinate fusion processing is performed on the distribution information of the morphological defects and the distribution information of the sealing defects to generate a composite defect feature set that simultaneously labels the morphological defects and sealing defects.

[0226] After obtaining the distribution information of morphological defects and the distribution information of sealing defects, it is necessary to perform spatial coordinate fusion processing on these two types of information to generate a composite defect feature set that simultaneously labels morphological defects and sealing defects.

[0227] First, spatial coordinate information is added to each defect feature in the morphological defect feature distribution information and the sealing defect feature distribution information. This spatial coordinate information records the specific location of each defect feature on the enteral nutrition infusion device and is the key basis for realizing spatial coordinate fusion.

[0228] Then, a fusion algorithm based on spatial coordinate matching is used to fuse the distribution information of morphological defects and the distribution information of sealing defects. This algorithm iterates through each defect feature in both the morphological and sealing defect distribution information to find their correspondence in spatial coordinates. If both morphological and sealing defects exist at a certain location, these two defect features are associated and merged.

[0229] During the integration process, defect characteristics will be comprehensively evaluated and labeled. For example, for locations where both morphological and sealing defects exist, the defects at that location will be comprehensively rated and labeled based on their severity and impact on product performance.

[0230] By using spatial coordinate fusion processing, the distribution information of morphological defects and sealing defects is merged into a unified composite defect feature set. This composite defect feature set simultaneously labels both morphological and sealing defects on the enteral nutrition infusion set.

[0231] This disclosure also provides a testing device for the production of enteral nutrition infusion sets, see [link to relevant documentation]. Figure 2 As shown, the device includes:

[0232] The acquisition module 210 is configured to acquire, during the manufacturing process of the enteral nutrition infusion set, a continuous image sequence of the surface area of ​​the enteral nutrition infusion set by means of an optical detection device deployed at a preset first detection station in a linear scanning manner, and to acquire the dynamic pressure data sequence of the enteral nutrition infusion set by means of a pressure sensor array deployed at a preset second detection station in a periodic pressure loading test.

[0233] The first generation module 220 is configured to construct an initial detection data set based on the continuous image sequence and the dynamic pressure data sequence, and perform a hierarchical feature parsing operation on the initial detection data set to generate a composite defect feature set containing morphological defect feature distribution information and sealing defect feature distribution information.

[0234] The second generation module 230 is configured to call a preset defect classification model based on the composite defect feature set to generate defect type labeling results that indicate the failure mode category and corresponding defect spatial location information of the enteral nutrition infusion device.

[0235] The third generation module 240 is configured to perform an adaptive adjustment operation of production parameters based on the defect type labeling results, generate a set of process optimization parameters that match the failure mode category, reactivate the production line operation according to the set of process optimization parameters, and perform a secondary detection and verification operation to generate optimization verification results.

[0236] In one possible implementation, the third generation module 240 is configured as follows:

[0237] The set of process optimization parameters is sent to the production line control terminal, triggering the injection mold parameter adjustment module, injection machine parameter adjustment module, automatic assembly robotic arm control module and sealing surface polishing equipment control module to simultaneously perform parameter update operations.

[0238] After the parameter update operation is completed, the production line is restarted, and a full-coverage detection operation is performed on the subsequently produced enteral nutrition infusion sets to obtain optimized continuous image sequences and dynamic pressure data sequences.

[0239] Real-time contour comparison processing is performed on the optimized continuous image sequence to generate optimized morphological defect feature distribution information;

[0240] The optimized dynamic pressure data sequence is subjected to pressure decay mode verification processing to generate optimized sealing defect feature distribution information.

[0241] The optimized morphological defect feature distribution information and the optimized sealing defect feature distribution information are compared and analyzed for defect density to generate optimization verification results that include defect reduction rate and pass rate improvement indicators.

[0242] In one possible implementation, the real-time contour comparison processing performed on the optimized continuous image sequence to generate optimized morphological defect feature distribution information includes:

[0243] A high-speed imaging device is deployed in a preset real-time detection station to capture a sequence of surface contour images of the enteral nutrition infusion device in motion at a millisecond-level sampling frequency.

[0244] The edge computing node is invoked to perform online contour extraction on the surface contour image sequence, generating real-time pipe connection area contour data, real-time sealing ring installation area contour data, and real-time interface adaptation area contour data.

[0245] The real-time pipe body connection area contour data, real-time sealing ring installation area contour data, and real-time interface adaptation area contour data are dynamically compared with the preset standard contour template to generate optimized morphological defect feature distribution information including real-time size deviation, real-time shape distortion, and real-time surface crack length.

