Commodity full-life-cycle management service platform based on electromechanical industry supply chain

Abstract

The invention discloses a commodity whole life cycle management service platform based on an electromechanical industry supply chain, and relates to the technical field of electromechanical industry life cycle management. Comprising an image acquisition and analysis module, a format feature construction module, a complexity prediction modeling module, a rhythm adaptive regulation and control module, an information decoding and acquisition module and a data binding and filing module. And the image acquisition and analysis module is used for carrying out real-time pre-shooting on the two-dimensional code image about to enter the identification window through industrial-grade visual equipment deployed at the front end of the identification area of the high-speed conveyor belt, and obtaining the structural format information of the two-dimensional code by analyzing the two-dimensional code image. Through a conveyor belt speed regulation and control mechanism driven by complexity scoring, two-dimensional code recognition rhythm self-adaptive matching is realized, the high-complexity two-dimensional code recognition accuracy and the low-complexity two-dimensional code recognition efficiency are improved, complete collection and binding of equipment life cycle data are ensured, and the intelligent and digital level of supply chain management is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of life cycle management of the electromechanical industry, and in particular to a commodity full life cycle management service platform based on the electromechanical industry supply chain. Background Art

[0002] The product lifecycle management service platform, based on the electromechanical industry supply chain, is a comprehensive service system that integrates information technology and industry chain management concepts. It aims to integrate the entire product lifecycle management process, from design and development, manufacturing, circulation, sales, use, to scrapping and recycling. By integrating technologies such as the Internet of Things, big data, and cloud computing, the platform collects and analyzes product operational data at each stage of the lifecycle in real time, supporting enterprises in achieving precision production, inventory optimization, quality traceability, repair and maintenance forecasting, and green recycling. It also strengthens information collaboration and resource sharing across the supply chain, improving supply chain transparency and responsiveness, thereby driving the electromechanical industry's transformation toward intelligent manufacturing and sustainable development.

[0003] When managing the entire life cycle of goods based on the supply chain of the electromechanical industry, QR codes, as the primary form of identification code, have become the preferred choice for information-intensive equipment management due to their large encoding capacity, high reading efficiency, encryption, and fault tolerance. Their core function is to give each electromechanical product or component a unique and identifiable "digital label," enabling efficient encoding and rapid reading of information. QR codes can not only store key basic information such as the equipment's model, production batch, manufacturer, and installation time, but can also dynamically link to cloud databases, updating core data such as maintenance records, usage status, and operating parameters in real time, fully supporting accurate traceability, intelligent operation and maintenance, and remote monitoring. At the same time, the encryption and fault-tolerance mechanisms of QR codes can also effectively prevent data from being tampered with or forged, and are an important technical support for ensuring the transparency, security, and traceability of information throughout the equipment's life cycle.

[0004] The existing technology has the following deficiencies:

[0005] In high-speed industrial scenarios, such as high-speed conveyor belts that automatically identify and read QR codes on electromechanical equipment, the purpose is to achieve real-time identity confirmation and information collection of electromechanical products or components in a high-beat production environment, facilitate rapid acquisition of the equipment's unique identification code, and associate or record key lifecycle data such as its production batch, model, assembly status, quality inspection results, operation records, and maintenance information to support subsequent intelligent sorting, quality control, process traceability, automatic assembly, and systematic management, ensuring the digitalization and automation of equipment management throughout its entire lifecycle.

[0006] However, in actual applications, due to the complexity differences in the QR code formats within the system, some devices use a long sequence encryption format, while others use a simplified encoding format. The recognition system lacks the ability to dynamically adapt to the parsing time of QR codes of varying complexity, leading to mismatched recognition times during high-speed recognition. This problem can easily cause the system to skip high-complexity QR codes or misread low-complexity QR codes, which in turn leads to inaccurate or erroneous collection of some device information, resulting in a break in the lifecycle data chain. This not only affects equipment quality traceability and maintenance decisions, but also significantly impacts the operational efficiency and management accuracy of the entire supply chain, making it one of the key technical issues that urgently need to be addressed in the full lifecycle management of products in the electromechanical industry supply chain.

[0007] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0008] The purpose of the present invention is to provide a commodity life cycle management service platform based on the supply chain of the electromechanical industry. By pre-shooting and analyzing the structural format information of the QR code image, combining feature engineering with the complexity score output by the convolutional neural network model, the conveyor belt speed is dynamically adjusted to achieve recognition rhythm matching, thereby ensuring that high-complexity QR codes have sufficient recognition time and low-complexity QR codes can pass at high speed. Ultimately, the accurate binding of QR code recognition results and equipment life cycle data is achieved, ensuring the complete collection and traceability of the full life cycle information of electromechanical products, not only improving the stability and beat efficiency of the recognition system, but also promoting the efficient collaboration of supply chain management towards digitalization and intelligence, so as to solve the problems in the above-mentioned background technology.

[0009] To achieve the above objectives, the present invention provides the following technical solutions: a commodity lifecycle management service platform based on the supply chain of the electromechanical industry, comprising an image acquisition and analysis module, a format feature construction module, a complexity prediction modeling module, a rhythm adaptive control module, an information decoding and acquisition module, and a data binding and archiving module;

[0010] The image acquisition and analysis module uses industrial-grade visual equipment (such as a high-speed CMOS camera) deployed at the front end of the recognition area on the high-speed conveyor belt to pre-shoot the QR code image that is about to enter the recognition window in real time, and obtains the structural format information of the QR code by analyzing the QR code image;

[0011] The format feature construction module constructs an independent format evaluation set for the structural format information collected from each QR code, applies feature engineering technology to extract key indicators reflecting the difficulty of QR code recognition, analyzes the extracted indicators, and constructs a profile for identifying the complexity characteristics of the QR code based on the analyzed features;

[0012] The complexity prediction modeling module uses the features extracted from each format evaluation set as feature vectors and inputs them into a pre-trained convolutional neural network model. The convolutional neural network model then outputs a complexity score to intelligently predict the format complexity of the QR code.

