Intelligent production line master control platform system based on blockchain and industrial big data

Abstract

The application relates to the technical field of intelligent manufacturing and industrial blockchain application, in particular to an intelligent production line general control platform system based on a blockchain and industrial big data, which comprises a data acquisition module, a causality analysis module, an intelligent contract execution module and a traceability management module; the data acquisition module acquires data through an industrial internet of things interface; the causality analysis module generates a dynamic defect risk index; the intelligent contract execution module compares the dynamic defect risk index with a preset risk threshold; if the dynamic defect risk index is greater than or equal to the preset risk threshold, process compensation instructions of a downstream node are generated, corresponding deviation compensation action information is updated and is stored in a chain; otherwise, a state maintenance mark is generated to control the downstream node to maintain a current parameter state, and the state maintenance mark is stored in a chain; the traceability management module generates a product full-life-cycle causality traceability tree; the application solves the problems of difficult cross-process causality tracing of a complex production line and unclear responsibility definition.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing and industrial blockchain application technology, specifically to an intelligent production line control platform system based on blockchain and industrial big data. Background Technology

[0002] A smart central control platform for continuous production lines is the core support for ensuring product quality and production operation management. Continuous production lines typically contain multiple nodes arranged in sequence according to the process flow, with products flowing sequentially along a pre-set process. To ensure the quality of the final product, it is usually necessary to acquire business data such as the timing status, flow logs and quality inspection of multiple nodes in the process for comprehensive monitoring. Product quality is affected by the status of multiple processes, and the status fluctuations of each node are inevitable in actual operation. Existing technologies for processing production line status data and controlling quality often rely solely on alarm thresholds for single nodes or final inspection results. In actual operation, parameter fluctuations at preceding nodes typically do not exceed tolerance ranges individually, preventing the system from triggering single-point alarms. However, these parameter fluctuations, though within preset tolerance ranges, gradually accumulate and couple during the flow of subsequent nodes, often manifesting as product failures only at the final inspection or usage stage. Existing technologies struggle to dynamically quantify the cumulative effects of cross-node deviations and make feedforward compensation decisions, leading to delayed identification of hidden defects and difficulties in tracing causal relationships throughout the product lifecycle, thus resulting in high operational and management risks. Therefore, a solution is urgently needed to address the problems present in existing technologies. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides an intelligent production line control platform system based on blockchain and industrial big data. Specifically, the technical solution of this invention includes: The data acquisition module is configured to acquire time-series status data and flow log data of each node in a multi-node production line process through an industrial IoT interface; The causal analysis module is configured to process time-series state data and flow log data based on a preset causal coupling model, calculate the transmission influence factor of the state fluctuation of the upstream node on the quality of the downstream node, and quantify the cumulative effect of cross-node deviation based on the transmission influence factor to generate a dynamic defect risk index. The smart contract execution module is configured to input the dynamic defect risk index into a smart contract deployed on the blockchain network via a blockchain remote call interface. The smart contract compares the dynamic defect risk index with a preset risk threshold. If the dynamic defect risk index is greater than or equal to the preset risk threshold, the smart contract generates a process compensation instruction for the downstream node and sends it to the corresponding downstream node through a programmable logic controller or device gateway to update the node parameter status. At the same time, the deviation compensation action information corresponding to the parameter status update is recorded on the blockchain. If the dynamic defect risk index is less than the preset risk threshold, a state maintenance flag is generated to control the downstream node to maintain the current parameter status, and the state maintenance flag is recorded on the blockchain. The traceability management module is configured to generate a causal traceability tree for the entire product lifecycle based on flow log data and deviation compensation action information or status maintenance markers stored on the blockchain.

[0004] Optionally, the causal coupling model includes Bayesian networks and physical mechanism knowledge graphs; the causal analysis module is specifically configured as follows when calculating the transmission influence factor of the state fluctuations of upstream nodes on the quality of downstream nodes: Extracting cross-node variable relationships based on physical mechanism knowledge graphs; Input the cross-node variable relationships and time-series state data into the Bayesian network to calculate the transmission influence factor.

[0005] Optionally, when the causal analysis module quantifies the cumulative effect of cross-node deviations based on transmission influencing factors and generates a dynamic defect risk index, the specific configuration is as follows: Calculate the difference between the timing state data of the current node and the preset baseline state data in each node, and use it as the parameter deviation value; The single-node risk value is obtained by multiplying the parameter deviation value with the transmission influence factor. Based on the node order determined by the flow log data, an exponential decay function with decreasing weights based on node order is used to aggregate the risk values ​​of a single node, generating a dynamic defect risk index.

[0006] Optionally, the smart contract execution module is specifically configured as follows when generating process compensation instructions for downstream nodes through smart contracts: Extract the target downstream node corresponding to the dynamic defect risk index; The system uses a built-in database to match the range of available adjustable parameters corresponding to the target downstream node from a preset process knowledge base. With minimizing the dynamic defect risk index as the objective function, the optimal process parameters are calculated by iteratively solving the problem within the range of available adjustable parameters using a preset optimization algorithm. The optimal process parameters are encapsulated as process compensation instructions.