[0246] The optimized morphological defect feature distribution information is transmitted to the central processing unit for defect density statistics, generating optimized morphological defect feature distribution information.

[0247] In one possible implementation, the third generation module 240 is configured as follows:

[0248] A stepped pressure test protocol is loaded in the preset airtightness test station, and the enteral nutrition infusion device is subjected to increasing pressure according to the preset pressure gradient, and the dynamic response data of each pressure stage is recorded.

[0249] Extract the real-time pressure rise curve, real-time pressure steady-state fluctuation curve, and real-time pressure release curve from the dynamic response data.

[0250] The slope similarity of the real-time pressure rise curve and the preset standard pressure rise template is calculated to generate the real-time pressure gradient matching degree.

[0251] The amplitude difference analysis is performed between the real-time pressure steady-state fluctuation curve and the preset standard pressure steady-state template to generate the real-time pressure fluctuation deviation.

[0252] The real-time pressure release curve is compared with the preset standard pressure release template to generate a real-time pressure drop rate deviation value.

[0253] Based on the real-time pressure gradient matching degree, real-time pressure fluctuation deviation degree, and real-time pressure drop rate deviation value, a three-dimensional sealing performance evaluation matrix is ​​constructed, and optimized sealing defect feature distribution information with marked potential leakage point locations and leakage risk levels is generated.

[0254] In one possible implementation, the third generation module 240 is configured as follows:

[0255] Extract the historical morphological defect feature distribution information and historical sealing defect feature information from the historical database before adjustment;

[0256] Calculate the ratio of the number of each type of defect in the optimized morphological defect feature distribution information to the corresponding number of defects in the historical morphological defect feature distribution information, and generate a set of morphological defect reduction rates;

[0257] Calculate the ratio of the number of each type of leakage risk level in the optimized sealing defect feature distribution information to the number of corresponding leakage risk levels in the historical sealing defect feature distribution information, and generate a set of sealing defect reduction rates;

[0258] According to the preset qualification judgment rules, the optimized morphological defect feature distribution information and the optimized sealing defect feature distribution information are checked for compliance item by item, the number of qualified products is counted and the improvement of the qualification rate is calculated.

[0259] The set of reduction rates for morphological defects, the set of reduction rates for sealing defects, and the improvement in pass rate are visualized and encapsulated to generate a multi-dimensional optimization verification result report.

[0260] In one possible implementation, the third generation module 240 is configured as follows:

[0261] When the defect type labeling result indicates that the pipe connection tolerance exceeds the standard, the mold closing pressure parameter and cooling time parameter of the injection mold are adjusted according to the distribution range of the dimensional deviation characteristics to generate a first process optimization parameter subset;

[0262] When the defect type labeling result indicates that there is an injection molding abnormality, the screw speed parameters and melt temperature parameters of the injection molding machine are adjusted according to the distortion direction of the shape distortion characteristics to generate a second set of process optimization parameters.

[0263] When the defect type labeling result indicates that there is a sealing ring assembly failure, the positioning accuracy parameters and clamping force parameters of the automatic assembly robot arm are adjusted according to the spatial coordinates of the pressure leakage point location information to generate a third process optimization parameter subset.

[0264] When the defect type labeling result indicates that there is wear on the interface sealing surface, the grinding speed parameters and feed parameters of the sealing surface polishing equipment are adjusted according to the numerical range of the leakage estimate to generate a fourth process optimization parameter subset.

[0265] The first, second, third, and fourth subsets of process optimization parameters are collaboratively optimized to generate a set of process optimization parameters that matches the current production line configuration.

[0266] In one possible implementation, the first generation module 220 is configured as follows:

[0267] Perform multi-scale contour extraction on the continuous image sequence to generate a first contour topology of the tube connection area of ​​the enteral nutrition infusion device, a second contour topology of the sealing ring installation area, and a third contour topology of the interface adaptation area.

[0268] The first contour topology, the second contour topology, and the third contour topology are compared pixel by pixel with a preset standard contour template to generate morphological defect feature distribution information including size deviation features, shape distortion features, and surface crack features.

[0269] Perform pressure decay pattern recognition on the dynamic pressure data sequence to extract pressure gradient features during the pressure rise phase, pressure steady-state fluctuation amplitude features during the pressure stabilization phase, and pressure decrease rate features during the pressure release phase.