[0013] The rhythm adaptive control module sends speed adjustment instructions to the conveyor belt scheduling unit. Based on the complexity score output by the convolutional neural network model, it dynamically adjusts the conveyor belt speed to adapt to the time requirements of decoding, forming a rhythm matching. Specifically, for high-complexity QR codes, the conveyor belt automatically slows down within the recognition window (for example, by 100 to 300 milliseconds) to ensure that the recognition system has sufficient time to complete data analysis. For low-complexity QR codes, the conveyor belt maintains normal speed to improve overall beat efficiency.

[0014] The information decoding and acquisition module, after the conveyor belt speed is adjusted and enters the rhythm matching state, uses an information reading device (such as a high-speed barcode scanner, industrial camera + decoding unit) to decode the QR code officially sent into the identification area, extract the unique identification code of the product from the QR code, and record the associated life cycle information, including key data such as equipment model, batch, quality inspection status, maintenance records, etc.

[0015] The data binding and archiving module, after successful QR code recognition, matches and binds the collected unique identification code with multi-dimensional data such as the current production rhythm information, work station data, equipment code, etc., and records it in the life cycle data platform.

[0016] Preferably, the structural format information of the QR code is obtained by parsing the QR code image, and the specific steps are as follows:

[0017] First, the QR code is identified and located within the image. QR codes typically contain specific positioning patterns (such as the "positioning detection patterns" at the three corners of a QR code or the "L-shaped boundary" of a DataMatrix). The parsing system uses these patterns to quickly determine the spatial position and orientation of the QR code within the image. This step determines the boundaries of the QR code's coding area and the sampling reference, providing a position reference and angle correction basis for subsequent structural decoding.

[0018] Based on the positioning pattern, the system divides the coded area into several small blocks (modules) according to the standard version rules of the QR code, forming a module matrix (such as 21×21, 25×25, etc.). The dimensions of this matrix directly reflect the "version number" of the QR code. For example, QR codes go from version 1 (21×21) to version 40 (177×177). By reading the black and white states of these modules point by point, the system constructs the "code surface structure diagram" of the entire QR code and counts structural parameters such as the number of modules and the proportion of the area occupied by the error correction code.

[0019] After the module matrix is ​​identified, the system parses the dedicated areas containing format information based on QR code encoding standards (such as ISO / IEC 18004), such as the Format Information area and the Version Information area. These areas store core structural parameters such as the QR code's error correction level (such as L, M, Q, H), mask mode, version number, and encoding method (such as byte mode and numeric mode). By parsing these fields, the system can accurately extract the QR code's structural format information.

[0020] Preferably, feature engineering technology is applied to extract key indicators reflecting the difficulty of QR code recognition from the format evaluation set, wherein the extracted indicators include the proportion of redundant data caused by the error correction level adopted by the QR code and the degree of disturbance caused to the overall pattern of the code surface by the mask mode used by the QR code. After analyzing the extracted indicators, the error correction redundancy factor and the mask complexity factor are generated respectively, and a portrait for identifying the complexity characteristics of the QR code is constructed through the error correction redundancy factor and the mask complexity factor.

[0021] Preferably, the error correction redundancy factor and mask complexity factor extracted from each format evaluation set are input as key feature vectors into a pre-trained convolutional neural network model for format complexity modeling. The convolutional neural network model outputs the modeled complexity score, and the format complexity of the target QR code is intelligently predicted based on the complexity score, reflecting the recognition time intensity and parsing difficulty level in the QR code decoding process.

[0022] Preferably, the conveyor belt speed is dynamically adjusted based on the complexity score output by the convolutional neural network model to adapt to the time requirement of decoding. The specific steps are as follows:

[0023] The complexity score output when the convolutional neural network model is used to intelligently predict the format complexity of the QR code is recorded as Γ p , where p represents the current p-th QR code sample, and the complexity score Γ p Calculate the theoretical optimal recognition time required for QR code decoding. The calculation expression is:

[0024]

[0025] ,in: T is the estimated decoding time corresponding to the p-th QR code; min is the lower limit of the basic recognition delay (i.e., the shortest recognition time for low-complexity QR codes, empirically set to 50ms); β1 and β2 are the nonlinear growth control coefficients of the recognition time, which control the sensitivity of the decoding time to the complexity score (adjusted according to the processing power of industrial recognition equipment, such as β1 = 100, β2 = 60); ρ is a high-order nonlinear weight factor used to amplify the influence of high-complexity areas, and its value range is usually 2≤ρ≤4;

[0026] α m is the growth rate adjustment coefficient of the square term of complexity score, which regulates the differential weight of high complexity score (the recommended value is 1≤α m ≤5);

[0027] After obtaining the estimated decoding time of the QR code Then, based on the current conveyor belt default linear speed v0 and the scan code recognition area length l scan Calculate the scanning time of each QR code without speed adjustment in,

[0028] like Trigger dynamic deceleration control and adjust the conveyor belt speed according to the following formula to generate a new dynamic recognition speed. The speed adjustment expression is:

[0029] ,in: The conveyor speed is adjusted within the QR code recognition area, changing dynamically to match the decoding time; m The speed adjustment coefficient of the conveyor belt is used to control the speed adjustment range (recommended range is 20 to 100ms); The power expansion term for the complexity score,

[0030] ω is the sensitivity index (usually 3 to 5), which is used to highlight the deceleration trend of highly complex codes;

[0031] μ m A safety margin threshold for complexity scoring, below which no significant slowdown is triggered (e.g. μ m =0.2); Introducing the periodic fluctuation compensation term, in the medium and high complexity segments ( ) Smoothly control speed jitter and avoid sudden speed drops.

