Blockchain-based school uniform quality information traceability method and system

Abstract

The application provides a school uniform quality information traceability method and system based on a blockchain, relates to the technical field of the blockchain, and comprises the following steps: collecting school uniform production process data, screening quality key factors, generating a link conversion verification code and a unique identifier, forming a blockchain, and distributing the blockchain to distributed nodes. Each node constructs a quality state conversion graph, performs traceability verification on a convergence node, and locates a deviated link. The school uniform quality information is traceable in a credible manner through the blockchain technology, the information transparency and traceability accuracy are improved, and the school uniform quality management capability is enhanced.

Description

Technical Field

[0001] This invention relates to the field of blockchain technology, and in particular to a blockchain-based method and system for tracing school uniform quality information. Background Technology

[0002] As school uniforms are worn daily by students, their quality and safety have always been a major concern for society. The transparency and traceability of school uniform quality information are becoming increasingly important. The production process of school uniforms involves multiple stages, including raw material procurement, fabric processing, cutting, sewing, and ironing, each of which significantly impacts the final product quality. Traditional school uniform quality management mainly relies on internal quality inspections and random checks by the manufacturing company, lacking complete data recording of the entire production process and third-party supervision and verification mechanisms.

[0003] Blockchain technology, due to its decentralized, immutable, and traceable characteristics, has been widely applied in supply chain management and product traceability in recent years. In the area of ​​school uniform quality traceability, blockchain technology can achieve data recording and sharing throughout the entire production process, providing reliable technical assurance for school uniform quality. Some school uniform manufacturers have begun exploring the use of information technology to manage the school uniform production process, but a systematic solution has not yet been formed.

[0004] Existing technologies for school uniform quality traceability still have some shortcomings and deficiencies. Current school uniform quality traceability systems lack the ability to effectively identify and analyze key quality factors in the production process. The school uniform production process involves numerous technological parameters, and the lack of scientific sensitivity analysis methods to determine the key factors truly affecting school uniform quality leads to data redundancy or missing key data. Traditional school uniform quality traceability methods, due to the lack of effective data verification mechanisms, are susceptible to human manipulation or omission of quality data during the production process, severely impacting the reliability and authority of the traceability system and failing to provide credible assurance for school uniform quality. Existing technologies struggle to accurately trace and locate the source of quality anomalies. When quality problems occur in school uniforms, it is impossible to quickly pinpoint the specific production stage and responsible party, resulting in untimely and ineffective resolution of quality issues. This also increases the difficulty of pursuing quality accountability and hinders the continuous improvement and enhancement of school uniform production quality. Summary of the Invention

[0005] This invention provides a blockchain-based method and system for tracing school uniform quality information, which can solve the problems in the prior art.

[0006] A first aspect of this invention provides a blockchain-based method for tracing school uniform quality information, comprising:

[0007] Collect process data at each stage of the school uniform production process;

[0008] Extract process parameters from each step of the process data, calculate the sensitivity coefficient of the process parameters to the final quality indicators of the school uniform, screen process parameters whose sensitivity coefficients exceed the preset sensitivity threshold to identify key quality factors, and record the numerical change trajectory of key quality factors in each step.

[0009] A time-series coding matrix is ​​generated based on the numerical change trajectory of key quality factors. The numerical differences of multiple key quality factors are matched and verified with the collaborative conversion rules to generate a link conversion verification code.

[0010] A unique identifier is generated based on key quality factors and process transition verification codes. The process data and the unique identifier are encapsulated into blocks to form a blockchain and distributed to distributed nodes.

[0011] Each distributed node constructs a quality state transition graph based on the link transition verification code, identifies the convergence node with multiple predecessor paths in the quality state transition graph, and performs backtracking verification on the multiple predecessor paths of the convergence node.

[0012] When any predecessor path verification fails, a backtracking data chain is constructed along the predecessor path to locate the first deviation link and write the deviation tracing record into the blockchain.

[0013] Receive traceability requests containing unique identifiers, retrieve the numerical change trajectory and verification results of key quality factors from the blockchain, and generate a school uniform quality information traceability report.

[0014] In one optional embodiment, process parameters for each stage are extracted from the process data, the sensitivity coefficient of the process parameters to the final quality indicators of the school uniform is calculated, process parameters with sensitivity coefficients exceeding a preset sensitivity threshold are selected to determine key quality factors, and the numerical change trajectory of key quality factors in each stage is recorded, including:

[0015] The process data is segmented according to the execution sequence of each stage. Process parameters are extracted from each stage segment and the stage location information to which the process parameters belong is marked. A parameter vector set containing process parameter values ​​and stage location information is constructed.

[0016] For each process parameter in the parameter vector set, multiple sets of disturbance values ​​are set within its historical value range. The disturbance values ​​are combined with the measured values ​​of other process parameters in the parameter vector set. According to the quality index calculation rules, the calculation results of the final quality index of the school uniform corresponding to each combination are obtained. The fluctuation range of the final quality index calculation results of the school uniform caused by the disturbance values ​​is statistically analyzed. The fluctuation range is calculated by the ratio of the change amplitude of the disturbance values ​​to obtain the sensitivity coefficient of the process parameter to the final quality index of the school uniform.

[0017] Process parameters with sensitivity coefficients exceeding a preset sensitivity threshold are identified as key quality factors. Measured values ​​of these key quality factors in each stage are extracted and arranged according to stage location information to form a time-series numerical sequence. Stage switching positions are marked in the time-series numerical sequence, and the numerical differences before and after the switching positions are extracted. The numerical differences, along with the corresponding stage switching positions and acquisition times, are encapsulated to form a numerical change trajectory.

[0018] In one optional embodiment, a time-series coding matrix is ​​generated based on the numerical change trajectory of key quality factors. The numerical differences of multiple key quality factors are matched and verified with collaborative conversion rules to generate a stage conversion verification code, including:

[0019] From the numerical change trajectory, the numerical differences of multiple key quality factors at the same stage switching position are extracted, and a multi-dimensional numerical difference vector is constructed. The numerical differences of each key quality factor in the multi-dimensional numerical difference vector are weighted and fused according to preset weights to obtain comprehensive difference features. The comprehensive difference features are time-series correlated with the stage switching position and the acquisition time to generate a time-series coding matrix.

[0020] The collaborative conversion rules of multiple quality key factors corresponding to the switching position are retrieved from the preset process conversion rule base. The collaborative conversion rules define the linkage change mode and mutual constraint relationship of multiple quality key factors during the switching of the process. The comprehensive differential features in the time-series coding matrix are matched with the linkage change mode to calculate the similarity. The numerical difference of multiple quality key factors is verified based on the mutual constraint relationship.

[0021] When the similarity exceeds the preset matching threshold and the consistency verification passes, the feature summary of the time-series coding matrix is ​​combined with the stage switching position identifier and the verification pass mark to perform a hash operation to generate a stage conversion verification code.

[0022] In one optional embodiment, each distributed node constructs a quality state transition graph based on the link transition verification code, identifies a convergence node with multiple predecessor paths in the quality state transition graph, and performs backtracking verification on the multiple predecessor paths of the convergence node, including:

[0023] Each distributed node receives the stage transition verification code, parses the stage switching position identifier and the feature summary of the timing coding matrix from the stage transition verification code, maps the stage switching position identifier to the state node in the quality state transition graph, establishes directed transition edges between state nodes according to the temporal sequence relationship between stage switching position identifiers, and marks the feature summary of the timing coding matrix on the corresponding directed transition edges to construct the quality state transition graph.

[0024] Traverse all state nodes in the quality state transition graph, count the number of incoming edges for each state node, filter state nodes whose number of incoming edges exceeds a preset convergence threshold to obtain convergence nodes, extract the predecessor paths corresponding to all incoming edges for the convergence nodes, and record the sequence of state nodes traversed by each predecessor path and the feature summary sequence marked on the directed transition edges.

[0025] For each predecessor path of the aggregation node, the directed transformation edges are visited sequentially in the reverse order of the state node sequence. The feature digests marked on the directed transformation edges are extracted, and hash verification is performed to determine their integrity. The correlation between the feature digests of adjacent directed transformation edges is calculated. When the integrity verification of the feature digests of all directed transformation edges passes and the correlation between adjacent feature digests exceeds the preset continuity threshold, the predecessor path backtracking verification is determined to be successful.

[0026] In one optional embodiment, calculating the correlation between feature summaries of adjacent directed transformation edges includes:

[0027] Extract two adjacent state nodes from the state node sequence, obtain the forward directed transition edge and the backward directed transition edge connecting the two state nodes, and extract the preorder feature summary marked on the forward directed transition edge and the postorder feature summary marked on the backward directed transition edge respectively.

[0028] A hash transformation is performed on the preceding feature summary to generate a preceding hash fingerprint, and a hash transformation is performed on the subsequent feature summary to generate a subsequent hash fingerprint. The preceding hash fingerprint is compared and verified with the preceding hash reference field embedded in the subsequent feature summary. The matching similarity between the preceding hash fingerprint and the preceding hash reference field is calculated to determine the correlation.

[0029] When the correlation reaches the preset continuity threshold, the temporal connection relationship between adjacent directed transformation edges is marked, and the temporal connection relationship is recorded in the verification result of the predecessor path.

[0030] In one optional embodiment, when any predecessor path verification fails, constructing a backtracking data chain along the predecessor path, locating the first deviation point, and writing the deviation tracing record into the blockchain includes:

[0031] When the backtracking verification of the current driving path fails, the state node sequence of the failed previous path is extracted. Based on the link switching position identifier associated with the state node sequence, the corresponding numerical change trajectory is retrieved. The numerical difference of the key quality factors and the feature summary of the time-series coding matrix are extracted from the numerical change trajectory. The backtracking data chain is constructed according to the reverse order of the link switching positions.