[0007] Optionally, the system also includes a privacy computing module, which is configured as follows: Before storing deviation compensation action information or state maintenance markers on the blockchain, the parameters or state maintenance markers in the process compensation instructions are encrypted using a zero-knowledge proof algorithm to generate quality compliance proof data. The quality compliance certificate data is broadcast to the blockchain network as part of the deviation compensation action information or state maintenance mark for distributed consensus and storage.

[0008] Optionally, the traceability management module can be configured as follows when generating a causal traceability tree for the entire product lifecycle: Obtain the product identifier of the target product; Based on the product identifier, retrieve the corresponding circulation log data, deviation compensation action information, or state maintenance marker in the blockchain network; and based on the circulation log data and the cross-node variable correlation obtained by the causal analysis module, parse out the node state of the corresponding node and the deviation evolution path across nodes. By linking nodes in the flow log data according to the time series, as well as deviation compensation action information or state retention markers, a causal traceability tree for the entire product lifecycle is constructed, which includes node status, deviation evolution path, and deviation compensation action information or state retention markers.

[0009] Optionally, the system also includes a resilience assessment module, which is configured as follows: The non-negative triggering ratio of smart contracts triggering process compensation instructions and the corresponding compensation success rate are statistically analyzed within a preset time period; where the triggering ratio is the proportion of the number of product units that trigger process compensation instructions within the preset time period to the total number of product units within that period. If the trigger ratio is greater than zero, the ratio of the compensation success rate to the trigger ratio will be used as the overall anti-interference resilience index of the production line; if the trigger ratio is equal to zero, the overall anti-interference resilience index of the production line will be set to the preset full score benchmark value; the preset full score benchmark value is used to characterize that the production line has not been subjected to interference exceeding the preset tolerance within the preset time period. The overall anti-interference resilience index of the production line is packaged into a resilience assessment data package and output.

[0010] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention quantifies the transmission influence factors of upstream state fluctuations on downstream through a causal analysis module, generates a dynamic defect risk index, and issues process compensation instructions by smart contract before the risk exceeds the limit. This mechanism breaks the limitation of single-node alarm, effectively blocks the cumulative effect of fluctuations within the preset tolerance range, transforms post-final inspection rejection into feedforward dynamic compensation during production line operation, and reduces the failure risk caused by hidden defects. 2. This invention records deviation compensation actions and status maintenance tags on the blockchain for evidence storage, and combines them with circulation logs to construct a product lifecycle causal traceability tree in time sequence; this mechanism not only ensures the tamper-proof protection of key control data, but also enables rapid backtracking to minor anomalies and compensation processes at any intermediate node in the event of quality disputes, solving the problems of difficult causal traceability across processes and unclear responsibility definition in complex production lines; 3. This invention employs a model that integrates physical mechanism knowledge graphs and Bayesian networks to calculate transmission influence factors. By extracting variable correlations that conform to process rules through mechanism graphs, and then using Bayesian networks for probabilistic inference, it avoids the defect of pure data-driven models that easily misjudge accidental correlations as causal paths, realizes interpretable calculation of transmission mechanisms, and significantly improves the stability and accuracy of quality risk assessment under complex working conditions. 4. When generating the dynamic defect risk index, this invention innovatively introduces a weighted exponential decay function to aggregate the risk values ​​of a single node according to the product flow sequence. The algorithm fully considers the difference between the self-absorption and immediate impact of deviations in the flow, and assigns higher intervention weights to anomalies that are closer to the current decision node, thereby transforming discrete cross-node fluctuations that do not exceed the single-point alarm threshold into a scientific and unified cross-process risk measurement benchmark. 5. When generating compensation instructions, the smart contract of this invention automatically performs optimization iteration within the range of available parameters matched in the process knowledge base, with the goal of minimizing the risk index. This mechanism solves the problem of relying on manual experience and having a long response cycle after traditional early warnings, and directly transforms abstract quality risks into optimal process parameters that can be automatically executed by field equipment, thus meeting the stringent requirements of continuous production lines for real-time dynamic optimization. 6. Before the compensation action is uploaded to the blockchain, the present invention introduces a zero-knowledge proof algorithm to generate quality compliance proof data and broadcasts it to the blockchain network as evidence payload. This mechanism allows the parties involved in collaborative production to prove that the product has been regulated according to the rules and meets the quality constraints without disclosing the specific process parameters in plaintext. This achieves on-chain trusted verification and secure privacy protection of core manufacturing formulas and production line optimization experience. Attached Figure Description

[0011] The present invention will be further explained below with reference to the accompanying drawings and embodiments: Figure 1 This is a schematic diagram of the modules of the intelligent production line control platform system based on blockchain and industrial big data provided in the embodiments of the present invention. Detailed Implementation

[0012] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0013] The intelligent production line control platform system based on blockchain and industrial big data includes: a data acquisition module configured to acquire time-series status data and flow log data of each node in the multi-node production line process through an industrial IoT interface; The causal analysis module is configured to process time-series state data and flow log data based on a preset causal coupling model, calculate the transmission influence factor of the state fluctuation of the upstream node on the quality of the downstream node, and quantify the cumulative effect of cross-node deviation based on the transmission influence factor to generate a dynamic defect risk index. The smart contract execution module is configured to input the dynamic defect risk index into a smart contract deployed on the blockchain network via a blockchain remote call interface. The smart contract compares the dynamic defect risk index with a preset risk threshold. If the dynamic defect risk index is greater than or equal to the preset risk threshold, the smart contract generates a process compensation instruction for the downstream node and sends it to the corresponding downstream node through a programmable logic controller or device gateway to update the node parameter status. At the same time, the deviation compensation action information corresponding to the parameter status update is recorded on the blockchain. If the dynamic defect risk index is less than the preset risk threshold, a state maintenance flag is generated to control the downstream node to maintain the current parameter status, and the state maintenance flag is recorded on the blockchain. The traceability management module is configured to generate a causal traceability tree for the entire product lifecycle based on flow log data and deviation compensation action information or status maintenance markers stored on the blockchain.