[0032] Preferably, in the format evaluation set constructed from the QR code structure format information, the specific steps of generating the error correction redundancy factor after analyzing the redundant data ratio caused by the error correction level adopted by the QR code are as follows:

[0033] Extract two core parameters from the QR code structure format information: error correction level coefficient α e and module code load ratio β d , where the error correction level coefficient α e The redundancy ratio corresponding to the error correction level used by the QR code (such as L = 0.07, M = 0.15, Q = 0.25, H = 0.30) reflects the redundancy strength; while the module coding load ratio β d =n d / n t , where n d is the number of data modules, n t is the total number of QR code modules (including error correction modules and data modules), which is used to indicate the dilution degree of data proportion. The redundancy density weight factor is constructed by combining the error correction level coefficient and the module coding load ratio. The constructed expression is:

[0034]

[0035] , where: γ is the control amplification factor (recommended value range is 1.5~3.0), which is used to adjust the redundancy intensity and data dilution ratio (i.e. ) applies nonlinear enhancement or compression control; Λ x The redundancy density weighting factor quantifies the nonlinear interaction between the error correction redundancy strength and the effective data density of the QR code, and is used to characterize the complexity pressure intensity imposed on the recognition system. The redundancy density weighting factor is used to characterize the density aggregation caused by the redundant information on the effective load compression. The tanh(·) function introduces nonlinear boundary restrictions to enhance the gradient response to high redundancy situations.

[0036] After obtaining the redundancy density weight factor, the QR code version level parameter v (ranging from 1 to 40, representing the upgrade level of the QR code size and the total number of modules) is introduced to construct a weighted amplification model to generate the final error correction redundancy factor (the higher the version, the more QR code modules, and the greater the processing burden of redundant information on the decoding system). The generation expression of the error correction redundancy factor is:

[0037] E cri =Λ x ·ln(1+δ·v 2 )

[0038] , where δ is the version amplification factor (recommended value range is 0.01~0.05), which is used to control the degree of increase of version level, ln(1+δ·v 2) structure enhances the response to the complexity of high-version QR codes but avoids explosive growth; E cri is the error correction redundancy factor; the final error correction redundancy factor E cri It can be used as a core complexity characteristic indicator for sorting and classification in structural format evaluation.

[0039] Preferably, in the format evaluation set constructed from the QR code structure format information, the specific steps of generating the mask complexity factor after analyzing the degree of disturbance caused by the mask pattern used in the QR code to the overall pattern of the code surface are as follows:

[0040] After preprocessing, the QR code image is binarized (black and white) and divided into multiple fixed-size sliding windows (the size of each window is set to w×w, and the recommended value is w=5 or w=7 to ensure locality). For each window, the density perturbation function is used to calculate the difference between the number of black modules and the number of white modules. The calculation expression is:

[0041]

[0042] , where: B i，j is the number of black modules in the (i, j)th window; W i，j is the number of white modules in the (i, j)th window; i，j is the intensity of local structural disturbance. The closer the value is to 1, the greater the imbalance between black and white, and the greater the disturbance.

[0043] After obtaining all local structural perturbation intensities, the perturbation paths are further extracted from the QR code image in four main directions (i.e., horizontal 0°, vertical 90°, diagonal 45°, and anti-diagonal 135°), and the total perturbation intensity sequence corresponding to each direction is constructed. Based on the total perturbation intensity sequence, the distribution difference between the perturbations in each direction is calculated to generate a mask complexity factor, which is used to quantify the impact of the mask pattern on the image structure balance. The generation expression of the mask complexity factor is:

[0044]

[0045] , where: L θ represents the total disturbance intensity along the direction θ; max(L θ )、min(L θ ) represent the maximum and minimum values ​​of the directional disturbance respectively; M ci It is the mask complexity factor. The larger the value, the more uneven the disturbance in each direction, the more chaotic the structure, and the higher the difficulty of recognition.

[0046] Preferably, And θ∈{0°, 90°, 45°, 135°}.

[0047] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0048] The present invention can effectively solve the recognition time mismatch problem caused by the difference in QR code structure in existing high-speed industrial scenarios by constructing an identification mechanism based on intelligent prediction of QR code format complexity and dynamic regulation of conveyor belt speed. By pre-shooting the QR code image and analyzing the structural format information, combining feature engineering with the complexity score output by the convolutional neural network model, the conveyor belt speed is dynamically adjusted to achieve recognition rhythm matching, thereby ensuring that high-complexity QR codes have sufficient recognition time and low-complexity QR codes can achieve high-speed passage. Ultimately, the accurate binding of QR code recognition results and equipment life cycle data is achieved, ensuring the complete collection and traceability of information on the entire life cycle of electromechanical products, which not only improves the stability and beat efficiency of the recognition system, but also promotes efficient collaboration of supply chain management towards digitalization and intelligence, and has good application value and promotion prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction to the drawings required for use in the embodiments will be given below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0050] Figure 1 This is a module diagram of a commodity full life cycle management service platform based on the electromechanical industry supply chain of the present invention. DETAILED DESCRIPTION

[0051] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art.

[0052] The present invention provides Figure 1 The illustrated platform is a product lifecycle management service platform based on the supply chain of the electromechanical industry, comprising an image acquisition and analysis module, a format feature construction module, a complexity prediction modeling module, a rhythm adaptive control module, an information decoding and acquisition module, and a data binding and archiving module.

[0053] The image acquisition and analysis module uses industrial-grade visual equipment (such as a high-speed CMOS camera) deployed at the front end of the recognition area on the high-speed conveyor belt to pre-shoot the QR code image that is about to enter the recognition window in real time, and obtains the structural format information of the QR code by analyzing the QR code image;

[0054] The core goal of this process is to capture clear image data of the QR code in advance before it enters the formal scanning and recognition area. The image acquisition module usually adopts a short exposure and high frame rate configuration to ensure that the image is not blurred due to high-speed motion. At the same time, the image recognition algorithm is used to perform contour extraction, positioning correction and image enhancement on the captured QR code image to further improve the code surface image quality. The system can then extract preliminary structural information based on the QR code format characteristics (such as Data Matrix, QR Code standard version, error correction level) and mark its corresponding format type as the basic input for subsequent format complexity evaluation.