[0270] The pressure gradient characteristics, pressure steady-state fluctuation amplitude characteristics, and pressure drop rate characteristics are matched with preset sealing performance benchmark parameters to generate sealing defect feature distribution information that includes pressure leak point location information and leakage amount estimation.

[0271] The spatial coordinate fusion processing of the morphological defect feature distribution information and the sealing defect feature distribution information generates a composite defect feature set that simultaneously labels morphological defects and sealing defects.

[0272] This disclosure also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described in the foregoing embodiments.

[0273] This disclosure also provides an electronic device, including:

[0274] A memory on which computer programs are stored;

[0275] A processor for executing the computer program in the memory to implement the steps of any of the methods described in the foregoing embodiments.

[0276] Figure 3 The enteral nutrition infusion set production testing device 100 shown includes a processor 1001 and a memory 1003. The processor 1001 and memory 1003 are connected, for example, via a bus 1002. Optionally, the enteral nutrition infusion set production testing device 100 may further include a communication component 1004, which can be used for data interaction between the device 100 and other devices, such as data transmission and / or data reception. It should be noted that in actual operation, the communication component 1004 is not limited to one, and the structure of this enteral nutrition infusion set production testing device 100 does not constitute a limitation on the embodiments of this application.

[0277] Processor 1001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 1001 may also be a combination that implements computing functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0278] Bus 1002 may include a pathway for transmitting information between the aforementioned components. Bus 1002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 1002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0279] The memory 1003 may be ROM (Read Only Memory) or other types of static storage devices capable of storing static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices capable of storing information and instructions, or EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media, other magnetic storage devices, or any other medium capable of carrying or storing program code and capable of being read by a computer, without limitation herein.

[0280] The memory 1003 is used to store program code for executing embodiments of the present disclosure, and its execution is controlled by the processor 1001. The processor 1001 is used to execute the program code stored in the memory 1003 to implement the steps shown in the aforementioned embodiments of the detection method for producing enteral nutrition infusion sets.

[0281] This disclosure also provides a computer-readable storage medium storing program code, which, when executed by a processor, can implement the steps and corresponding content of the aforementioned detection method embodiment for producing enteral nutrition infusion devices.

[0282] The preferred embodiments of the present disclosure have been described in detail above with reference to the accompanying drawings. However, the present disclosure is not limited to the specific details of the above embodiments. Within the scope of the technical concept of the present disclosure, various changes, modifications, substitutions and variations can be made to these embodiments, and all such changes, modifications, substitutions and variations fall within the protection scope of the present disclosure.

[0283] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction, and such combinations should also be considered as part of this disclosure. To avoid unnecessary repetition, this disclosure will not further describe the various possible combinations. The technical scope of this application is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for detecting a defect in the production of an enteral nutrition infusion set, characterized in that The method includes: During the manufacturing process of the enteral nutrition infusion set, a continuous image sequence of the surface area of ​​the enteral nutrition infusion set is acquired by an optical inspection device deployed at a preset first inspection station in a linear scanning manner, and a dynamic pressure data sequence of the enteral nutrition infusion set is acquired by a pressure sensor array deployed at a preset second inspection station in a periodic pressure loading test. Based on the continuous image sequence and the dynamic pressure data sequence, an initial detection data set is constructed, and a hierarchical feature parsing operation is performed on the initial detection data set to generate a composite defect feature set containing morphological defect feature distribution information and sealing defect feature distribution information. Based on the composite defect feature set, a preset defect classification model is invoked to generate defect type labeling results that indicate the failure mode category and corresponding defect spatial location information of the enteral nutrition infusion device; Based on the defect type labeling results, an adaptive adjustment operation of production parameters is performed to generate a set of process optimization parameters that match the failure mode category. The production line is then reactivated based on the set of process optimization parameters, and a secondary detection and verification operation is performed to generate optimization verification results. The step of performing a hierarchical feature parsing operation on the initial detection data set to generate a composite defect feature set containing morphological defect feature distribution information and sealing defect feature distribution information includes: Perform multi-scale contour extraction on the continuous image sequence to generate the first contour topology of the tube connection area of ​​the enteral nutrition infusion device, the second contour topology of the sealing ring installation area, and the third contour topology of the interface adaptation area. The first contour topology, the second contour topology, and the third contour topology are compared pixel by pixel with a preset standard contour template to generate morphological defect feature distribution information including size deviation features, shape distortion features, and surface crack features. Performing a pressure decay pattern recognition operation on the dynamic pressure data sequence includes: segmenting the dynamic pressure data sequence into a pressure rise phase, a pressure stabilization phase, and a pressure release phase; extracting pressure gradient features using a slope calculation method during the pressure rise phase; extracting pressure steady-state fluctuation amplitude features during the pressure stabilization phase; and extracting pressure drop rate features during the pressure release phase using a differential calculation method. The pressure gradient characteristics, pressure steady-state fluctuation amplitude characteristics, and pressure drop rate characteristics are matched with preset sealing performance benchmark parameters to generate sealing defect feature distribution information that includes pressure leak point location information and leakage amount estimation. The spatial coordinate fusion processing of the morphological defect feature distribution information and the sealing defect feature distribution information generates a composite defect feature set that simultaneously labels morphological defects and sealing defects.