[0014] This embodiment provides a general control mechanism for a power battery module assembly production line. Specifically, a continuous production line is set up to include electrode drying node N1, winding node N2, liquid injection node N3, packaging node N4, and final inspection node N5 in sequence. The target product is a single module cell batch P202501. In actual operation, the following situation often occurs: the parameter fluctuations of the preceding nodes do not exceed the tolerance range on their own, but their impact accumulates gradually in subsequent nodes, and only manifests as abnormal capacity decay in the final inspection stage or customer use stage. To address this issue, this embodiment employs a closed-loop mechanism of data collection, causal analysis, on-chain decision-making, and source tracing to dynamically compensate for and reliably record cross-node deviations. N1 to N5 serve as node identifiers in the production line ordered by process sequence, and P202501 serves as the batch identifier for the production batch to be tracked. Specifically, the data acquisition module establishes communication connections with various device controllers, the MES system, and online monitoring equipment; the industrial IoT interface can adopt a unified OPC architecture. Data is acquired through transmission control protocols, process field networks, or equipment gateway forwarding methods. For each node, the collected time-series status data includes continuous variables such as temperature, pressure, speed, tension, and vacuum. The flow log data includes product identification, node entry time, exit time, equipment number, operator number, and batch number. The quality inspection data includes intermediate sampling results and final inspection results. For ease of explanation, the following simplified example data demonstrates the data flow. For a specific cell unit C17 in batch P202501, the data acquisition module obtains the following node data between timestamps t1 and t4: In N1, the drying temperatures are 81, 82, 84, and 83; in N2, the winding tensions are 10.0, 10.3, 10.5, and 10.2; in N3, the liquid injection volumes are 5.00, 5.02, 5.05, and 5.01; and in N4, the packaging pressures are 1.20, 1.18, 1.25, and 1.19. Correspondingly, the transfer log is recorded as C17:N1→N2→N3→N4→N5, and each transfer is accompanied by a timestamp; the final inspection node N5 gives the capacity test value, for example, 97.8%; where C17 serves as the identifier of a single cell unit in batch P202501, and t1 to t4 serve as the sampling time identifiers arranged in the order of sampling, and the four values ​​in the aforementioned nodes correspond to the sampling results at the four sampling times t1, t2, t3, and t4, respectively; After receiving the aforementioned time-series status data and flow log data, the causal analysis module associates different processes according to a predefined node sequence; the transmission influence factor here can be understood as the strength of the influence of a certain type of state fluctuation of an upstream node on the quality characteristics of a downstream node. For example, a simplified three-node relationship can be established: N1 temperature fluctuation affects N3 liquid absorption uniformity, N2 tension fluctuation affects N4 encapsulation stress concentration, and both affect N5 capacity stability; assuming that after model calculation, the conduction influence factor of N1 on N5 is 0.4, the conduction influence factor of N2 on N5 is 0.3, the conduction influence factor of N3 on N5 is 0.2, and the conduction influence factor of N4 on N5 is 0.1; Furthermore, the causal analysis module calculates the cumulative risk of deviation based on the difference between the current state of each node and the corresponding baseline state. In the example, the baseline states can be set as follows: drying temperature 83, winding tension 10.1, liquid injection volume 5.00, and packaging pressure 1.20. If representative values ​​of 84, 10.5, 5.05, and 1.25 are taken for a certain sampling moment, the deviation values ​​are 1, 0.4, 0.05, and 0.05, respectively. After calculating with the corresponding transmission influence factor, several single-node risk values ​​can be obtained, such as 0.4, 0.12, 0.01, and 0.005. Then, combined with the flow sequence, the dynamic defect risk index at that moment is formed. If the aggregation result is 0.62, and the system threshold is set to 0.55, it indicates that although the product unit has not triggered an alarm in a single process, there is already a clear tendency for cumulative failure across processes. To avoid confusion with the terminology used for intermediate aggregation quantities in the following text, the dynamic defect risk index used for direct comparison with risk thresholds in this embodiment refers to the standardized output value. If the unstandardized intermediate aggregation quantity is obtained first through the preceding calculation, it can be transformed by a preset mapping rule, such as S-shaped function mapping or minimum-maximum normalization processing, to form the dynamic defect risk index in the range of 0 to 1. Therefore, the above 0.62 should be understood as a standardized risk index that can be directly input