[0055] Obtain the structural format information of the QR code by parsing the QR code image. The specific steps are as follows:

[0056] Step 1: QR code positioning and positioning graphic recognition;

[0057] First, the QR code is identified and located within the image. QR codes typically contain specific positioning patterns (such as the "positioning detection patterns" at the three corners of a QR code or the "L-shaped boundary" of a DataMatrix). The parsing system uses these patterns to quickly determine the spatial position and orientation of the QR code within the image. This step determines the boundaries of the QR code's coding area and the sampling reference, providing a position reference and angle correction basis for subsequent structural decoding.

[0058] Step 2: Module matrix construction and grid division;

[0059] Based on the positioning pattern, the system divides the coded area into several small blocks (modules) according to the standard version rules of the QR code, forming a module matrix (such as 21×21, 25×25, etc.). The dimensions of this matrix directly reflect the "version number" of the QR code. For example, QR codes go from version 1 (21×21) to version 40 (177×177). By reading the black and white states of these modules point by point, the system constructs the "code surface structure diagram" of the entire QR code and counts structural parameters such as the number of modules and the proportion of the area occupied by the error correction code.

[0060] Step 3: Format information parsing and structure field extraction;

[0061] After the module matrix is ​​identified, the system parses the dedicated areas containing format information based on QR code encoding standards (such as ISO / IEC 18004), such as the Format Information area and the Version Information area. These areas store core structural parameters such as the QR code's error correction level (such as L, M, Q, H), mask mode, version number, and encoding method (such as byte mode and numeric mode). By parsing these fields, the system can accurately extract the QR code's structural format information.

[0062] The format feature construction module constructs an independent format evaluation set for the structural format information collected from each QR code, applies feature engineering technology to extract key indicators reflecting the difficulty of QR code recognition, analyzes the extracted indicators, and constructs a profile for identifying the complexity characteristics of the QR code based on the analyzed features;

[0063] Feature engineering technology is applied to extract key indicators reflecting the difficulty of QR code recognition from the format evaluation set. The extracted indicators include the proportion of redundant data caused by the error correction level adopted by the QR code and the degree of disturbance caused to the overall pattern of the code surface by the mask mode used by the QR code. After analyzing the extracted indicators, the error correction redundancy factor and mask complexity factor are generated respectively. The error correction redundancy factor and mask complexity factor are used to construct a portrait for identifying the complexity characteristics of the QR code.

[0064] The higher the error correction level used in a QR code, the greater the proportion of redundant data, which generally translates to a more complex structural format and greater recognition difficulty. This is because QR codes utilize built-in error correction mechanisms (typically based on Reed-Solomon coding) to enhance their resistance to corruption. The corresponding error correction capabilities for different levels (L, M, Q, and H) are approximately 7%, 15%, 25%, and 30%, respectively. As the error correction level increases, the proportion of original data in the QR code decreases, while the proportion of redundant checksum data increases. This significantly increases the total number of QR code modules and pattern density. High redundancy not only increases the computational burden of the decoding algorithm but also raises the requirements for image acquisition quality (such as clarity, contrast, and lighting balance). This is particularly true in high-speed industrial recognition scenarios, where denser modules lead to higher image resolution requirements while reducing the tolerance for the sampling frame rate and stability of industrial cameras, thereby increasing overall recognition complexity. Therefore, a higher error correction level generally indicates a more complex QR code format and is a key indicator in assessing recognition difficulty.

[0065] In the format evaluation set constructed from the QR code structure format information, the specific steps for generating the error correction redundancy factor after analyzing the redundant data ratio caused by the error correction level adopted by the QR code are as follows:

[0066] Extract two core parameters from the QR code structure format information: error correction level coefficient α e and module code load ratio β d , where the error correction level coefficient α e The redundancy ratio corresponding to the error correction level used by the QR code (such as L = 0.07, M = 0.15, Q = 0.25, H = 0.30) reflects the redundancy strength; while the module coding load ratio β d =n d / n t, where n d is the number of data modules, n t is the total number of QR code modules (including error correction modules and data modules), which is used to indicate the dilution degree of data proportion. The redundancy density weight factor is constructed by combining the error correction level coefficient and the module coding load ratio. The constructed expression is:

[0067]

[0068] , where: γ is the control amplification factor (recommended value range is 1.5~3.0), which is used to adjust the redundancy intensity and data dilution ratio (i.e. ) applies nonlinear enhancement or compression control; Λ x The redundancy density weighting factor quantifies the nonlinear interaction between the error correction redundancy strength and the effective data density of the QR code, and is used to characterize the complexity pressure intensity imposed on the recognition system. The redundancy density weighting factor is used to characterize the density aggregation caused by the redundant information on the effective load compression. The tanh(·) function introduces nonlinear boundary restrictions to enhance the gradient response to high redundancy situations.

[0069] The error correction coefficient is directly derived from the QR code's format information area based on the error correction level (ECL). QR code standards (such as ISO / IEC 18004 QR Code) define four error correction levels: L (low), M (medium), Q (high), and H (high), corresponding to approximately 7%, 15%, 25%, and 30% error resilience, respectively. This information is embedded in the QR code's format information area, typically encoded as a 15-bit format string containing both the error correction level and the mask pattern. The recognition system determines the specific ECL by parsing the first two bits or corresponding bit segments in the format information string (according to a standard mapping table) and maps this to an ECL coefficient, such as L = 0.07, M = 0.15, Q = 0.25, and H = 0.30. This ECL coefficient can be used as a quantitative parameter in complexity assessment modeling to reflect the computational load and image processing pressure imposed on the recognition system by the proportion of redundant data in the QR code.