[0032] Traverse the backtracking data chain, compare the numerical difference of key quality factors with the standard numerical change range in the collaborative conversion rules to calculate the deviation magnitude, perform cross-link correlation analysis on the feature summary of the time-series coding matrix, and determine the deviation position when the deviation magnitude exceeds the preset tolerance range or the cross-link correlation is lower than the preset correlation threshold, and extract the deviation magnitude and correlation decay as deviation features.

[0033] The first deviation position is selected from the starting position of the backtracking data chain to determine the first deviation link. The link switching position identifier and deviation characteristics of the first deviation link are encapsulated into a deviation tracing record, which is then written into the blockchain after being verified by distributed node consensus.

[0034] In one optional embodiment, performing cross-stage correlation analysis on the feature summaries of the time-series coding matrix includes:

[0035] Extract the temporal coding matrix feature summaries corresponding to the switching positions of adjacent links from the backtracking data chain, construct a feature summary pairing sequence, and convert the hash value of each pair of adjacent feature summaries in the feature summary pairing sequence into a coding vector;

[0036] Calculate the cosine of the angle between the encoded vectors of adjacent feature summaries as the local correlation degree. Obtain the local correlation degree of all adjacent link switching positions. Assign attenuation weights according to the reverse position of the link switching position in the backtracking data chain. Sum the local correlation degree with the corresponding attenuation weights to obtain the global cross-link correlation degree.

[0037] When the global cross-stage correlation is lower than the preset correlation threshold, the adjacent feature summary pairs with the lowest local correlation are selected, the corresponding stage switching positions are extracted, the correlation decay positions are determined, the difference between the preset correlation threshold and the global cross-stage correlation is calculated, and the correlation decay amount is determined.

[0038] The blockchain-based school uniform quality information traceability system of this invention includes:

[0039] The data acquisition unit is used to collect process data at each stage of the school uniform production process;

[0040] The parameter extraction unit is used to extract process parameters from each step of the process data, calculate the sensitivity coefficient of the process parameters to the final quality indicators of the school uniform, screen process parameters whose sensitivity coefficient exceeds the preset sensitivity threshold to determine key quality factors, and record the numerical change trajectory of key quality factors in each step.

[0041] The coding verification unit is used to generate a time-series coding matrix based on the numerical change trajectory of key quality factors, match and verify the numerical differences of multiple key quality factors with the collaborative conversion rules, and generate a stage conversion verification code.

[0042] The identifier encapsulation unit is used to generate unique identifiers based on key quality factors and process transformation verification codes, encapsulate process data and unique identifiers into blocks, form a blockchain, and distribute them to distributed nodes.

[0043] The state construction unit is used by each distributed node to construct a quality state transition graph based on the link transition verification code, identify the convergence node with multiple predecessor paths in the quality state transition graph, and perform backtracking verification on the multiple predecessor paths of the convergence node respectively.

[0044] The backtracking verification unit is used to construct a backtracking data chain along the predecessor path when any predecessor path verification fails, locate the first deviation link and write the deviation tracing record into the blockchain.

[0045] The traceability reporting unit is used to receive traceability requests containing unique identifiers, retrieve the numerical change trajectory and verification results of key quality factors from the blockchain, and generate a school uniform quality information traceability report.

[0046] A second aspect of this invention provides a blockchain-based school uniform quality information traceability system, comprising:

[0047] The data acquisition unit is used to collect process data at each stage of the school uniform production process;

[0048] The parameter extraction unit is used to extract process parameters from each step of the process data, calculate the sensitivity coefficient of the process parameters to the final quality indicators of the school uniform, screen process parameters whose sensitivity coefficient exceeds the preset sensitivity threshold to determine key quality factors, and record the numerical change trajectory of key quality factors in each step.

[0049] The coding verification unit is used to generate a time-series coding matrix based on the numerical change trajectory of key quality factors, match and verify the numerical differences of multiple key quality factors with the collaborative conversion rules, and generate a stage conversion verification code.

[0050] The identifier encapsulation unit is used to generate unique identifiers based on key quality factors and process transformation verification codes, encapsulate process data and unique identifiers into blocks, form a blockchain, and distribute them to distributed nodes.

[0051] The state construction unit is used by each distributed node to construct a quality state transition graph based on the link transition verification code, identify the convergence node with multiple predecessor paths in the quality state transition graph, and perform backtracking verification on the multiple predecessor paths of the convergence node respectively.

[0052] The backtracking verification unit is used to construct a backtracking data chain along the predecessor path when any predecessor path verification fails, locate the first deviation link and write the deviation tracing record into the blockchain.

[0053] The traceability reporting unit is used to receive traceability requests containing unique identifiers, retrieve the numerical change trajectory and verification results of key quality factors from the blockchain, and generate a school uniform quality information traceability report.

[0054] A third aspect of the present invention provides an electronic device, comprising:

[0055] processor;

[0056] Memory used to store processor-executable instructions;

[0057] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0058] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0059] In this embodiment of the invention, by calculating the sensitivity coefficient of process parameters to the final quality indicators of school uniforms and screening key quality factors, the traceability process is made more accurate and efficient, avoiding the data redundancy and computational complexity problems caused by traditional full-parameter traceability. A time-series coding matrix is ​​generated based on the numerical change trajectory of key quality factors, and a link conversion verification code is generated by matching and verifying the numerical differences of multiple key quality factors with collaborative conversion rules, establishing an innovative quality data verification mechanism to ensure the authenticity and integrity of traceability data. By constructing a quality state transition diagram and performing backtracking verification on convergence nodes with multiple preceding paths, precise location of quality anomalies in the school uniform production process is achieved, enabling rapid identification of the first deviation link and improving the accuracy and efficiency of quality problem traceability. Blockchain technology is used to store key quality factors and link conversion verification codes, making school uniform quality information tamper-proof and capable of distributed verification, significantly improving the credibility and security of quality information traceability. Attached Figure Description

[0060] Figure 1 This is a flowchart illustrating the blockchain-based school uniform quality information traceability method according to an embodiment of the present invention.

[0061] Figure 2 Generate a logic flowchart for the step-by-step verification code generation. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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.

[0063] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0064] Figure 1 This is a flowchart illustrating the blockchain-based school uniform quality information traceability method according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0065] Collect process data at each stage of the school uniform production process;

[0066] Extract process parameters from each step of the process data, calculate the sensitivity coefficient of the process parameters to the final quality indicators of the school uniform, screen process parameters whose sensitivity coefficients exceed the preset sensitivity threshold to identify key quality factors, and record the numerical change trajectory of key quality factors in each step.

[0067] A time-series coding matrix is ​​generated based on the numerical change trajectory of key quality factors. The numerical differences of multiple key quality factors are matched and verified with the collaborative conversion rules to generate a link conversion verification code.

[0068] A unique identifier is generated based on key quality factors and process transition verification codes. The process data and the unique identifier are encapsulated into blocks to form a blockchain and distributed to distributed nodes.

[0069] Each distributed node constructs a quality state transition graph based on the link transition verification code, identifies the convergence node with multiple predecessor paths in the quality state transition graph, and performs backtracking verification on the multiple predecessor paths of the convergence node.

[0070] When any predecessor path verification fails, a backtracking data chain is constructed along the predecessor path to locate the first deviation link and write the deviation tracing record into the blockchain.

[0071] Receive traceability requests containing unique identifiers, retrieve the numerical change trajectory and verification results of key quality factors from the blockchain, and generate a school uniform quality information traceability report.

[0072] In one optional implementation, process parameters for each stage are extracted from the process data, the sensitivity coefficient of these process parameters to the final quality indicators of the school uniform is calculated, process parameters with sensitivity coefficients exceeding a preset sensitivity threshold are selected to determine key quality factors, and the numerical change trajectory of these key quality factors in each stage is recorded, including:

[0073] The process data is segmented according to the execution sequence of each stage. Process parameters are extracted from each stage segment and the stage location information to which the process parameters belong is marked. A parameter vector set containing process parameter values ​​and stage location information is constructed.

[0074] For each process parameter in the parameter vector set, multiple sets of disturbance values ​​are set within its historical value range. The disturbance values ​​are combined with the measured values ​​of other process parameters in the parameter vector set. According to the quality index calculation rules, the calculation results of the final quality index of the school uniform corresponding to each combination are obtained. The fluctuation range of the final quality index calculation results of the school uniform caused by the disturbance values ​​is statistically analyzed. The fluctuation range is calculated by the ratio of the change amplitude of the disturbance values ​​to obtain the sensitivity coefficient of the process parameter to the final quality index of the school uniform.

[0075] Process parameters with sensitivity coefficients exceeding a preset sensitivity threshold are identified as key quality factors. Measured values ​​of these key quality factors in each stage are extracted and arranged according to stage location information to form a time-series numerical sequence. Stage switching positions are marked in the time-series numerical sequence, and the numerical differences before and after the switching positions are extracted. The numerical differences, along with the corresponding stage switching positions and acquisition times, are encapsulated to form a numerical change trajectory.