into the smart contract for threshold comparison. The smart contract execution module receives the dynamic defect risk index through a blockchain remote call interface. The blockchain network can be deployed in a consortium blockchain environment jointly participated in by manufacturing enterprises, contract manufacturers, quality inspectors, and supply chain regulators. After receiving the product identifier C17, the target downstream node N4, the risk index 0.62, and the timestamp T, the contract executes the threshold comparison logic. If the index is greater than or equal to the threshold, the contract generates a process compensation instruction; if it is less than the threshold, a status preservation marker is generated. The timestamp T serves as the time identifier for this smart contract call event, used to record the sequence of risk assessment and instruction generation on the chain. In the compensation branch, process compensation instructions are issued to the corresponding nodes through programmable logic controllers or device gateways; for example, if the contract determines that the N4 packaging pressure and packaging holding time need to be fine-tuned, then control parameters such as increasing the packaging pressure from 1.20 to 1.23 and increasing the holding time from 2.0 seconds to 2.3 seconds are issued. After the device executes the action, the system records the product identifier, target node, parameters before compensation, parameters after compensation, execution time, and execution result, and stores this information on the blockchain as deviation compensation action information. In the hold branch, if the risk index of another cell unit C18 in the same batch is only 0.41, the system issues a hold status flag, controls the corresponding node to maintain the current parameters unchanged, and simultaneously stores the judgment result and time information of the non-compensation trigger on the blockchain. Here, C18 serves as the identifier of another cell unit in the same batch as C17. The traceability management module generates a causal traceability tree for the entire product lifecycle based on flow logs and on-chain records. This traceability tree can be organized in the manner of product identifier—process node—state evolution—compensation action—final inspection result. For cell unit C17, the root of the tree can be product identifier C17, the first layer is the flow link from N1 to N5, the second layer is attached to the time sequence state summary corresponding to each node, the third layer is attached to the causal transmission relationship and risk evolution path, and the fourth layer is attached to the compensation action information or state maintenance mark. In this way, once a capacity anomaly occurs later, it is possible to trace back to the upstream micro-fluctuation and its intermediate compensation process by going up the tree. Furthermore, the following measures can be taken for abnormal situations: when the data acquisition module experiences a sampling interruption, the interruption window can be marked as missing, and the most recent valid value can be used to maintain or interpolate the data within the window before proceeding with the analysis; if the continuous missing duration exceeds a preset threshold, such as more than 30 seconds, the product unit will be marked as having incomplete data, automatic compensation will be suspended, and manual confirmation will be required before the data can continue to be processed. If a remote blockchain call fails, the record to be uploaded to the chain is first written to the local anti-tampering cache queue. After the network is restored, the on-chain timestamp is rewritten, and the delayed on-chain status is marked in the traceability tree. If the programmable logic controller does not send back execution confirmation, the compensation action is not considered to have taken effect. The system can reissue it once. After the number of retries exceeds the preset number, it will be handled manually. For example, in the power battery module assembly scenario, during the night shift production of batch P202501, the temperature of the N1 drying oven slightly increased. Although observed individually, it was still within the allowable range of the equipment. However, after N2 winding and N3 liquid injection, the system predicted that the final inspection capacity consistency would decrease. Instead of waiting for the final inspection results, the central control platform issued a compensation instruction in advance at the N4 packaging node to fine-tune the pressure holding parameters of some products and synchronized this feedforward compensation process to the blockchain. When the sample was inspected the next day, the capacity deviation of this batch was controlled within the qualified range, and the source of deviation, compensation process, and responsibility of each cell could be traced. The purpose of this step is to transform the local fluctuations that were originally scattered in various processes into a calculable, interveneable, and traceable end-to-end quality risk closed loop, so as to achieve feedforward suppression of cross-node deviations and the reliable construction of the product life cycle quality causal chain.