[0070] The role of this step is to transform the nonlinear interaction between error correction level and data density into a weighted model of structural complexity pressure.

[0071] After obtaining the redundancy density weight factor, the QR code version level parameter v (ranging from 1 to 40, representing the upgrade level of the QR code size and the total number of modules) is introduced to construct a weighted amplification model to generate the final error correction redundancy factor (the higher the version, the more QR code modules, and the greater the processing burden of redundant information on the decoding system). The generation expression of the error correction redundancy factor is:

[0072] E cri =Λ x ·ln(1+δ·v 2 )

[0073] , where δ is the version amplification factor (recommended value range is 0.01~0.05), which is used to control the degree of increase of version level, ln(1+δ·v 2 ) structure enhances the response to the complexity of high-version QR codes but avoids explosive growth; E cri is the error correction redundancy factor; the final error correction redundancy factor E cri It can be used as a core complexity characteristic indicator for sorting and classification in structural format evaluation;

[0074] The version level parameter (usually denoted as v) of a QR code indicates the size level of the QR code, ranging from 1 to 40, corresponding to the module size of the QR code (e.g., version 1 is a 21×21 module, and version 40 is a 177×177 module). This parameter can be obtained by parsing the version information area within the QR code image. For QR codes of version 7 and above, the standard requires an 18-bit version information code to be embedded in the image, located at a specific location near the positioning mark. The version number can be directly obtained by decoding this area. For small QR codes of versions 1 to 6, since no version information area is provided, the system can infer the corresponding version level by identifying the total number of modules in the QR code (i.e., the number of rows and columns of black and white module grids in the image). The purpose of obtaining the version parameter v is to facilitate subsequent recognition complexity modeling and control strategy development. Since higher QR code versions contain more and denser coding modules, the processing load on the recognition algorithm is also heavier, generally indicating greater recognition difficulty.

[0075] The purpose of this step is to couple the redundant pressure with the structural size of the QR code to generate a result that is both sensitive and stable, characterizing the complexity characteristics of the QR code.

[0076] The error correction redundancy factor (ERF) shows that, in a format evaluation set constructed from QR code structural format information, the proportion of redundant data introduced by the error correction level employed by the QR code is analyzed. A larger ERF indicates a more complex QR code structure and format, making it more difficult to recognize. Conversely, a smaller ERF indicates a less complex QR code structure and format, making it easier to recognize. This is because the ERF integrates the error correction level (reflecting redundancy strength) with the effective data dilution ratio (data density) through a nonlinear fusion model, comprehensively characterizing the processing load imposed on the decoding system by the QR code structure. When the error correction level is higher (such as Q or H), the proportion of redundant modules in the QR code increases significantly, the overall code density increases, the boundaries between modules narrow, the image sampling accuracy requirements become higher, and the error tolerance of the recognition algorithm decreases. Furthermore, higher versions and sparser data lead to more space filling in the structure, further increasing the spatial mapping difficulty of the decoding algorithm. Therefore, a larger ERF indicates a higher system resource usage per unit recognition time and a narrower error tolerance window, making it an important metric for assessing recognition complexity.

[0077] The mask pattern used by the QR code has a significant impact on its recognizability. The higher the degree of mask perturbation, the more complex the structure of the QR code, which increases the difficulty of recognition. The mask pattern is essentially to avoid the appearance of long areas of the same color in the QR code (such as large continuous black or white blocks). By performing bit-by-bit logical transformations on the encoded data, the pattern is broken up to make it more visually evenly distributed. However, some mask patterns (such as Mask Pattern 6 and 7) will cause the code surface to form dense complex textures or high-frequency patterns, especially in high-contrast backgrounds, reflective environments or high-speed shooting, which can easily cause stripe blurring, boundary aliasing or sampling jitter on the image sensor, thereby affecting the accuracy of module positioning and data recovery. In addition, certain mask patterns will form structures similar to interference patterns, which are easily confused with the locator and correction symbols of the QR code, causing the recognition algorithm to make errors. Therefore, in the structural format information, the degree of disturbance caused by the mask pattern (i.e., "mask complexity") is one of the key parameters for evaluating the difficulty of QR code recognition. A high mask disturbance value often indicates that the QR code is more difficult to be accurately recognized in high-speed industrial scenarios.

[0078] In the format evaluation set constructed from the QR code structure format information, the specific steps for generating the mask complexity factor after analyzing the degree of disturbance caused by the mask pattern used in the QR code to the overall pattern of the code surface are as follows:

[0079] After preprocessing, the QR code image is binarized (black and white) and divided into multiple fixed-size sliding windows (the size of each window is set to w×w, and the recommended value is w=5 or w=7 to ensure locality). For each window, the density perturbation function is used to calculate the difference between the number of black modules and the number of white modules. The calculation expression is:

[0080]

[0081] , where: B i，j is the number of black modules in the (i, j)th window; W i，j is the number of white modules in the (i, j)th window; i，j is the intensity of local structural disturbance. The closer the value is to 1, the greater the imbalance between black and white, and the greater the disturbance.

[0082] The purpose of this step is to capture the "oppression" and "structural fragmentation" of the local image disturbance caused by the mask by scanning the local black and white density imbalance of the QR code pattern, providing a disturbance distribution basis for the subsequent global mask complexity factor.

[0083] After obtaining all local structural perturbation intensities, the perturbation paths are further extracted from the QR code image in four main directions (i.e., horizontal 0°, vertical 90°, diagonal 45°, and anti-diagonal 135°), and the total perturbation intensity sequence corresponding to each direction is constructed. Based on the total perturbation intensity sequence, the distribution difference between the perturbations in each direction is calculated to generate a mask complexity factor, which is used to quantify the impact of the mask pattern on the image structure balance. The generation expression of the mask complexity factor is:

[0084]

[0085] , where: L θ represents the total perturbation intensity along the direction θ, And θ∈{0°, 90°, 45°, 135°}; max(L θ )、min(L θ ) represent the maximum and minimum values ​​of the directional disturbance respectively; M ci is the mask complexity factor. A larger value indicates that the disturbance in each direction is strongly uneven, the structure is more chaotic, and the recognition difficulty is higher.