[0076] In one specific implementation, process data for the entire school uniform production process is acquired, including execution timestamps and corresponding process parameter information for each stage. The process data is segmented according to the execution sequence of each stage, such as dividing the school uniform production process into fabric inspection, cutting, sewing, ironing, and packaging stages. Process parameters are extracted from each stage segment. For example, fabric strength, shrinkage rate, and color fastness are extracted from the fabric inspection stage; cutting accuracy and cutting speed are extracted from the cutting stage; stitch length, sewing speed, and thread tension are extracted from the sewing stage; temperature, pressure, and ironing time are extracted from the ironing stage; and packaging material performance and sealing strength are extracted from the packaging stage. Simultaneously, the stage location information to which each process parameter belongs is labeled, such as stage number, stage name, and time position within the stage, constructing a parameter vector set containing process parameter values ​​and stage location information. For example, in the production process of a certain batch of school uniforms, the fabric strength value of the fabric in the fabric testing stage is 500N, the shrinkage rate is 3%, the color fastness is level 4, the stage number is E01, and the time position is 0-30 minutes after the start of production; the cutting accuracy of the cutting stage is 0.5mm, the cutting speed is 20m / min, the stage number is E02, and the time position is 31-60 minutes after the start of production.

[0077] For each process parameter in the parameter vector set, multiple sets of perturbation values ​​are set within the historical value range. For example, for fabric strength values ​​within the range of 450N-550N, five sets of perturbation values ​​are set: 450N, 475N, 500N, 525N, and 550N; for shrinkage rates within the range of 2%-4%, five sets of perturbation values ​​are set: 2%, 2.5%, 3%, 3.5%, and 4%. These perturbation values ​​are then combined with the measured values ​​of other process parameters in the parameter vector set. Using pre-established quality index calculation rules, the final quality index of the school uniform corresponding to each combination is calculated. For example, when the fabric strength is 450N and other parameters remain unchanged from the actual measured values, the calculated abrasion resistance index of the school uniform is 85 points; when the fabric strength is 475N, the calculated abrasion resistance index is 87 points; when the fabric strength is 500N, the calculated abrasion resistance index is 90 points; when the fabric strength is 525N, the calculated abrasion resistance index is 92 points; and when the fabric strength is 550N, the calculated abrasion resistance index is 93 points.

[0078] The fluctuation range of the final quality index calculation results of the school uniform caused by the statistical perturbation values ​​was analyzed. For the fabric strength value, a perturbation value from 450N to 550N caused the abrasion resistance index of the school uniform to change from 85 points to 93 points, a fluctuation range of 8 points. The ratio of the fluctuation range to the change amplitude of the perturbation value was calculated to obtain the sensitivity coefficient of the process parameters to the final quality index of the school uniform. The change amplitude of the fabric strength value was 100N, and the sensitivity coefficient was calculated as 8 points / 100N = 0.08 points / N. Similarly, the sensitivity coefficients of other process parameters were calculated, such as the sensitivity coefficient of shrinkage rate (7.5 points / N), the sensitivity coefficient of sewing stitch length (0.5 points / mm), and the sensitivity coefficient of ironing temperature (0.15 points / ℃), etc.

[0079] Process parameters with sensitivity coefficients exceeding a preset sensitivity threshold are identified as key quality factors. If the preset sensitivity threshold is 0.05 points / unit, then parameters such as fabric strength, shrinkage rate, and sewing stitch length all exceed the threshold and are identified as key quality factors for school uniforms. The measured values ​​of these key quality factors at each stage are extracted and arranged according to their stage location to form a time-series numerical sequence. For example, the fabric strength is 500N in the fabric testing stage, 490N after the sewing stage, 485N after the ironing stage, and 485N after the packaging stage, forming a time-series numerical sequence [500N, 490N, 485N, 485N].

[0080] The switching positions of each stage are marked in the time-series numerical sequence, and the numerical differences before and after the switching positions are extracted. The fabric strength value differs by 10N at the switching position from the fabric inspection stage to the sewing stage, by 5N at the switching position from the sewing stage to the ironing stage, and by 0N at the switching position from the ironing stage to the packaging stage. The numerical differences are encapsulated with the corresponding stage switching positions and acquisition times to form a numerical change trajectory. For example, the numerical change trajectory of the fabric strength value is {(E01→E03, 10N, T1), (E03→E04, 5N, T2), (E04→E05, 0N, T3)}, where T1, T2, and T3 are the acquisition times of the corresponding stage switching.

[0081] For identified key quality factors, their complete change trajectory is recorded via blockchain. Each block in the blockchain contains batch information, process information, key quality factor values, change differences, timestamps, and other data. The immutability of the blockchain ensures the traceability of school uniform quality information. When quality issues arise with school uniforms, the blockchain records are queried to find the change trajectory of the corresponding batch's key quality factors, allowing analysis of the process and causes of the quality problem.

[0082] In practical application, a batch of school uniforms was found to have substandard abrasion resistance during quality inspection. A query of blockchain records revealed that the fabric strength value decreased by 10N after the sewing process, exceeding the normal fluctuation range of 5N. Further investigation of the sewing process records showed that the sewing speed was too fast, causing excessive tension on the fabric and resulting in the abnormally low strength value. Based on this finding, the operating parameters of the sewing process were adjusted to prevent similar quality issues from recurring.

[0083] The method described in this embodiment enables traceability of quality information throughout the entire school uniform production process, identifies key factors affecting school uniform quality and their changing trajectories, and provides an effective means for school uniform quality control and problem identification.

[0084] In one optional implementation, a time-series coding matrix is ​​generated based on the numerical change trajectory of key quality factors. The numerical differences of multiple key quality factors are matched and verified with collaborative conversion rules to generate a stage conversion verification code, including:

[0085] From the numerical change trajectory, the numerical differences of multiple key quality factors at the same stage switching position are extracted, and a multi-dimensional numerical difference vector is constructed. The numerical differences of each key quality factor in the multi-dimensional numerical difference vector are weighted and fused according to preset weights to obtain comprehensive difference features. The comprehensive difference features are time-series correlated with the stage switching position and the acquisition time to generate a time-series coding matrix.

[0086] The collaborative conversion rules of multiple quality key factors corresponding to the switching position are retrieved from the preset process conversion rule base. The collaborative conversion rules define the linkage change mode and mutual constraint relationship of multiple quality key factors during the switching of the process. The comprehensive differential features in the time-series coding matrix are matched with the linkage change mode to calculate the similarity. The numerical difference of multiple quality key factors is verified based on the mutual constraint relationship.

[0087] When the similarity exceeds the preset matching threshold and the consistency verification passes, the feature summary of the time-series coding matrix is ​​combined with the stage switching position identifier and the verification pass mark to perform a hash operation to generate a stage conversion verification code.

[0088] In one specific implementation, the numerical differences of multiple key quality factors at the same stage of switching positions are extracted from the aforementioned numerical change trajectory. For example, in the school uniform production process, at the position from the fabric inspection stage to the cutting stage, the numerical difference in fabric strength is extracted as -5N (500N in fabric inspection, 495N after cutting), the numerical difference in fabric shrinkage is +0.2% (3.0% in fabric inspection, 3.2% after cutting), and the numerical difference in color fastness is -0.5 grade (4 grade in fabric inspection, 3.5 grade after cutting). These numerical differences constitute a multidimensional numerical difference vector, represented as [fabric strength difference, fabric shrinkage difference, color fastness difference] = [-5N, +0.2%, -0.5 grade].

[0089] The numerical differences of each key quality factor in the multidimensional numerical difference vector are weighted and fused according to preset weights to obtain a comprehensive difference feature. Based on school uniform quality standards and industry experience, the preset weights for fabric strength, fabric shrinkage rate, and color fastness are 0.5, 0.3, and 0.2, respectively, reflecting the degree of influence of different key quality factors on school uniform quality. The numerical differences of each key quality factor are multiplied by their corresponding weights, and the results are combined into a scalar value to obtain the comprehensive difference feature. For the switch between fabric inspection and cutting stages, the comprehensive difference feature is calculated as: (-5N×0.5) + (0.2%×0.3) + (-0.5 level×0.2) = -2.5 - 0.06 - 0.1 = -2.66, representing the comprehensive impact of this stage switch on school uniform quality.

[0090] The comprehensive differential features are coupled with the switching locations and acquisition times for temporal correlation encoding to generate a temporal encoding matrix. During encoding, the switching location is represented as a combination of the starting and target switching numbers, and the acquisition time is represented as a timestamp relative to the production start time. For example, the switching from fabric inspection (E01) to cutting (E02), with an acquisition time of 60 minutes after production start, is encoded as [E01-E02, -2.66, 60min]. Similarly, the switching from cutting (E02) to sewing (E03) is encoded as [E02-E03, -3.45, 120min]; the switching from sewing (E03) to ironing (E04) is encoded as [E03-E04, -1.78, 180min]; and the switching from ironing (E04) to packaging (E05) is encoded as [E04-E05, -0.92, 210min]. These codes form a time-series coding matrix, which describes the transformation characteristics of key quality factors in the school uniform production process at each stage.

[0091] Retrieve the collaborative conversion rules of multiple quality key factors corresponding to the link switching position from the preset process conversion rule library. In the process conversion rule library, the expected change ranges and mutual relationships of quality key factors during the switching of each link are defined under normal process conditions. For example, the conversion rule from fabric inspection to cutting stipulates that: the normal change range of fabric strength value is from -7N to -3N, the normal change range of fabric shrinkage rate is from +0.1% to +0.3%, and the normal change range of color fastness is from -0.7 level to -0.3 level; at the same time, the change of fabric strength value and color fastness should be positively correlated, that is, the more the fabric strength value decreases, the more obvious the decrease of color fastness. These collaborative conversion rules define the linkage change patterns and mutual constraint relationships of multiple quality key factors during link conversion.