[0015] Furthermore, the causal coupling model includes a Bayesian network and a physical mechanism knowledge graph. When calculating the transmission influence factor of the state fluctuation of the upstream node on the quality of the downstream node in each node, the causal analysis module is specifically configured as follows: extracting cross-node variable correlations based on the physical mechanism knowledge graph; inputting the cross-node variable correlations and time-series state data into the Bayesian network to calculate the transmission influence factor.

[0016] This embodiment provides a causal coupling mechanism jointly driven by mechanism and data. Specifically, in the aforementioned power battery module assembly line, relying solely on historical statistical correlation is prone to the following defects: when the production line switches material batches, the environmental humidity changes, or the equipment maintenance status changes, the pure statistical model may misjudge accidental co-occurrence as a real causal relationship, resulting in distortion of the direction and magnitude of the transmission influencing factors. Therefore, this embodiment introduces a physical mechanism knowledge graph and a Bayesian network for collaborative solution to improve the stability of cross-process causal inference. Specifically, the physical mechanism knowledge graph is used to store the constraint relationships between process variables that conform to the process rules. In this graph, each node can correspond to a process variable or quality feature, and each edge represents the direction of influence between variables. For example, the following simplified associations can be established: drying temperature → residual moisture in electrode sheet; residual moisture in electrode sheet → uniformity of liquid injection wetting; winding tension → interlayer adhesion of electrode sheet; adhesion + uniformity of wetting → internal resistance stability after packaging; internal resistance stability → capacity consistency. In this way, the system will not directly connect all sensor variables indiscriminately, but will first screen out reasonable cross-node paths according to the process rules. For ease of explanation, a simplified set of relationships can be constructed; let variable V1 represent N1 drying temperature, V2 represent residual moisture in the electrode, V3 represent N2 winding tension, V4 represent N3 liquid injection uniformity, V5 represent N4 encapsulation stress, and V6 represent N5 final test capacity; the effective correlation paths extracted from the graph include V1→V2→V4→V6 and V3→V5→V6; if a certain path does not conform to known physical laws, for example, if the N1 drying temperature directly determines the N4 encapsulation pressure setting value, then it will not be included in this round of causal calculation graph; Based on this, the Bayesian network establishes a directed probability structure according to the above-mentioned variable relationships; for example, the continuous variable can be discretized into three states: low, normal, and high; taking the drying temperature of V1 as an example, below 82 is recorded as low, 82 to 84 is recorded as normal, and above 84 is recorded as high, and then the conditional probability is calculated based on historical production data. It should be noted that the drying temperature → residual moisture of the electrode in the graph is used to characterize the process constraint relationship, and is not limited to a simple monotonic linear increase or decrease. In the processing of power battery electrodes, a lower temperature is more likely to lead to a higher residual moisture, while a higher temperature may continue to affect the subsequent wetting uniformity and final capacity through paths such as over-drying of the electrode and changes in pore state. Therefore, in this set of discretization examples in this embodiment, the statistical results of 0.65 for V1 being low and V2 being high residual moisture, 0.70 for V2 being high and V4 being low wetting uniformity, and 0.60 for V4 being low and V6 having low capacity can be used to demonstrate a type of conduction strength calculation process from V1 to V6. In another set of samples that meet the spectrum constraints, a high V1 can also negatively affect V6 through other instability paths allowed by the spectrum. The two correspond to conduction branches under different working conditions and do not conflict. Furthermore, when cell unit C17 in batch P202501 detects that V1 is high and V3 is high at a certain moment, the system inputs this state into the Bayesian network; if the inference results show that: when only considering the V1 path, the posterior probability of V6 being low increases from 0.10 to 0.27; when only considering the V3 path, the posterior probability of V6 being low increases from 0.10 to 0.19; When considering both V1 and V3 paths, the posterior probability of a lower V6 increases to 0.36; therefore, the contributions of V1 and V3 paths to the final inspection quality can be converted into transmission influence factors, such as 0.17 and 0.09, respectively, or further normalized to form a relative contribution ratio of 0.65 and 0.35. Furthermore, for abnormal or incomplete modeling situations, the following can be done: If a variable in the knowledge graph lacks a corresponding physical relationship, for example, a new wind speed sensor for the drying chamber has been added but the process modeling has not yet been completed, then the variable can be used as a candidate auxiliary variable, only participating in statistical verification, and not directly entering the main causal path; If a conditional probability in the Bayesian network is unstable due to too few samples, for example, a piece of equipment has just been put into operation and there are less than 50 historical data, then a combination of an expert initial probability table and online incremental correction can be used to handle the situation. If the number of times the path given by the graph contradicts the statistical results within a preset time period reaches a preset threshold, for example, if the graph believes that V1 should positively influence V2, but recent data shows that it does so in the opposite direction, the system can mark the conflicting path as a relationship to be verified, and instead of deleting it immediately, it will trigger the process engineer to review the equipment calibration status and sensor orientation. For example, during the aforementioned night shift production, the system detected that the drying temperature corresponding to C17 was too high. If the analysis was based solely on time correlation, it might be mistakenly determined that it was directly related to subsequent packaging pressure fluctuations. However, after screening through mechanism graphs, the system only retained the reasonable link of drying temperature—residual moisture—liquid injection uniformity—capacity. After combining this with Bayesian network calculations, it was confirmed that the risk mainly came from the combined effect of upstream drying and winding, rather than the packaging equipment itself being out of balance. In this way, subsequent compensation became more targeted. The purpose of this mechanism is to couple the prior physical laws of the process with the inference results from historical data in order to achieve interpretable calculation of cross-node transmission influence factors and avoid misidentifying accidental correlations as real causal paths.

[0017] Furthermore, when the causal analysis module quantifies the cumulative effect of cross-node deviation based on the transmission influence factor to generate a dynamic defect risk index, the specific configuration is as follows: calculate the difference between the time-series state data of the current node and the preset baseline state data in each node as the parameter deviation value; multiply the parameter deviation value with the transmission influence factor to obtain the single-node risk value; and aggregate the single-node risk value according to the node order determined by the flow log data, using an exponential decay function based on the decreasing weight allocation of the node order to generate a dynamic defect risk index.