[0086] This step is to evaluate whether the perturbation distribution of the QR code pattern in multiple directions is balanced, and to measure the degree of interference of the mask on the overall image structure from a geometric perspective. If the mask causes dense stacking of image modules in some directions and sparse and broken images in other directions, the probability of positioning and decoding errors will increase significantly. Therefore, the higher the directional heterogeneity, the higher the mask complexity factor M. ci The larger it is, the higher the complexity of QR code recognition.

[0087] The mask complexity factor (MCF) indicates that, within a format evaluation set constructed from QR code structural format information, the degree of perturbation caused by the mask pattern used in the QR code to the overall pattern is analyzed. A higher value indicates a more complex and difficult-to-recognize QR code structure, while a lower value indicates a less complex QR code structure. This is because the MCF essentially reflects the degree of structural imbalance caused by mask perturbation in different directions, i.e., the geometric heterogeneity of the pattern. If the mask pattern introduces a large number of densely stacked modules in some directions and sparse blocks in other directions, the overall QR code pattern will exhibit an irregular texture, increasing the difficulty of image sampling, positioning alignment, and module discrimination. A higher MCF indicates more pronounced perturbation differences between directions, a more fragmented or uneven pattern, and a greater likelihood of misjudgment or failure during fast decoding by the recognition algorithm. Conversely, a lower MCF indicates a more evenly distributed and structurally stable pattern, enabling the recognition system to more quickly and accurately locate and decode the code, resulting in a lower recognition difficulty. Therefore, the level of the mask complexity factor can be used as an important quantitative basis for the complexity of the structural format, which is directly related to the recognizability of the QR code in high-speed industrial scenarios.

[0088] The complexity prediction modeling module uses the features extracted from each format evaluation set as feature vectors and inputs them into a pre-trained convolutional neural network model. The convolutional neural network model then outputs a complexity score to intelligently predict the format complexity of the QR code.

[0089] The error correction redundancy factor and mask complexity factor extracted from each format evaluation set are input as key feature vectors into the pre-trained convolutional neural network model for format complexity modeling. The convolutional neural network model outputs the modeled complexity score, and the format complexity of the target QR code is intelligently predicted based on the complexity score, reflecting the recognition time intensity and parsing difficulty level during the QR code decoding process.

[0090] The pre-trained convolutional neural network model refers to the convolutional neural network (CNN) used in the format complexity modeling method adopted in this solution. It is not trained only when the QR code recognition task is carried out, but before the system is deployed or the model is applied, it has been offline trained, parameter optimized and performance verified based on a large number of representative historical QR code data samples. Through the training process, this model learns the nonlinear mapping relationship between the input feature vector (such as error correction redundancy factor, mask complexity factor) and the target output variable (i.e., complexity score), and can quickly and efficiently predict the recognition complexity of the input feature vector in practical applications. The so-called pre-training is completed, which emphasizes that the model has real-time reasoning capabilities, that is, in the actual industrial recognition process, no time-consuming parameter updates or structure iterations are performed. Instead, the existing model structure and weights are directly called for forward reasoning (Forward Inference) to achieve rapid output of complexity scores.

[0091] In this approach, the convolutional neural network model is typically trained using supervised learning. This involves iteratively optimizing the network model by constructing a dataset containing input feature vectors (such as error correction redundancy factors and mask complexity factors) and complexity labels derived from manual annotations or empirical evaluations. During the model training phase, the system first extracts features from tens of thousands of captured QR code images and their format parameters to establish a format evaluation set. A target complexity score is then constructed using methods such as manual settings, historical recognition time records, and recognition success rates. This data is then fed into a designed multi-layer convolutional neural network. A loss function (such as mean squared error loss) is used to measure the error between the predicted results and the true labels. The network's convolution kernel parameters, fully connected weights, and bias terms are then adjusted through a backpropagation algorithm to continuously optimize the model's fitting capabilities. The entire training process can be completed on a high-performance computing platform (such as a GPU server). After training, the model parameters are frozen and deployed to edge devices or industrial control hosts. This allows for instant access to the corresponding complexity score by simply inputting a new QR code format feature vector, eliminating the need for additional training or parameter tuning. This mechanism ensures that the recognition system can maintain high-precision, high-robustness and low-latency intelligent prediction performance when facing large-scale, high-frequency recognition tasks. It is one of the key supporting units for realizing the "prediction-control-recognition" closed-loop mechanism of the QR code recognition process.

[0092] The rhythm adaptive control module sends speed adjustment instructions to the conveyor belt scheduling unit. Based on the complexity score output by the convolutional neural network model, it dynamically adjusts the conveyor belt speed to adapt to the time requirements of decoding, forming a rhythm matching. Specifically, for high-complexity QR codes, the conveyor belt automatically slows down within the recognition window (for example, by 100 to 300 milliseconds) to ensure that the recognition system has sufficient time to complete data analysis. For low-complexity QR codes, the conveyor belt maintains normal speed to improve overall beat efficiency.

[0093] This control is performed by a PLC controller or industrial bus (such as EtherCAT or PROFINET). Closed-loop control ensures response latency is less than 50ms, ensuring dynamic rhythm coordination in a high-beat environment. The core function of this step is to compensate for uneven scanning and recognition capabilities through flexible adjustment of the motion system, forming an adaptive recognition mechanism where "codes follow the beat and the beat adapts to the code."