[0092] Perform pattern matching calculation of similarity between the comprehensive differential features in the time series coding matrix and the linkage change pattern. The linkage change pattern is represented as the expected comprehensive differential feature range and change curve. Taking the conversion from fabric inspection to cutting as an example, the expected comprehensive differential feature range is from -3.0 to -2.0, and the actually calculated comprehensive differential feature is -2.66, which falls within the expected range. By calculating the proximity of the actual comprehensive differential feature to the midpoint of the expected range, the similarity is obtained as 0.87 (with a full score of 1).

[0093] Perform consistency verification on the numerical differences of multiple quality key factors based on the mutual constraint relationship. Check whether the numerical differences of each quality key factor meet the preset constraint conditions. For the conversion from fabric inspection to cutting, check whether the fabric strength value difference of -5N is within the expected range of -7N to -3N, whether the fabric shrinkage rate difference of +0.2% is within the expected range of +0.1% to +0.3%, and whether the color fastness difference of -0.5 level is within the expected range of -0.7 level to -0.3 level. At the same time, verify whether the changes of fabric strength value and color fastness conform to the positive correlation relationship, that is, whether the decrease of fabric strength value by 5N corresponding to the decrease of color fastness by about 0.5 level meets the expected ratio. If all tests pass, the consistency verification result is "passed".

[0094] When the similarity exceeds a preset matching threshold and the consistency check passes, the feature summary of the time-series coding matrix, the stage switching position identifier, and the verification pass marker are combined and hashed to generate a stage transition verification code. Assuming the preset matching threshold is 0.80, and the fabric detects a similarity of 0.87 in the cutting stage transition, exceeding the threshold, and the consistency check passes, a stage transition verification code is generated. The feature summary includes the comprehensive differential feature -2.66 and a similarity of 0.87, the stage switching position identifier is E01-E02, and the verification pass marker is "PASS". This information is combined and subjected to a SHA-256 hash operation to obtain a 64-bit hexadecimal string, such as "8f7d56a1e2b9c3d4f8e7a6b5c4d3e2f1a0b9c8d7e6f5a4b3c2d1e0f", which serves as the verification code for the fabric detecting the cutting stage transition.

[0095] These process transition verification codes are recorded on the blockchain, forming a complete quality traceability chain together with the corresponding batch information, process information, and original data of key quality factors. When it is necessary to verify the quality reliability of the school uniform production process, the comprehensive differential characteristics and consistency verification results of the process transition are recalculated, a verification code is generated, and compared with the verification code recorded on the blockchain. If they match, it proves that the quality data of the process transition has not been tampered with and meets the process requirements; if they do not match, it indicates that the quality data is abnormal or has been tampered with.

[0096] The method in this embodiment enables multi-dimensional verification and traceability management of changes in key quality factors during school uniform production, ensuring the authenticity and completeness of school uniform quality information and providing technical support for the accurate identification and accountability of school uniform quality problems.

[0097] like Figure 2 The diagram shown illustrates the logic flowchart for generating the verification code during the transition process.

[0098] In one optional implementation, each distributed node constructs a quality state transition graph based on a link transition verification code, identifies a convergence node with multiple predecessor paths in the quality state transition graph, and performs backtracking verification on the multiple predecessor paths of the convergence node, including:

[0099] Each distributed node receives the stage transition verification code, parses the stage switching position identifier and the feature summary of the timing coding matrix from the stage transition verification code, maps the stage switching position identifier to the state node in the quality state transition graph, establishes directed transition edges between state nodes according to the temporal sequence relationship between stage switching position identifiers, and marks the feature summary of the timing coding matrix on the corresponding directed transition edges to construct the quality state transition graph.

[0100] Traverse all state nodes in the quality state transition graph, count the number of incoming edges for each state node, filter state nodes whose number of incoming edges exceeds a preset convergence threshold to obtain convergence nodes, extract the predecessor paths corresponding to all incoming edges for the convergence nodes, and record the sequence of state nodes traversed by each predecessor path and the feature summary sequence marked on the directed transition edges.

[0101] For each predecessor path of the aggregation node, the directed transformation edges are visited sequentially in the reverse order of the state node sequence. The feature digests marked on the directed transformation edges are extracted, and hash verification is performed to determine their integrity. The correlation between the feature digests of adjacent directed transformation edges is calculated. When the integrity verification of the feature digests of all directed transformation edges passes and the correlation between adjacent feature digests exceeds the preset continuity threshold, the predecessor path backtracking verification is determined to be successful.

[0102] In one specific implementation, the distributed nodes include internal nodes of the school uniform manufacturing enterprise, nodes of the quality inspection agency, nodes of suppliers, and nodes for consumer inquiries. These nodes are connected through a blockchain network and jointly maintain school uniform quality information. Each distributed node receives a process transition verification code, which includes a process switching location identifier, a verification pass marker, and a feature summary of the time-series coding matrix. Taking a batch of school uniform production process as an example, the process transition verification code from the fabric inspection stage to the cutting stage is "8f7d56a1e2b9c3d4f8e7a6b5c4d3e2f1a0b9c8d7e6f5a4b3c2d1e0f", and the process transition verification code from the cutting stage to the sewing stage is "7e6d45b2c3a8d5e6f7g8h9i0j1k2l3m4n5o6p7q8r9s0t1u2v3w4x5y6z", and so on.

[0103] The feature summary of the stage transition position identifier and the temporal coding matrix is ​​parsed from the stage transition CAPTCHA. Using a predefined parsing algorithm, the stage transition position identifier "E01-E02" (from fabric inspection stage to cutting stage) is parsed from the CAPTCHA "8f7d56a1e2b9c3d4f8e7a6b5c4d3e2f1a0b9c8d7e6f5a4b3c2d1e0f", and the feature summary is "[-2.66, 0.87, PASS]", representing the combined differential feature value, similarity, and verification pass mark. Similarly, the link switching position identifier (from cutting to sewing) can be parsed from the verification code "7e6d45b2c3a8d5e6f7g8h9i0j1k2l3m4n5o6p7q8r9s0t1u2v3w4x5y6z" as "E02-E03" and the feature summary is "[-3.45, 0.92, PASS]".

[0104] The process switching locations are mapped to state nodes in the quality state transition diagram. In this diagram, each production process corresponds to a state node. Fabric inspection is mapped to state node S01, cutting to state node S02, sewing to state node S03, ironing to state node S04, and packaging to state node S05. Directed transition edges are established between state nodes based on the temporal relationship between the process switching locations. For example, "E01-E02" represents the switch from fabric inspection to cutting, corresponding to a directed transition edge from state node S01 to S02; "E02-E03" represents the switch from cutting to sewing, corresponding to a directed transition edge from state node S02 to S03.

[0105] The feature summaries of the temporal coding matrix are labeled on the corresponding directed transition edges to construct a complete quality state transition graph. For example, the feature summaries "[-2.66, 0.87, PASS]" are labeled on the directed transition edge from S01 to S02; "[-3.45, 0.92, PASS]" are labeled on the directed transition edge from S02 to S03; "[-1.78, 0.85, PASS]" are labeled on the directed transition edge from S03 to S04; and "[-0.92, 0.90, PASS]" are labeled on the directed transition edge from S04 to S05.

[0106] Traverse all state nodes in the quality state transition diagram and count the number of incoming edges for each state node. In this example, state node S01 has 0 incoming edges, state node S02 has 1, state node S03 has 1, state node S04 has 1, and state node S05 has 1. If there are branches and merging processes during production, such as fabrics coming from different suppliers or batches, or uniform parts being processed in parallel by different processes before sewing and then merged, then the number of incoming edges for some state nodes may be greater than 1.

[0107] State nodes with more than a preset convergence threshold are selected as convergence nodes. Assuming the preset convergence threshold is 2, a state node is considered a convergence node when the number of incoming edges is greater than or equal to 2. In more complex production scenarios, there may be convergence nodes with multiple incoming edges. For example, in the production process of a batch of school uniforms, state node S03 (sewing process) receives incoming edges from two different cutting processes, with a number of 2, exceeding the preset convergence threshold. Therefore, S03 is identified as a convergence node.

[0108] For each sink node, extract the predecessor paths corresponding to all incoming edges, and record the sequence of state nodes traversed by each predecessor path and the feature summary sequence labeled on the directed transformation edge. Taking sink node S03 as an example, there are two predecessor paths: Path 1 traverses the state node sequence [S01, S02, S03], with the corresponding feature summary sequence [[-2.66, 0.87, PASS], [-3.45, 0.92, PASS]]; Path 2 traverses the state node sequence [S01', S02', S03], with the corresponding feature summary sequence [[-2.75, 0.86, PASS], [-3.38, 0.91, PASS]]. Here, S01' and S02' represent the state nodes of another production branch.

[0109] For each predecessor path of the convergence node, the directed transformation edges are visited sequentially in reverse order of the state node sequence. Feature summaries labeled on these edges are extracted and their integrity is verified using hashing. Taking path 1 as an example, starting from S03, the directed transformation edges from S02 to S03 and from S01 to S02 are visited sequentially. The feature summary "[-3.45, 0.92, PASS]" on the directed transformation edge from S02 to S03 is extracted and compared with the original step transformation verification code using hashing to verify the integrity of the feature summary. If the hash comparison matches, it indicates that the feature summary has not been tampered with. Similarly, the integrity of the feature summary "[-2.66, 0.87, PASS]" on the directed transformation edge from S01 to S02 is extracted and verified.