[0018] This embodiment provides a quantitative generation mechanism for a dynamic defect risk index. Specifically, after obtaining the transmission influence factor through the aforementioned mechanism and data, if the decision is still based solely on whether a single-point alarm threshold is exceeded, the following defects will be exposed: when multiple nodes are in a state of slight deviation but have not yet exceeded the limit, traditional systems often do not issue alarms, which is precisely the situation where cumulative quality failures are most easily overlooked. Therefore, this embodiment forms a cross-process dynamic defect risk index by aggregating deviation values, transmission influence factors, and weighted exponential decay. Specifically, the system pre-sets baseline state data for each node; this baseline state can come from the statistical average within a stable production window or from the target value of the process formula; taking the aforementioned production line as an example, the following are set: N1 drying temperature baseline is 83, N2 winding tension baseline is 10.1, N3 liquid injection baseline is 5.00, and N4 packaging pressure baseline is 1.20; when a certain cell unit C17 flows through four nodes, the representative sampled values ​​are 84, 10.5, 5.05, and 1.25, respectively, and the parameter deviation values ​​are 1, 0.4, 0.05, and 0.05, respectively; Furthermore, the single-node risk value is obtained by multiplying each deviation value with the corresponding transmission influence factor. Assuming the transmission influence factors are 0.4, 0.3, 0.2, and 0.1, the single-node risk values ​​are: N1: 1 × 0.4 = 0.40; N2: 0.4 × 0.3 = 0.12; N3: 0.05 × 0.2 = 0.01; N4: 0.05 × 0.1 = 0.005. For the risk values ​​of the above single nodes, the system does not simply add them together, but instead introduces a weighted exponential decay function to aggregate them according to the product flow sequence. The reason for this setting is that the closer the deviation is to the current decision node, the higher the reference value for immediate compensation. Although the deviation of the earlier process has already occurred, its impact may have been partially absorbed in the intermediate stage. For ease of explanation, a simplified set of weights can be used: N1 corresponds to 0.5, N2 to 0.7, N3 to 0.85, and N4 to 1.0; then the dynamic defect risk index can be formed by weighted summation, for example, 0.40×0.5+0.12×0.7+0.01×0.85+0.005×1.0=0.2975; it should be further clarified here that the aforementioned 0.5, 0.7, 0.85, and 1.0 are not arbitrarily specified, but rather are the discrete results of the weighted exponential decay function at the four node positions; in one implementation method, the original decay weight corresponding to the node with node sequence number i can be calculated first according to the following formula. : in, The node sequence number is The original decay weights corresponding to the nodes; Indicates the sequential numbering of nodes; This indicates the total number of nodes in the current decision-making chain; This represents the dimensionless preset attenuation coefficient; is the base of the natural logarithm; The system then normalizes or rounds the obtained weights according to process requirements; taking the aforementioned four-node scenario as an example, when When choosing the appropriate value, a decay effect of the same magnitude as 0.5, 0.7, 0.85, and 1.0 can be obtained. Therefore, the aforementioned values ​​are used to intuitively demonstrate the aggregation logic that the closer to the current compensation node, the greater the weight. If combined with correction terms such as product category coefficient and detection reliability coefficient, a risk index between 0 and 1 can be finally obtained, such as 0.62. In other words, 0.2975 can be understood as the intermediate risk after exponential decay aggregation, while 0.62 corresponds to the standardized result after substituting the intermediate risk into the preset mapping rule, thus maintaining consistency with the threshold comparison caliber in the previous example. In another micro-level simulation, if the parameter values ​​of a subsequent product unit C18 are 83.2, 10.2, 5.01, and 1.21, then the deviation values ​​are only 0.2, 0.1, 0.01, and 0.01; the corresponding single-node risk values ​​are 0.08, 0.03, 0.002, and 0.001; after aggregation, a lower risk index can be obtained, such as 0.18; in this way, even if no single node triggers the device limit violation, the system can still distinguish between product units with minor but high cumulative risk and product units with minor and negligible risk. Furthermore, boundary cases can be handled as follows: If a baseline state data changes due to a formula switch, such as switching from a winter formula to a summer formula, the system should simultaneously replace the baseline value of that node when the new formula takes effect to avoid calculating false deviations using the old baseline; if a node deviation value is negative, such as when the actual temperature is lower than the baseline, the system can choose to retain the sign for calculation according to the process definition, or take the absolute value first and then combine it with the risk direction for correction; for variables that clearly only bring risk if they are too high, such as leakage caused by excessive injection volume, the case of being too low can be recorded as 0 risk contribution; if a node in exponential decay polymerization is missing data, it can be handled by redistributing the weights of adjacent nodes, such as proportionally transferring the weights of the missing node to the preceding and following nodes, or directly marking the product unit as having insufficient confidence in risk calculation; For example, during the night shift operation of batch P202501, C17 did not exceed the preset alarm threshold at any of the N1, N2, N3, or N4 nodes. However, the high upstream temperature and high winding tension continued to overlap, resulting in a risk index of 0.62 after aggregation. In contrast, although C18 also experienced slight fluctuations, its index was only 0.18 because the absolute value of the deviation was small and its impact was limited due to its proximity to the final process. Based on this, the central control platform only triggered compensation for C17, while maintaining the status quo for C18, thereby avoiding the use of a uniform range of parameter adjustments for all products. The purpose of this step is to transform discrete, weak-amplitude, cross-node process fluctuations into a unified risk metric, so as to enable early identification and graded handling of the accumulation process of hidden defects.

[0019] Furthermore, when the smart contract execution module generates process compensation instructions for downstream nodes through smart contracts, the specific configuration is as follows: extract the target downstream node corresponding to the dynamic defect risk index; match the available adjustable parameter range corresponding to the target downstream node from the preset process knowledge base through the built-in system database; calculate the optimal process parameters by iteratively solving within the available adjustable parameter range using a preset optimization algorithm, such as grid search algorithm, genetic algorithm, or simulated annealing algorithm, with minimizing the dynamic defect risk index as the objective function; and encapsulate the optimal process parameters into a process compensation instruction.