2. The method according to claim 1, wherein The step of reactivating the production line operation based on the set of process optimization parameters and performing a secondary detection and verification operation to generate optimization verification results includes: The set of process optimization parameters is sent to the production line control terminal, triggering the injection mold parameter adjustment module, injection machine parameter adjustment module, automatic assembly robotic arm control module and sealing surface polishing equipment control module to simultaneously perform parameter update operations. After the parameter update operation is completed, the production line is restarted, and a full-coverage detection operation is performed on the subsequently produced enteral nutrition infusion sets to obtain optimized continuous image sequences and dynamic pressure data sequences. Real-time contour comparison processing is performed on the optimized continuous image sequence to generate optimized morphological defect feature distribution information; The optimized dynamic pressure data sequence is subjected to pressure decay mode verification processing to generate optimized sealing defect feature distribution information; the optimized morphological defect feature distribution information and the optimized sealing defect feature distribution information are compared and analyzed for defect density to generate optimization verification results including defect reduction rate and pass rate improvement indicators.

3. The method according to claim 2, wherein The step of performing real-time contour comparison processing on the optimized continuous image sequence to generate optimized morphological defect feature distribution information includes: A high-speed imaging device is deployed in a preset real-time detection station to capture a sequence of surface contour images of the enteral nutrition infusion device in motion at a millisecond-level sampling frequency. The edge computing node is invoked to perform online contour extraction on the surface contour image sequence, generating real-time pipe connection area contour data, real-time sealing ring installation area contour data, and real-time interface adaptation area contour data. The real-time pipe body connection area contour data, real-time sealing ring installation area contour data, and real-time interface adaptation area contour data are dynamically compared with the preset standard contour template to generate optimized morphological defect feature distribution information including real-time size deviation, real-time shape distortion, and real-time surface crack length. The optimized morphological defect feature distribution information is transmitted to the central processing unit for defect density statistics, generating optimized morphological defect feature distribution information.

4. The method according to claim 2, wherein The optimized dynamic pressure data sequence is subjected to pressure decay mode verification processing to generate optimized sealing defect feature distribution information, including: A stepped pressure test protocol is loaded in the preset airtightness test station, and the enteral nutrition infusion device is subjected to increasing pressure according to the preset pressure gradient, and the dynamic response data of each pressure stage is recorded. Extract the real-time pressure rise curve, real-time pressure steady-state fluctuation curve, and real-time pressure release curve from the dynamic response data. The slope similarity of the real-time pressure rise curve and the preset standard pressure rise template is calculated to generate the real-time pressure gradient matching degree. The amplitude difference analysis is performed between the real-time pressure steady-state fluctuation curve and the preset standard pressure steady-state template to generate the real-time pressure fluctuation deviation. The real-time pressure release curve is compared with the preset standard pressure release template to generate a real-time pressure drop rate deviation value. Based on the real-time pressure gradient matching degree, real-time pressure fluctuation deviation degree, and real-time pressure drop rate deviation value, a three-dimensional sealing performance evaluation matrix is ​​constructed, and optimized sealing defect feature distribution information with marked potential leakage point locations and leakage risk levels is generated.

5. The testing method for producing enteral nutrition infusion sets according to claim 2, characterized in that, The step involves comparing the optimized morphological defect feature distribution information with the optimized sealing defect feature distribution information to generate optimization verification results that include defect reduction rate and pass rate improvement indicators, including: Extract the historical morphological defect feature distribution information and historical sealing defect feature information from the historical database before adjustment; Calculate the ratio of the number of each type of defect in the optimized morphological defect feature distribution information to the corresponding number of defects in the historical morphological defect feature distribution information, and generate a set of morphological defect reduction rates; Calculate the ratio of the number of each type of leakage risk level in the optimized sealing defect feature distribution information to the number of corresponding leakage risk levels in the historical sealing defect feature distribution information, and generate a set of sealing defect reduction rates; According to the preset qualification judgment rules, the optimized morphological defect feature distribution information and the optimized sealing defect feature distribution information are checked for compliance item by item, the number of qualified products is counted and the improvement of the qualification rate is calculated. The set of reduction rates for morphological defects, the set of reduction rates for sealing defects, and the improvement in pass rate are visualized and encapsulated to generate a multi-dimensional optimization verification result report.