[0110] The correlation between feature summaries of adjacent directed transformation edges is calculated to assess the continuity of the quality transformation process. The correlation calculation is based on the changing trend and magnitude of the comprehensive difference features in adjacent feature summaries. Taking path 1 as an example, the comprehensive difference feature in the feature summaries from S01 to S02 is -2.66, and from S02 to S03 it is -3.45, with a change magnitude of 0.79, both being negative, indicating that the quality parameters are continuously adjusting. Based on a predefined correlation evaluation criterion, the correlation between these two feature summaries is calculated to be 0.88 (out of 1).

[0111] The backtracking verification of the predecessor path is considered successful when the integrity verification of the feature summaries of all directed transformation edges passes and the correlation between adjacent feature summaries exceeds a preset continuity threshold. Assuming the preset continuity threshold is 0.75, the correlation between adjacent feature summaries in path 1 is 0.88, exceeding the threshold, and the integrity verification of all feature summaries passes; therefore, the backtracking verification of path 1 is successful. Similarly, the backtracking verification results of path 2 are analyzed. When the backtracking verification of all predecessor paths is successful, it indicates that all input quality states of the sink node are reliable and consistent.

[0112] By constructing the aforementioned quality state transition diagram and using the backtracking verification method, the quality state transition process throughout the entire school uniform production process can be effectively tracked, ensuring the authenticity and completeness of quality information. When quality problems occur in school uniforms, the quality state transition diagram can be traced back from the point of problem to pinpoint the specific stage and cause of the quality problem, providing a basis for quality improvement and accountability. It can also identify abnormal breakpoints and inconsistencies in the quality state transition process, preventing quality information from being tampered with or falsified.

[0113] In one alternative implementation, calculating the correlation between the feature summaries of adjacent directed transformation edges includes:

[0114] Extract two adjacent state nodes from the state node sequence, obtain the forward directed transition edge and the backward directed transition edge connecting the two state nodes, and extract the preorder feature summary marked on the forward directed transition edge and the postorder feature summary marked on the backward directed transition edge respectively.

[0115] A hash transformation is performed on the preceding feature summary to generate a preceding hash fingerprint, and a hash transformation is performed on the subsequent feature summary to generate a subsequent hash fingerprint. The preceding hash fingerprint is compared and verified with the preceding hash reference field embedded in the subsequent feature summary. The matching similarity between the preceding hash fingerprint and the preceding hash reference field is calculated to determine the correlation.

[0116] When the correlation reaches the preset continuity threshold, the temporal connection relationship between adjacent directed transformation edges is marked, and the temporal connection relationship is recorded in the verification result of the predecessor path.

[0117] In one specific implementation, during the backtracking verification process of the school uniform quality state transition diagram, two adjacent state nodes are extracted from the state node sequence, and the forward and backward directed transition edges connecting these two state nodes are obtained. Taking a batch of school uniform production process as an example, adjacent state nodes S02 and S03 are extracted from the state node sequence [S01, S02, S03, S04, S05], where S02 represents the cutting stage and S03 represents the sewing stage. The forward directed transition edge connecting S02 and S03 is the transition edge from S01 (fabric inspection stage) to S02, and the backward directed transition edge is the transition edge from S02 to S03.

[0118] The preorder feature summaries annotated on the forward directed transformation edges and the postorder feature summaries annotated on the backward directed transformation edges are extracted separately. The preorder feature summaries annotated on the forward directed transformation edges S01 to S02 are "[-2.66, 0.87, PASS, FP98A7C5]", where -2.66 is the comprehensive difference feature value, 0.87 is the similarity, PASS is the verification pass mark, and FP98A7C5 is the hash fingerprint identifier of the forward directed transformation edge. The post-order feature summary labeled on the backward directed transformation edges S02 to S03 is "[-3.45, 0.92, PASS, FP4D2E3F, FP98A7C5]", where -3.45 is the comprehensive difference feature value, 0.92 is the similarity, PASS is the verification pass mark, FP4D2E3F is the hash fingerprint identifier of the backward directed transformation edge, and FP98A7C5 is the embedded predecessor hash reference field, which is used to reference the hash fingerprint of the predecessor feature summary.

[0119] A hash transformation is performed on the preceding feature digest to generate a preceding hash fingerprint. The core content "[-2.66, 0.87, PASS, FP98A7C5]" in the preceding feature digest "[-2.66, 0.87, PASS]" is hashed using SHA-256 to obtain the preceding hash fingerprint "98A7C5D6E4F3A2B1C0D9E8F7A6B5C4D3". This hash fingerprint uniquely identifies the content of the preceding feature digest and is stored in the preceding feature digest as a simplified version (using the first 8 bits of the hash value) in the format "FP98A7C5".

[0120] A hash transformation is performed on the subsequent feature digest to generate a subsequent hash fingerprint. The core content "[-3.45, 0.92, PASS, FP4D2E3F, FP98A7C5]" is hashed using SHA-256 to obtain the subsequent hash fingerprint "4D2E3F1A2B3C4D5E6F7A8B9C0D1E2F3". This hash fingerprint uniquely identifies the content of the subsequent feature digest and is stored as a simplified version in the subsequent feature digest in the form of "FP4D2E3F".

[0121] The preceding hash fingerprint is compared and verified with the preceding hash reference field embedded in the subsequent feature digest. The simplified version of the preceding hash fingerprint, "98A7C5", is compared with the preceding hash reference field "FP98A7C5" in the subsequent feature digest. In this example, the simplified version of the preceding hash fingerprint is indeed "98A7C5", which perfectly matches the preceding hash reference field "FP98A7C5" in the subsequent feature digest. This indicates that the subsequent feature digest does indeed reference the preceding feature digest, establishing a temporal link between the two.

[0122] The similarity between the preceding hash fingerprint and the preceding hash reference field is calculated to determine the correlation. The similarity is 1.0 when the preceding hash reference field completely contains the simplified version of the preceding hash fingerprint; the similarity is between 0 and 1 when there is a partial match or hash collision; and the similarity is 0 when there is no match at all. In this example, the simplified version of the preceding hash fingerprint "98A7C5" completely matches the preceding hash reference field "FP98A7C5", therefore the similarity is 1.0, indicating a complete temporal connection between the two feature summaries.

[0123] When the correlation degree reaches the preset continuity threshold, it indicates that there is a temporal connection between adjacent directed transition edges. Assuming the preset continuity threshold is 0.8, the correlation degree of 1.0 in this example exceeds the threshold. Therefore, it indicates that there is a temporal connection between the directed transition edge from S01 to S02 and the directed transition edge from S02 to S03, indicating that the quality state transition of these two production links is continuous and consistent.

[0124] The temporal connections are recorded in the verification results of the predecessor paths. For the predecessor path represented by the state node sequence [S01, S02, S03, S04, S05], the verification results record that there are temporal connections between S01-S02 and S02-S03, between S02-S03 and S03-S04, and between S03-S04 and S04-S05. These recorded temporal connections constitute a complete quality state transition chain, ensuring the continuity and integrity of quality information during the school uniform production process.

[0125] In practical applications, multiple parallel precursor paths exist. For example, in the production process of another batch of school uniforms, there are two precursor paths: path 1 traverses the state node sequence [S01, S02, S03], and path 2 traverses the state node sequence [S01', S02', S03]. For path 1, the temporal connection between S01-S02 and S02-S03 is verified; for path 2, the temporal connection between S01'-S02' and S02'-S03 is verified. When the temporal connection verification of all paths passes, it indicates that the quality state transitions of all production branches are continuous and consistent.

[0126] In some cases, there is a partial match between the preceding hash fingerprint and the preceding hash reference field. For example, in a production batch, the simplified version of the preceding hash fingerprint is "A1B2C3", while the preceding hash reference field in the subsequent feature summary is "FPA1B2D4". There is a partial match between the two, and the calculated similarity is 0.75. If the preset continuity threshold is 0.7, it will still mark the temporal connection between adjacent directed transformation edges, but in the verification results, it will be marked as "partial connection", indicating a slight data inconsistency or error.

[0127] Different preset continuity thresholds are set for different types of school uniform products. For high-standard school uniform products, a higher preset continuity threshold, such as 0.9, is set to ensure a high degree of continuity in quality status transitions; for ordinary school uniform products, a moderate preset continuity threshold, such as 0.7, is set to allow for a certain degree of data fluctuation. By adjusting the preset continuity thresholds, the strictness and adaptability of quality traceability can be flexibly balanced.

[0128] The blockchain storage of school uniform quality information not only records the quality status data of each stage but also the temporal connections between adjacent stages. When querying the quality traceability information of a batch of school uniforms, the complete state transition path and temporal connection verification results are retrieved from the blockchain, clearly showing the quality status changes of the school uniform from fabric to finished product, and ensuring the continuity and consistency of these status changes. When a quality anomaly is found in a certain stage, the cause can be traced back to the stage through the temporal connection relationship, achieving precise location of the quality problem.

[0129] The aforementioned time-series connection verification mechanism not only ensures the integrity and continuity of quality information during school uniform production but also effectively prevents the tampering and forgery of quality data, providing technical assurance for reliable traceability of school uniform quality. This hash-link-based time-series verification method, combined with the immutability of blockchain, constructs a trustworthy and transparent school uniform quality information traceability system.

[0130] In one optional implementation, when any predecessor path verification fails, a backtracking data chain is constructed along the predecessor path to locate the first deviation and write the deviation tracing record into the blockchain, including:

[0131] When the backtracking verification of the current driving path fails, the state node sequence of the failed previous path is extracted. Based on the link switching position identifier associated with the state node sequence, the corresponding numerical change trajectory is retrieved. The numerical difference of the key quality factors and the feature summary of the time-series coding matrix are extracted from the numerical change trajectory. The backtracking data chain is constructed according to the reverse order of the link switching positions.