[0020] This embodiment provides a process compensation solution mechanism for downstream nodes. Specifically, after the aforementioned risk index can identify hidden risks, if the system only gives an alarm indicating the existence of risks without outputting executable parameters, there is still a significant bottleneck: human engineers need to make a comprehensive judgment within a preset time threshold on which node to adjust, how much to adjust, and whether it will cause new side effects, which is difficult to meet the real-time requirements of high-speed production lines. Therefore, this embodiment further combines the process knowledge base and optimization algorithm to automatically generate feasible compensation instructions after on-chain triggering. Specifically, the smart contract extracts the target downstream node based on the transmission path corresponding to the risk index. For example, for C17, the system determines through causal path that the main risk will be amplified in the N4 encapsulation stage, so the target node is N4, rather than the already past N1 or N2. If a risk path points to multiple candidate nodes at the same time, the nearest downstream node that has not yet been executed and has room for compensation can be selected first. If the nearest downstream node does not support automatic parameter tuning, the second nearest node is selected. The available adjustable parameter range corresponding to the target node is read from the process knowledge base. The process knowledge base can be preset by process engineers or incrementally updated from historical successful execution cases. Taking the N4 package node as an example, the available adjustable parameters can include packaging pressure, holding time, fixture temperature, etc. In the example, the allowable range of packaging pressure is 1.18 to 1.28, with a step of 0.01; the allowable range of holding time is 1.8 to 2.6 seconds, with a step of 0.1 seconds; and the allowable range of fixture temperature is 35 to 40, with a step of 1. The knowledge base can also record taboo rules, such as high temperature and high pressure combinations cannot reach the upper limit at the same time; if the current product model is a high energy density model, the fixture temperature must not exceed 38. It should be noted that the range of available adjustable parameters does not only refer to the upper and lower limits of each parameter's value, but also includes the set of feasible parameters jointly defined by taboo rules, equipment load limits, energy consumption constraints, yield side effect constraints, and product model restrictions; the subsequent optimization algorithm iterates within this set of feasible parameters, rather than unconditionally searching for all combinations of values. Furthermore, the smart contract solves for the problem by minimizing the dynamic defect risk index. For ease of explanation, a simplified iterative approach can be used. Assume the current parameters are a packaging pressure of 1.20, a holding time of 2.0 seconds, and a fixture temperature of 36°C, corresponding to a risk index of 0.62. The system attempts several candidate combinations within the allowable range: Combination A is 1.22, 2.2 seconds, and 36 seconds, corresponding to a predicted risk of 0.50; Combination B is 1.23, 2.3 seconds, and 36 seconds, corresponding to a predicted risk of 0.41; Combination C is 1.24, 2.3 seconds, and 37 seconds, corresponding to a predicted risk of 0.39; Combination D is 1.26, 2.5 seconds, and 38 seconds, corresponding to a predicted risk of 0.38, but triggers energy consumption constraints or yield side effect limitations. To maintain consistency with the objective function of this embodiment, the aforementioned equipment load, energy consumption, and parameter stability are preferably treated as feasibility constraints or secondary selection conditions in this embodiment: that is, candidate combinations that do not meet the constraints of the process knowledge base are first eliminated. Therefore, although combination D in the example has a lower predicted risk, it is not included in the final comparison because it exceeds the set of feasible parameters. Among the remaining feasible combinations A, B, and C, the one with the lowest dynamic defect risk index is selected as the preferred result. Thus, combination C can be determined as the optimal process parameter. If the risk outcomes of multiple feasible combinations are the same or the differences are less than the preset tolerance, such as not exceeding 0.01, then the combination with smaller parameter changes and lower equipment disturbances is selected as the final output. After selection, it is packaged into a process compensation instruction. For example, when product unit C17 reaches N4, the packaging pressure is set to 1.24, the holding pressure is set to 2.3 seconds, and the fixture temperature is set to 37. The encapsulation result is forwarded to the field device via the device gateway or PLC protocol, and the device is required to send back execution confirmation. The instruction message may include product identifier, node number, parameter item, parameter value, effective time, valid batch quantity, and signature digest to ensure that the field execution object and timing are accurate. As an anomaly handling mechanism, if there are no available adjustment parameters for the target node, such as the process having only a fixed cycle time and being unable to be fine-tuned, the system can generate manual review suggestions instead of automatic compensation; if all candidate combinations fail to reduce the risk index below the threshold during the optimization process, the parameter combination with the greatest risk reduction can be selected as the mitigation solution, and subsequent stricter inspection strategies can be added. If the risk results of multiple candidate combinations are close, for example, the difference is less than the preset tolerance of 0.01, the scheme with smaller parameter change can be selected first to reduce the disturbance to equipment stability; if the field equipment returns a parameter writing failure or the equipment is busy, the compensation instruction will automatically become invalid, the system will switch to state hold and prompt manual intervention. For example, when production of batch P202501 continued, C17 was analyzed and confirmed to require compensation at N4. The system retrieved from the knowledge base that the packaging process was suitable for absorbing upstream deviations through a combination of pressure and holding pressure. After rapid iteration, it selected a parameter set of 1.23 for packaging pressure and 2.3 seconds for holding pressure, and issued an instruction 0.5 seconds before C17 entered the packaging station. After the equipment executed the instruction, the subsequent final inspection capacity was restored to 98.4%, and no deviations were found. The purpose of this step is to transform abstract risk assessments into specific process parameters that can be directly executed by field equipment, so as to achieve feedforward compensation for cumulative deviations across processes, rather than remaining at the alarm level.