6. The testing method for producing an enteral nutrition infusion set according to any one of claims 1-5, characterized in that, The step of performing adaptive adjustment of production parameters based on the defect type labeling results to generate a set of process optimization parameters matching the failure mode category includes: When the defect type labeling result indicates that the pipe connection tolerance exceeds the standard, the mold closing pressure parameter and cooling time parameter of the injection mold are adjusted according to the distribution range of the dimensional deviation characteristics to generate a first process optimization parameter subset; When the defect type labeling result indicates that there is an injection molding abnormality, the screw speed parameters and melt temperature parameters of the injection molding machine are adjusted according to the distortion direction of the shape distortion characteristics to generate a second set of process optimization parameters. When the defect type labeling result indicates that there is a sealing ring assembly failure, the positioning accuracy parameters and clamping force parameters of the automatic assembly robot arm are adjusted according to the spatial coordinates of the pressure leakage point location information to generate a third process optimization parameter subset. When the defect type labeling result indicates that there is wear on the interface sealing surface, the grinding speed parameters and feed parameters of the sealing surface polishing equipment are adjusted according to the numerical range of the leakage estimate to generate a fourth process optimization parameter subset. The first, second, third, and fourth subsets of process optimization parameters are collaboratively optimized to generate a set of process optimization parameters that matches the current production line configuration.

7. A testing device for the production of enteral nutrition infusion sets, characterized in that, The device includes: The acquisition module is configured to acquire, during the manufacturing process of the enteral nutrition infusion set, a continuous image sequence of the surface area of ​​the enteral nutrition infusion set by means of an optical detection device deployed at a preset first detection station in a linear scanning manner, and to acquire the dynamic pressure data sequence of the enteral nutrition infusion set by means of a pressure sensor array deployed at a preset second detection station in a periodic pressure loading test. The first generation module is configured to construct an initial detection data set based on the continuous image sequence and the dynamic pressure data sequence, and perform a hierarchical feature parsing operation on the initial detection data set to generate a composite defect feature set containing morphological defect feature distribution information and sealing defect feature distribution information. The second generation module is configured to call a preset defect classification model based on the composite defect feature set to generate defect type labeling results that indicate the failure mode category and corresponding defect spatial location information of the enteral nutrition infusion device. The third generation module is configured to perform adaptive adjustment of production parameters based on the defect type labeling results, generate a set of process optimization parameters that match the failure mode category, reactivate the production line operation according to the set of process optimization parameters, and perform secondary detection and verification operations to generate optimization verification results. The first generation module is configured as follows: Perform multi-scale contour extraction on the continuous image sequence to generate the first contour topology of the tube connection area of ​​the enteral nutrition infusion device, the second contour topology of the sealing ring installation area, and the third contour topology of the interface adaptation area. The first contour topology, the second contour topology, and the third contour topology are compared pixel by pixel with a preset standard contour template to generate morphological defect feature distribution information including size deviation features, shape distortion features, and surface crack features. Performing a pressure decay pattern recognition operation on the dynamic pressure data sequence includes: segmenting the dynamic pressure data sequence into a pressure rise phase, a pressure stabilization phase, and a pressure release phase; extracting pressure gradient features using a slope calculation method during the pressure rise phase; extracting pressure steady-state fluctuation amplitude features during the pressure stabilization phase; and extracting pressure drop rate features during the pressure release phase using a differential calculation method. The pressure gradient characteristics, pressure steady-state fluctuation amplitude characteristics, and pressure drop rate characteristics are matched with preset sealing performance benchmark parameters to generate sealing defect feature distribution information that includes pressure leak point location information and leakage amount estimation. The spatial coordinate fusion processing of the morphological defect feature distribution information and the sealing defect feature distribution information generates a composite defect feature set that simultaneously labels morphological defects and sealing defects.

8. 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 described in any one of claims 1-6.

9. An electronic device, characterized in that, include: A memory on which computer programs are stored; A processor for executing the computer program in the memory to implement the steps of the method according to any one of claims 1-6.

Images