[0132] Traverse the backtracking data chain, compare the numerical difference of key quality factors with the standard numerical change range in the collaborative conversion rules to calculate the deviation magnitude, perform cross-link correlation analysis on the feature summary of the time-series coding matrix, and determine the deviation position when the deviation magnitude exceeds the preset tolerance range or the cross-link correlation is lower than the preset correlation threshold, and extract the deviation magnitude and correlation decay as deviation features.

[0133] The first deviation position is selected from the starting position of the backtracking data chain to determine the first deviation link. The link switching position identifier and deviation characteristics of the first deviation link are encapsulated into a deviation tracing record, which is then written into the blockchain after being verified by distributed node consensus.

[0134] In one specific implementation, when the backtracking verification of the preceding path fails, the state node sequence of the failed preceding path is extracted. Taking a batch of school uniform production process as an example, there exists a preceding path passing through the state node sequence [S01, S02, S03, S04, S05], corresponding to the fabric inspection, cutting, sewing, ironing, and packaging stages, respectively. During the backtracking verification process, it was found that the temporal connection relationship between the directed transition edge from S03 to S04 and the directed transition edge from S04 to S05 failed verification due to abnormal or tampered quality data in a certain stage. The complete state node sequence [S01, S02, S03, S04, S05] of the failed preceding path is extracted for subsequent deviation tracing analysis.

[0135] Based on the process transition location identifiers associated with the state node sequence, the corresponding numerical change trajectory is retrieved. State node S01 corresponds to the fabric inspection process, S02 to the cutting process, and the process transition location identifier is "E01-E02"; state node S02 corresponds to the cutting process, S03 to the sewing process, and the process transition location identifier is "E02-E03"; state node S03 corresponds to the sewing process, S04 to the ironing process, and the process transition location identifier is "E03-E04"; state node S04 corresponds to the ironing process, S05 to the packaging process, and the process transition location identifier is "E04-E05". Using these process transition location identifiers, the corresponding numerical change trajectory is retrieved from the blockchain or distributed database.

[0136] Numerical differences and feature summaries of the temporal coding matrix for key quality factors were extracted from the numerical change trajectories. In the numerical change trajectory at the stage transition position "E01-E02", the numerical difference in fabric strength was -5N, the numerical difference in fabric shrinkage was +0.2%, and the numerical difference in color fastness was -0.5 grade. The feature summary of the temporal coding matrix was "[-2.66, 0.87, PASS, FP98A7C5]". Similarly, corresponding numerical differences and feature summaries were extracted from the numerical change trajectories at other stage transition positions. In the process switching position "E03-E04", it was found that the difference in fabric strength value was -12N (far greater than the normal expected range of -3N to -7N), the difference in fabric shrinkage rate was +0.5%, the difference in color fastness was -1.2, and the feature summary of the time-series coding matrix was "[-7.33, 0.65, FAIL, FP6B7C8D, FP3A4B5C]", where "FAIL" indicates that the verification failed.

[0137] A backtracking data chain is constructed by reversing the order of process switching positions. Starting from the packaging stage, the process switching positions and their corresponding critical quality factor numerical differences and feature summaries are arranged in reverse order to construct the backtracking data chain: [("E04-E05", [-0.92, 0.90, PASS, FP9D8E7F, FP6B7C8D]), ("E03-E04", [-7.33, 0.65, FAIL, FP6B7C8D, FP3A4B5C]), ("E02-E03", [-3.45, 0.92, PASS, FP3A4B5C, FP98A7C5]), ("E01-E02", [-2.66, 0.87, PASS, FP98A7C5])]. This backtracking data chain is arranged in reverse chronological order, facilitating tracing from the final product back to the original process and identifying the origin of quality problems.

[0138] The data chain is traversed and traced back to calculate the deviation range by comparing the numerical differences of key quality factors with the standard numerical variation range in the collaborative conversion rules. For the link switching position "E01-E02", the numerical difference of fabric strength value -5N is within the standard numerical variation range [-7N, -3N], and the deviation range is 0%; the numerical difference of fabric shrinkage rate +0.2% is within the standard numerical variation range [+0.1%, +0.3%], and the deviation range is 0%; the numerical difference of color fastness grade -0.5 is within the standard numerical variation range [-0.7 grade, -0.3 grade], and the deviation range is 0%. For the switching positions "E03-E04", the difference in fabric strength value of -12N exceeds the standard value variation range [-7N, -3N], with a deviation of 71.4% (calculation method: |actual value - midpoint of interval| / half width of interval = (|-12-(-5)| / 2 = 7 / 2 = 3.5, deviation range = 3.5 / 4.9 × 100% = 71.4%); the difference in fabric shrinkage rate of +0.5% exceeds the standard value variation range [+0.1%, +0.3%], with a deviation of 100%; the difference in color fastness of -1.2 grade exceeds the standard value variation range [-0.7 grade, -0.3 grade], with a deviation of 150%.

[0139] Cross-stage correlation analysis was performed on the feature summaries of the time-series coding matrix. The matching between the predecessor hash reference field and the predecessor hash fingerprint in the feature summaries of adjacent stages was analyzed. In stage switching positions "E02-E03", the predecessor hash reference field "FP98A7C5" of the feature summary completely matches the hash fingerprint of the feature summaries in stage switching positions "E01-E02", with a correlation score of 1.0. In stage switching positions "E03-E04", the predecessor hash reference field "FP3A4B5C" of the feature summary completely matches the hash fingerprint of the feature summaries in stage switching positions "E02-E03", with a correlation score of 1.0. In stage switching positions "E04-E05", the predecessor hash reference field "FP6B7C8D" of the feature summary completely matches the hash fingerprint of the feature summaries in stage switching positions "E03-E04", with a correlation score of 1.0. Although the hash reference relationship is correct, the similarity value in the feature summary of the link switching position "E03-E04" is only 0.65, which is lower than the expected 0.80 or above.

[0140] When the deviation exceeds the preset tolerance range or the cross-stage correlation is lower than the preset correlation threshold, the deviation location is determined, and the deviation magnitude and correlation attenuation are extracted as deviation features. Assume the preset tolerance range is 30% and the preset correlation threshold is 0.75. In the stage transition position "E03-E04", the deviation magnitude of the fabric strength value is 71.4%, exceeding the preset tolerance range of 30%; the deviation magnitude of the fabric shrinkage rate is 100%, exceeding the preset tolerance range; and the deviation magnitude of the color fastness is 150%, exceeding the preset tolerance range. Simultaneously, the similarity value of the feature summary (0.65) is lower than the preset correlation threshold of 0.75. Therefore, the stage transition position "E03-E04" is determined as the deviation location. The deviation magnitude of the fabric strength value (71.4%), the deviation magnitude of the fabric shrinkage rate (100%), the deviation magnitude of the color fastness (150%), and the similarity attenuation of 0.15 (calculated as expected similarity 0.80 minus actual similarity 0.65) are extracted as deviation features.

[0141] By filtering the data from the beginning of the backtracking data chain, the first deviation point was identified. Starting from the beginning of the backtracking data chain, "E04-E05", the first deviation point was found to be "E03-E04". Therefore, the ironing stage (E04) was determined to be the first deviation point. This indicates that the school uniform quality problem first occurred in the ironing stage. Due to excessively high temperatures or improper operation during the ironing process, key quality indicators such as fabric strength, shrinkage rate, and color fastness experienced abnormal changes.

[0142] The initial deviation point's switching location identifier and deviation characteristics are encapsulated into a deviation traceability record. This record includes the switching location identifier "E03-E04," and deviation characteristics such as fabric strength deviation of 71.4%, fabric shrinkage deviation of 100%, colorfastness deviation of 150%, and similarity attenuation of 0.15. This comprehensive information records the specific manifestations and extent of the quality anomaly, providing a basis for resolving quality issues and assigning responsibility.

[0143] After verification through distributed node consensus, deviation traceability records are written to the blockchain. These distributed nodes include internal nodes of the school uniform manufacturer, quality inspection agency nodes, supplier nodes, and consumer query nodes. These nodes verify the validity of the deviation traceability records through a consensus mechanism, ensuring the accuracy and credibility of the records. Once verified, the deviation traceability record is written to the blockchain as a permanent record of quality traceability, which anyone can query and verify, but which cannot be tampered with. Blockchain technology ensures the transparency, immutability, and verifiability of quality traceability.

[0144] The method described in this embodiment can not only accurately pinpoint the initial occurrence of quality problems in school uniforms, but also quantify the specific manifestations and extent of quality anomalies, providing strong support for resolving quality issues and assigning responsibility. This blockchain-based quality traceability method effectively improves the quality control level of school uniform products and safeguards the health and safety of students.

[0145] In one optional implementation, cross-stage correlation analysis of the feature summaries of the time-series coding matrix includes:

[0146] Extract the temporal coding matrix feature summaries corresponding to the switching positions of adjacent links from the backtracking data chain, construct a feature summary pairing sequence, and convert the hash value of each pair of adjacent feature summaries in the feature summary pairing sequence into a coding vector;

[0147] Calculate the cosine of the angle between the encoded vectors of adjacent feature summaries as the local correlation degree. Obtain the local correlation degree of all adjacent link switching positions. Assign attenuation weights according to the reverse position of the link switching position in the backtracking data chain. Sum the local correlation degree with the corresponding attenuation weights to obtain the global cross-link correlation degree.

[0148] When the global cross-stage correlation is lower than the preset correlation threshold, the adjacent feature summary pairs with the lowest local correlation are selected, the corresponding stage switching positions are extracted, the correlation decay positions are determined, the difference between the preset correlation threshold and the global cross-stage correlation is calculated, and the correlation decay amount is determined.