[0094] Based on the complexity score output by the convolutional neural network model, the conveyor belt speed is dynamically adjusted to adapt to the decoding time requirements. The specific steps are as follows:

[0095] The complexity score output when the convolutional neural network model is used to intelligently predict the format complexity of the QR code is recorded as Γ p , where p represents the current p-th QR code sample, and the complexity score Γ p Calculate the theoretical optimal recognition time required for QR code decoding. The calculation expression is:

[0096]

[0097] ,in: T is the estimated decoding time corresponding to the p-th QR code; min is the lower limit of the basic recognition delay (i.e., the shortest recognition time of a low-complexity QR code, empirically set to 50ms); β1 and β2 are the nonlinear growth control coefficients of the recognition time, which control the sensitivity of the decoding time to the complexity score (adjusted according to the processing power of the industrial recognition equipment, such as β1 = 100, β2 = 60); ρ is a high-order nonlinear weight factor used to amplify the influence of high-complexity areas, and its value range is usually 2≤ρ≤4; α m is the growth rate adjustment coefficient of the square term of complexity score, which regulates the differential weight of high complexity score (the recommended value is 1≤α m ≤5);

[0098] This step converts the complexity score into a measurable indicator of the time required for recognition by modeling the high-order combination of the output scores of the convolutional neural network model. This reflects the nonlinear enhancement effect of the complexity of the QR code structure on the actual recognition time, and provides an accurate time basis for conveyor belt speed control.

[0099] After obtaining the estimated decoding time of the QR code Then, based on the current conveyor belt default linear speed v0 and the scan code recognition area length L scan Calculate the scanning time of each QR code without speed adjustment in,

[0100] like Trigger dynamic deceleration control and adjust the conveyor belt speed according to the following formula to generate a new dynamic recognition speed. The speed adjustment expression is:

[0101] ,in: The conveyor speed is adjusted within the QR code recognition area, changing dynamically to match the decoding time; m The speed adjustment coefficient of the conveyor belt is used to control the speed adjustment range (recommended range is 20 to 100ms); The power expansion term for the complexity score,

[0102] ω is the sensitivity index (usually 3 to 5), which is used to highlight the deceleration trend of highly complex codes;

[0103] μ m A safety margin threshold for complexity scoring, below which no significant slowdown is triggered (e.g. μ m =0.2); Introducing periodic fluctuation compensation term, in the medium and high complexity segments Smoothly control speed jitter to avoid sudden speed drops;

[0104] This control strategy can achieve:

[0105] Low complexity QR code (Γ p <<μ m ), the system maintains or slightly increases the speed Ensure high beat efficiency;

[0106] High complexity QR code (Γ p →1), the recognition window length is automatically extended and the speed is dynamically reduced to effectively match the estimated decoding time and avoid recognition failures or information omissions.

[0107] The purpose of this step is to implement an adaptive matching mechanism between the time consumption of QR code recognition and decoding and the rhythm of the conveyor belt. By introducing multiple nonlinear terms and intelligent speed regulation strategies, it ensures that the system can stably complete code scanning and data collection under different recognition load conditions. It is the key control logic for realizing the "flexible decoding rhythm" of industrial identification.

[0108] The information decoding and acquisition module, after the conveyor belt speed is adjusted and enters the rhythm matching state, uses an information reading device (such as a high-speed barcode scanner, industrial camera + decoding unit) to decode the QR code officially sent into the identification area, extract the unique identification code of the product from the QR code, and record the associated life cycle information, including key data such as equipment model, batch, quality inspection status, maintenance records, etc.

[0109] The purpose of this step is to achieve stable information collection of the recognition system under dynamic rhythm conditions, ensure that data is not lost or misinterpreted, and provide stable and accurate basic data for subsequent management links (such as intelligent sorting, quality control, and process traceability).

[0110] The data binding and archiving module, after successful QR code recognition, matches and binds the collected unique identification code with multi-dimensional data such as the current production beat information, work station data, and equipment code, and records it in the lifecycle data platform;

[0111] This binding operation is written to systems such as MES, PLM, and ERP through middleware or API interfaces, ensuring that the equipment has traceable digital identity information throughout its entire process, from production line entry to circulation, use, maintenance, and recycling. This step integrates the QR code recognition results with the entire lifecycle management system to achieve a closed-loop data integration, supporting data source consistency and availability in subsequent intelligent manufacturing, visual scheduling, and digital twin scenarios.

[0112] The present invention can effectively solve the recognition time mismatch problem caused by the difference in QR code structure in existing high-speed industrial scenarios by constructing an identification mechanism based on intelligent prediction of QR code format complexity and dynamic regulation of conveyor belt speed. By pre-shooting the QR code image and analyzing the structural format information, combining feature engineering with the complexity score output by the convolutional neural network model, the conveyor belt speed is dynamically adjusted to achieve recognition rhythm matching, thereby ensuring that high-complexity QR codes have sufficient recognition time and low-complexity QR codes can achieve high-speed passage. Ultimately, the accurate binding of QR code recognition results and equipment life cycle data is achieved, ensuring the complete collection and traceability of information on the entire life cycle of electromechanical products, which not only improves the stability and beat efficiency of the recognition system, but also promotes efficient collaboration of supply chain management towards digitalization and intelligence, and has good application value and promotion prospects.

[0113] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0114] The above description is merely illustrative of certain exemplary embodiments of the present invention. It goes without saying that those skilled in the art will be able to modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and description are illustrative in nature and should not be construed as limiting the scope of protection of the claims.