[0149] In one specific implementation, the temporal coding matrix feature summaries corresponding to the switching positions of adjacent links are extracted from the backtracking data chain, and a feature summary pairing sequence is constructed. Taking the production process of a certain batch of school uniforms as an example, the backtracking data chain includes the link switching positions "E01-E02" (fabric detection to cutting), "E02-E03" (cutting to sewing), "E03-E04" (sewing to ironing), and "E04-E05" (ironing to packaging). The corresponding temporal coding matrix feature summaries are "[-2.66, 0.87, PASS, FP98A7C5]", "[-3.45, 0.92, PASS, FP3A4B5C, FP98A7C5]", "[-7.33, 0.65, FAIL, FP6B7C8D, FP3A4B5C]" and "[-0.92, 0.90, PASS, FP9D8E7F, FP6B7C8D]". The constructed feature summary pairing sequence is: [("[-2.66, 0.87, PASS, FP98A7C5]", "[-3.45, 0.92, PASS, FP3A4B5C, FP98A7C5]"), ("[-3.45, 0.92, PASS, FP3A4B5C, FP98A7C5]", "[-7.33, 0.65, FAIL, FP6B7C8D, FP3A4B5C]"), ("[-7.33, 0.65, FAIL, FP6B7C8D, FP3A4B5C]", "[-0.92, 0.90, PASS, FP9D8E7F, FP6B7C8D]")]. Each pair of paired feature summaries represents the quality state transition relationship between adjacent stages.

[0150] For each pair of adjacent feature digests in the feature digest pairing sequence, the hash value of the feature digest is converted into an encoding vector. A SHA-256 hash operation is performed on the feature digest, and the first 32 bits of the hash value are taken. Each 4 bits are treated as a hexadecimal number and converted to the corresponding decimal number, forming an 8-dimensional encoding vector. For example, the first 32 bits of the hash value of the feature digest "[-2.66, 0.87, PASS, FP98A7C5]" are "98A7C5D6E4F3A2B1", which is converted into an 8-dimensional encoding vector [9, 8, 10, 7, 12, 5, 13, 6]; the first 32 bits of the hash value of the feature digest "[-3.45, 0.92, PASS, FP3A4B5C, FP98A7C5]" are "3A4B5C6D7E8F9A0B", which is converted into an 8-dimensional encoding vector [3, 10, 4, 11, 5, 12, 6, 13]. Similarly, the encoded vectors for other feature summaries are calculated. This encoded vector conversion method maps the string-based feature summaries to numerical vectors, facilitating vector computation.

[0151] The cosine of the angle between the encoded vectors of adjacent feature summaries is calculated as the local correlation. The cosine of the vector angle is obtained by dividing the dot product of the two vectors by the product of their respective magnitudes. The result is between -1 and 1; the closer the value is to 1, the more similar the two vectors are. For example, the encoded vectors [9, 8, 10, 7, 12, 5, 13, 6] for the feature summary "[-2.66, 0.87, PASS, FP98A7C5]" and [3, 10, 4, 11, 5, 12, 6, 13] for the feature summary "[-3.45, 0.92, PASS, FP3A4B5C, FP98A7C5]" have a calculated cosine of 0.85, indicating a high similarity. This value is used as the local correlation between the stage transition position "E01-E02" and "E02-E03".

[0152] Obtain the local correlation of all adjacent link switching positions. Using the method described above, the local correlation of the link switching position "E01-E02" to "E02-E03" is calculated to be 0.85, the local correlation of "E02-E03" to "E03-E04" is 0.56 (indicating a significant decrease in correlation), and the local correlation of "E03-E04" to "E04-E05" is 0.74. These local correlation values ​​reflect the coherence of quality state transitions between adjacent links; higher values ​​indicate better coherence, while lower values ​​indicate quality anomalies or data tampering.

[0153] The weights are assigned based on the reverse position of the process switching location in the backtracking data chain. Processes earlier in the reverse sequence have a greater impact on the current quality status and are therefore assigned higher weights; processes later in the reverse sequence have a relatively smaller impact and are assigned lower weights. For example, starting the backtracking from the packaging process, the reverse position of the process switching location "E04-E05" is 1, assigned a weight of 0.4; "E03-E04" is 2, assigned a weight of 0.3; "E02-E03" is 3, assigned a weight of 0.2; and "E01-E02" is 4, assigned a weight of 0.1. This weight allocation reflects the degree of influence of each process on the final product quality, and the sum of the weights is 1.

[0154] The global cross-process correlation degree is obtained by weighting and summing the local correlation degrees with their corresponding attenuation weights. The local correlation degree from "E01-E02" to "E02-E03" (0.85) multiplied by a weight of 0.2 yields a weighted correlation degree of 0.17; the local correlation degree from "E02-E03" to "E03-E04" (0.56) multiplied by a weight of 0.3 yields a weighted correlation degree of 0.168; and the local correlation degree from "E03-E04" to "E04-E05" (0.74) multiplied by a weight of 0.4 yields a weighted correlation degree of 0.296. These weighted correlation degrees are then summed to obtain the global cross-process correlation degree of 0.634. This value represents the overall coherence of the quality information chain throughout the entire production process; a higher value indicates better coherence.

[0155] When the global cross-stage correlation is lower than a preset correlation threshold, the adjacent feature summary pairs with the lowest local correlation are selected, and the corresponding stage switching positions are extracted to determine the correlation decay positions. Assuming the preset correlation threshold is 0.7, the global cross-stage correlation of 0.634 in this example is lower than the threshold, indicating insufficient continuity of quality information. By comparing the local correlation of each stage switching position, the local correlation of "E02-E03" to "E03-E04" is found to be the lowest at 0.56. Therefore, the stage switching position "E03-E04" is determined to be the correlation decay position. This indicates that there are quality anomalies or data tampering in the process from sewing to ironing.

[0156] The difference between the preset correlation threshold and the global cross-stage correlation degree is calculated to determine the correlation degree attenuation. Subtracting the global cross-stage correlation degree of 0.634 from the preset correlation threshold of 0.7 yields a correlation degree attenuation of 0.066. This value quantifies the degree of break in the quality information chain; a larger value indicates a more severe break. A correlation degree attenuation of 0.066 indicates a slight break in quality information, requiring attention but not yet reaching a level that seriously affects quality.

[0157] In practical applications, the global cross-process correlation analysis adjusts parameters according to the characteristics of different types of school uniform products. For high-standard school uniform products, a higher preset correlation threshold is set, such as 0.8, to ensure a high degree of continuity in the quality information chain; for ordinary school uniform products, a moderate preset correlation threshold is set, such as 0.65, allowing for a certain degree of data fluctuation. By adjusting the preset correlation threshold, the strictness and adaptability of quality traceability are flexibly balanced.

[0158] The results of global cross-process correlation analysis, combined with other deviation tracing indicators, can more accurately pinpoint quality issues in the school uniform production process. For example, if the correlation decay position coincides with the deviation position of the difference in the values ​​of key quality factors, both pointing to the process switching position "E03-E04," it further confirms the possibility of quality problems in the ironing process. Through this multi-dimensional quality traceability analysis, not only can the location of quality problems be determined, but also the nature and severity of the problems can be described, providing targeted guidance for quality improvement.

[0159] The results of the global cross-process correlation analysis are recorded on the blockchain, forming a complete quality information chain together with other quality traceability data. When it is necessary to query the quality status of a batch of school uniforms, the correlation analysis results are retrieved from the blockchain to understand the consistency of quality information and identify weaknesses. This information is of great value to manufacturers to improve processes, quality inspection agencies to supervise quality, and consumers to understand product history.

[0160] Through the global cross-process correlation analysis mechanism in this embodiment, weak links in the quality information chain during the school uniform production process can be accurately identified, providing a scientific basis for locating and solving quality problems, improving the refinement and traceability of school uniform quality management, and ensuring that students wear safe and healthy school uniforms.

[0161] The blockchain-based school uniform quality information traceability system of this invention includes:

[0162] The data acquisition unit is used to collect process data at each stage of the school uniform production process;

[0163] The parameter extraction unit is used to extract process parameters from each step of the process data, calculate the sensitivity coefficient of the process parameters to the final quality indicators of the school uniform, screen process parameters whose sensitivity coefficient exceeds the preset sensitivity threshold to determine key quality factors, and record the numerical change trajectory of key quality factors in each step.

[0164] The coding verification unit is used to generate a time-series coding matrix based on the numerical change trajectory of key quality factors, match and verify the numerical differences of multiple key quality factors with the collaborative conversion rules, and generate a stage conversion verification code.

[0165] The identifier encapsulation unit is used to generate unique identifiers based on key quality factors and process transformation verification codes, encapsulate process data and unique identifiers into blocks, form a blockchain, and distribute them to distributed nodes.

[0166] The state construction unit is used by each distributed node to construct a quality state transition graph based on the link transition verification code, identify the convergence node with multiple predecessor paths in the quality state transition graph, and perform backtracking verification on the multiple predecessor paths of the convergence node respectively.

[0167] The backtracking verification unit is used to construct a backtracking data chain along the predecessor path when any predecessor path verification fails, locate the first deviation link and write the deviation tracing record into the blockchain.

[0168] The traceability reporting unit is used to receive traceability requests containing unique identifiers, retrieve the numerical change trajectory and verification results of key quality factors from the blockchain, and generate a school uniform quality information traceability report.

[0169] A third aspect of the present invention provides an electronic device, comprising:

[0170] processor;

[0171] Memory used to store processor-executable instructions;

[0172] The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0173] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0174] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0175] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A blockchain-based method for tracing school uniform quality information, characterized in that, include: Collect process data at each stage of the school uniform production process; Extract process parameters from each step of the process data, calculate the sensitivity coefficient of the process parameters to the final quality indicators of the school uniform, screen process parameters whose sensitivity coefficients exceed the preset sensitivity threshold to identify key quality factors, and record the numerical change trajectory of key quality factors in each step. A time-series coding matrix is ​​generated based on the numerical change trajectory of key quality factors. The numerical differences of multiple key quality factors are matched and verified with the collaborative conversion rules to generate a stage conversion verification code. The process includes: extracting the numerical differences of multiple key quality factors at the same stage switching position from the numerical change trajectory, constructing a multi-dimensional numerical difference vector, weighting and fusing the numerical differences of each key quality factor in the multi-dimensional numerical difference vector according to preset weights to obtain a comprehensive difference feature, and performing time-series correlation coding with the stage switching position and the acquisition time to generate a time-series coding matrix. The collaborative conversion rules of multiple quality key factors corresponding to the switching position are retrieved from the preset process conversion rule base. The collaborative conversion rules define the linkage change mode and mutual constraint relationship of multiple quality key factors during the switching of the process. The comprehensive differential features in the time-series coding matrix are matched with the linkage change mode to calculate the similarity. The numerical difference of multiple quality key factors is verified based on the mutual constraint relationship. When the similarity exceeds the preset matching threshold and the consistency verification passes, the feature summary of the time-series coding matrix is ​​combined with the stage switching position identifier and the verification pass mark to perform a hash operation to generate a stage conversion verification code; A unique identifier is generated based on key quality factors and process transition verification codes. The process data and the unique identifier are encapsulated into blocks to form a blockchain and distributed to distributed nodes. Each distributed node constructs a quality state transition graph based on the stage transition verification code, identifies the convergence node with multiple predecessor paths in the quality state transition graph, and performs backtracking verification on the multiple predecessor paths of the convergence node. This includes: each distributed node receiving the stage transition verification code, parsing the stage switching position identifier and the feature summary of the timing coding matrix from the stage transition verification code, mapping the stage switching position identifier to the state node in the quality state transition graph, establishing directed transition edges between state nodes according to the temporal sequence relationship between stage switching position identifiers, and marking the feature summary of the timing coding matrix on the corresponding directed transition edges to construct the quality state transition graph. Traverse all state nodes in the quality state transition graph, count the number of incoming edges for each state node, filter state nodes whose number of incoming edges exceeds a preset convergence threshold to obtain convergence nodes, extract the predecessor paths corresponding to all incoming edges for the convergence nodes, and record the sequence of state nodes traversed by each predecessor path and the feature summary sequence marked on the directed transition edges. For each predecessor path of the aggregation node, the directed transformation edges are visited in reverse order of the state node sequence. The feature digests marked on the directed transformation edges are extracted, and hash verification is performed to determine the integrity. The correlation between the feature digests of adjacent directed transformation edges is calculated. When the integrity verification of the feature digests of all directed transformation edges passes and the correlation between adjacent feature digests exceeds the preset continuity threshold, the predecessor path backtracking verification is determined to be successful. When any predecessor path verification fails, a backtracking data chain is constructed along the predecessor path to locate the first deviation link and write the deviation tracing record into the blockchain. Receive traceability requests containing unique identifiers, retrieve the numerical change trajectory and verification results of key quality factors from the blockchain, and generate a school uniform quality information traceability report.

2. The method according to claim 1, characterized in that, Process parameters for each stage are extracted from the process data. The sensitivity coefficients of these parameters to the final quality indicators of the school uniforms are calculated. Process parameters with sensitivity coefficients exceeding a preset sensitivity threshold are selected to identify key quality factors. The numerical changes of these key quality factors at each stage are recorded, including: The process data is segmented according to the execution sequence of each stage. Process parameters are extracted from each stage segment and the stage location information to which the process parameters belong is marked. A parameter vector set containing process parameter values ​​and stage location information is constructed. For each process parameter in the parameter vector set, multiple sets of disturbance values ​​are set within its historical value range. The disturbance values ​​are combined with the measured values ​​of other process parameters in the parameter vector set. According to the quality index calculation rules, the calculation results of the final quality index of the school uniform corresponding to each combination are obtained. The fluctuation range of the final quality index calculation results of the school uniform caused by the disturbance values ​​is statistically analyzed. The fluctuation range is calculated by the ratio of the change amplitude of the disturbance values ​​to obtain the sensitivity coefficient of the process parameter to the final quality index of the school uniform. Process parameters with sensitivity coefficients exceeding a preset sensitivity threshold are identified as key quality factors. Measured values ​​of these key quality factors in each stage are extracted and arranged according to stage location information to form a time-series numerical sequence. Stage switching positions are marked in the time-series numerical sequence, and the numerical differences before and after the switching positions are extracted. The numerical differences, along with the corresponding stage switching positions and acquisition times, are encapsulated to form a numerical change trajectory.

3. The method according to claim 1, characterized in that, Calculating the correlation between feature summaries of adjacent directed transformation edges includes: Extract two adjacent state nodes from the state node sequence, obtain the forward directed transition edge and the backward directed transition edge connecting the two state nodes, and extract the preorder feature summary marked on the forward directed transition edge and the postorder feature summary marked on the backward directed transition edge respectively. A hash transformation is performed on the preceding feature summary to generate a preceding hash fingerprint, and a hash transformation is performed on the subsequent feature summary to generate a subsequent hash fingerprint. The preceding hash fingerprint is compared and verified with the preceding hash reference field embedded in the subsequent feature summary. The matching similarity between the preceding hash fingerprint and the preceding hash reference field is calculated to determine the correlation. When the correlation reaches the preset continuity threshold, the temporal connection relationship between adjacent directed transformation edges is marked, and the temporal connection relationship is recorded in the verification result of the predecessor path.

4. The method according to claim 1, characterized in that, When any preceding path verification fails, a backtracking data chain is constructed along the preceding path to locate the first deviation and write the deviation tracing record to the blockchain, including: When the backtracking verification of the current driving path fails, the state node sequence of the failed previous path is extracted. Based on the link switching position identifier associated with the state node sequence, the corresponding numerical change trajectory is retrieved. The numerical difference of the key quality factors and the feature summary of the time-series coding matrix are extracted from the numerical change trajectory. The backtracking data chain is constructed according to the reverse order of the link switching positions. Traverse the backtracking data chain, compare the numerical difference of key quality factors with the standard numerical change range in the collaborative conversion rules to calculate the deviation magnitude, perform cross-link correlation analysis on the feature summary of the time-series coding matrix, and determine the deviation position when the deviation magnitude exceeds the preset tolerance range or the cross-link correlation is lower than the preset correlation threshold, and extract the deviation magnitude and correlation decay as deviation features. The first deviation position is selected from the starting position of the backtracking data chain to determine the first deviation link. The link switching position identifier and deviation characteristics of the first deviation link are encapsulated into a deviation tracing record, which is then written into the blockchain after being verified by distributed node consensus.

5. The method according to claim 4, characterized in that, Cross-stage correlation analysis of the feature summaries of the time-series coding matrix includes: Extract the temporal coding matrix feature summaries corresponding to the switching positions of adjacent links from the backtracking data chain, construct a feature summary pairing sequence, and convert the hash value of each pair of adjacent feature summaries in the feature summary pairing sequence into a coding vector; Calculate the cosine of the angle between the encoded vectors of adjacent feature summaries as the local correlation degree. Obtain the local correlation degree of all adjacent link switching positions. Assign attenuation weights according to the reverse position of the link switching position in the backtracking data chain. Sum the local correlation degree with the corresponding attenuation weights to obtain the global cross-link correlation degree. When the global cross-stage correlation is lower than the preset correlation threshold, the adjacent feature summary pairs with the lowest local correlation are selected, the corresponding stage switching positions are extracted, the correlation decay positions are determined, the difference between the preset correlation threshold and the global cross-stage correlation is calculated, and the correlation decay amount is determined.

6. A blockchain-based school uniform quality information traceability system, used to implement the method of any one of claims 1-5, characterized in that, include: The data acquisition unit is used to collect process data at each stage of the school uniform production process; The parameter extraction unit is used to extract process parameters from each step of the process data, calculate the sensitivity coefficient of the process parameters to the final quality indicators of the school uniform, screen process parameters whose sensitivity coefficient exceeds the preset sensitivity threshold to determine key quality factors, and record the numerical change trajectory of key quality factors in each step. The coding verification unit is used to generate a time-series coding matrix based on the numerical change trajectory of key quality factors, match and verify the numerical differences of multiple key quality factors with the collaborative conversion rules, and generate a stage conversion verification code. The identifier encapsulation unit is used to generate unique identifiers based on key quality factors and process transformation verification codes, encapsulate process data and unique identifiers into blocks, form a blockchain, and distribute them to distributed nodes. The state construction unit is used by each distributed node to construct a quality state transition graph based on the link transition verification code, identify the convergence node with multiple predecessor paths in the quality state transition graph, and perform backtracking verification on the multiple predecessor paths of the convergence node respectively. The backtracking verification unit is used to construct a backtracking data chain along the predecessor path when any predecessor path verification fails, locate the first deviation link and write the deviation tracing record into the blockchain. The traceability reporting unit is used to receive traceability requests containing unique identifiers, retrieve the numerical change trajectory and verification results of key quality factors from the blockchain, and generate a school uniform quality information traceability report.

7. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 5.

Images