[0115] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A product life cycle management service platform based on the supply chain of the electromechanical industry, characterized by: It includes image acquisition and analysis module, format feature construction module, complexity prediction modeling module, rhythm adaptive control module, information decoding and acquisition module and data binding and archiving module; The image acquisition and analysis module uses industrial-grade visual equipment deployed at the front end of the high-speed conveyor belt recognition area to pre-shoot the QR code image that is about to enter the recognition window in real time, and obtains the structural format information of the QR code by analyzing the QR code image; The format feature construction module constructs an independent format evaluation set for the structural format information collected from each QR code, applies feature engineering technology to extract key indicators reflecting the difficulty of QR code recognition, analyzes the extracted indicators, and constructs a profile for identifying the complexity characteristics of the QR code based on the analyzed features; The complexity prediction modeling module uses the features extracted from each format evaluation set as feature vectors and inputs them into a pre-trained convolutional neural network model. The convolutional neural network model then outputs a complexity score to intelligently predict the format complexity of the QR code. The rhythm adaptive control module sends speed adjustment instructions to the conveyor belt scheduling unit. Based on the complexity score output by the convolutional neural network model, the conveyor belt speed is dynamically adjusted to adapt to the time requirements of decoding, forming a rhythm matching; The information decoding and acquisition module, after the conveyor belt speed adjustment is completed and enters the rhythm matching state, uses the information reading device to decode the QR code officially sent into the identification area, extract the unique identification code of the product from the QR code, and record the associated life cycle information; The data binding and archiving module matches and binds the collected multi-dimensional data after successful QR code recognition and records it in the lifecycle data platform.

2. A commodity life cycle management service platform based on the supply chain of the electromechanical industry according to claim 1, characterized in that: Obtain the structural format information of the QR code by parsing the QR code image. The specific steps are as follows: Identify and locate the graphic position of the QR code in the QR code image; Based on the positioning pattern, the coding area is divided into several small blocks according to the standard version rules of the QR code to form a module matrix. The version number of the QR code is reflected by the dimension of the module matrix. After the module matrix is ​​identified, the dedicated area containing format information is parsed according to the QR code encoding standard, and the structural format information of the QR code is extracted by parsing the dedicated area field.

3. The product life cycle management service platform based on the electromechanical industry supply chain according to claim 1 is characterized in that: Feature engineering technology is applied to extract key indicators reflecting the difficulty of QR code recognition from the format evaluation set. The extracted indicators include the proportion of redundant data caused by the error correction level adopted by the QR code and the degree of disturbance caused to the overall pattern of the code surface by the mask mode used by the QR code. After analyzing the extracted indicators, the error correction redundancy factor and mask complexity factor are generated respectively. The error correction redundancy factor and mask complexity factor are used to construct a portrait for identifying the complexity characteristics of the QR code.

4. The product life cycle management service platform based on the electromechanical industry supply chain according to claim 3 is characterized in that: The error correction redundancy factor and mask complexity factor extracted from each format evaluation set are input as key feature vectors into the pre-trained convolutional neural network model for format complexity modeling. The convolutional neural network model outputs the modeled complexity score, and the format complexity of the target QR code is intelligently predicted based on the complexity score, reflecting the recognition time intensity and parsing difficulty level during the QR code decoding process.

5. The product life cycle management service platform based on the supply chain of the electromechanical industry according to claim 4 is characterized in that: Based on the complexity score output by the convolutional neural network model, the conveyor belt speed is dynamically adjusted to adapt to the decoding time requirements. The specific steps are as follows: The complexity score output when the convolutional neural network model is used to intelligently predict the format complexity of the QR code is recorded as ,in p Indicates the current p QR code samples, scored according to complexity Calculate the theoretical optimal recognition time required for QR code decoding. The calculation expression is: ,in: For the p The estimated decoding time corresponding to each QR code; As the basic identification delay lower limit; and To identify the regulation coefficient of temporal nonlinear growth and control the sensitivity of decoding time to complexity score; is a high-order nonlinear weight factor used to amplify the influence of high-complexity areas; is the growth rate adjustment coefficient, which regulates the differential weight of the high complexity score; After obtaining the estimated decoding time of the QR code Then, according to the current conveyor belt default line speed Length of the scanned code recognition area Calculate the scanning time of each QR code without speed adjustment ,in, ;like , triggering dynamic deceleration control, adjusting the conveyor belt speed according to the following formula to generate a new dynamic recognition speed. The speed adjustment expression is: ,in: The conveyor speed is adjusted within the QR code recognition area, changing dynamically to match the decoding time; It is the conveyor belt speed adjustment coefficient, which is used to control the speed adjustment range; is a sensitivity index used to highlight the deceleration trend of high-complexity codes; A safety margin threshold for scoring complexity below which no significant slowdown is triggered.

6. The product life cycle management service platform based on the electromechanical industry supply chain according to claim 3 is characterized in that: In the format evaluation set constructed from the QR code structure format information, the specific steps for generating the error correction redundancy factor after analyzing the redundant data ratio caused by the error correction level adopted by the QR code are as follows: Extract two core parameters from the QR code structure format information: error correction level coefficient and module code load ratio , where the error correction level coefficient The redundancy ratio corresponding to the error correction level used by the QR code reflects the redundancy strength; while the module coding load ratio ,in is the number of data modules, is the total number of QR code modules, which is used to indicate the dilution degree of data proportion. The redundancy density weight factor is constructed by combining the error correction level coefficient and the module coding load ratio. After obtaining the redundancy density weight factor, the QR code version level parameter is introduced v , a weighted amplification model is constructed to generate the final error correction redundancy factor.

7. The product life cycle management service platform based on the electromechanical industry supply chain according to claim 3 is characterized in that: In the format evaluation set constructed from the QR code structure format information, the specific steps for generating the mask complexity factor after analyzing the degree of disturbance caused by the mask pattern used in the QR code to the overall pattern of the code surface are as follows: The QR code image is binarized after preprocessing and divided into multiple sliding windows of fixed size. For each window, the difference between the number of black modules and the number of white modules is calculated using the density perturbation function. After obtaining all local structural perturbation intensities, the perturbation paths are extracted from the QR code image in the four main directions, and the total perturbation intensity sequence corresponding to each direction is constructed. Based on the total perturbation intensity sequence, the distribution differences between the perturbations in each direction are calculated, and a mask complexity factor is generated to quantify the impact of the mask pattern on the image structure balance.

Images