[0021] Furthermore, the system also includes a privacy computing module, which is configured to: before uploading deviation compensation action information or state maintenance markers to the blockchain for notarization, encrypt the parameters or state maintenance markers in the process compensation instruction based on a zero-knowledge proof algorithm, such as concise non-interactive zero-knowledge proof, scalable transparent zero-knowledge proof, or bulletproof proof protocol, to generate quality compliance proof data; and broadcast the quality compliance proof data as part of the deviation compensation action information or state maintenance markers to the blockchain network for distributed consensus and notarization.

[0022] This embodiment provides a privacy protection mechanism before on-chain evidence storage. Specifically, in the aforementioned scheme, if the deviation compensation action information is directly written into the consortium blockchain in plaintext, it will bring new problems: In cross-enterprise collaborative production scenarios, when contract manufacturers, brand owners, quality inspectors, and supply chain finance parties jointly participate in on-chain verification, if core process parameters such as packaging pressure, drying temperature, and liquid injection volume are fully disclosed, it is easy to leak manufacturing formulas and production line optimization experience. Therefore, this embodiment introduces zero-knowledge proof to generate quality compliance proof data before on-chain storage, proving that the batch of products has completed compensation according to the rules and meets quality constraints without disclosing the specific parameter values. Specifically, taking the C17 triggering compensation at the N4 package node as an example, the actual execution parameters on site may be a package pressure of 1.23, a holding pressure of 2.3 seconds, and a fixture temperature of 37°C. The privacy computing module does not directly broadcast the above plaintext values, but instead inputs the parameters along with preset quality rules into the proof circuit. The quality rules can be expressed as: the parameters must be within the process allowable range; the predicted risk index after compensation is less than the target upper limit; and the predicted results of the final inspection or intermediate inspection meet the qualification standards. In the simplified deduction, it can be understood as three judgment conditions: First, 1.18≤sealing pressure≤1.28; Second, 1.8≤holding time≤2.6; Third, the risk index after compensation<0.55; The privacy computing module generates the proof result Z1 based on these conditions. The on-chain verification node only needs to verify the validity of Z1 to confirm that there is a set of undisclosed parameters that meet the above conditions. Similar processing can be used for state maintenance markers; for example, if C18 does not trigger compensation, the system needs to prove that the risk index is already below the threshold in the current state, and maintaining the status quo will not cause deviation; at this time, proof result Z2 can be generated, the proof content corresponds to the current parameter being within the allowable range and the risk index <0.55, but does not disclose the specific process parameters of the product; Quality compliance proof data can be combined with deviation compensation action information or state maintenance markers to form an on-chain evidence payload; in the example, deviation compensation action information may include product identifier hash value, target node number, action type, execution time, proof result Z1, and proof verification public key index; state maintenance evidence may include product identifier hash, node number, maintenance type, judgment time, and proof result Z2; each verification node in the consortium blockchain can complete consensus confirmation without knowing the specific formula; As an exception handling mechanism, if the generation time of zero-knowledge proof exceeds the allowed window on site, for example, the cycle time of the target workstation is only 1 second while the proof calculation takes longer, the system can adopt a two-stage on-chain method: first issue on-site control instructions and cache the original data in the local security area, and then delay the generation of the proof result asynchronously and supplement it on the chain. If the proof generation fails, the system may choose to upload only the minimum necessary digest information on the chain and mark the batch as missing privacy proof to be supplemented; if the proof verification fails, the on-chain record will not enter the final confirmation state and the audit process will be triggered at the same time; if there are old version nodes among the participating nodes that do not support proof verification, the gateway node can pre-complete the verification and broadcast the verification digest, and the old nodes will only receive the result and will not participate in the deep verification. For example, in the scenario where batch P202501 was completed collaboratively by the main manufacturer and the outsourced packaging plant, the outsourced plant actually executed the compensation parameters, but did not want other participants besides the main manufacturer to directly know its equipment tuning range; the system broadcasts the fact that the compensation action that meets the qualification requirements has been executed to the consortium blockchain in the form of quality compliance certificate data, so that the brand can confirm that the product compensation is compliant, the quality inspector can verify that the quality conditions are met, and the parameter details of the outsourced plant will not be exposed in plain text on the chain; The purpose of this mechanism is to protect core technical information such as process parameters while maintaining on-chain credible evidence and multi-party verifiability, so as to achieve a balance between quality reliability and formula confidentiality.

[0023] Furthermore, when generating a product lifecycle causal traceability tree, the traceability management module is specifically configured as follows: obtain the product identifier of the target product; retrieve the corresponding circulation log data, deviation compensation action information, or state maintenance markers in the blockchain network based on the product identifier; and based on the circulation log data and the cross-node variable association relationships obtained by the causal analysis module, parse out the node status of the corresponding node and the deviation evolution path across nodes; link the circulation log data and deviation compensation action information or state maintenance markers according to the time series to construct a product lifecycle causal traceability tree containing node status, deviation evolution path, and deviation compensation action information or state maintenance markers.

[0024] This embodiment provides a mechanism for constructing a causal traceability tree for the entire lifecycle of a single product. Specifically, after compensation or record keeping has been completed on the aforementioned chain, if the system can only provide isolated log entries, the following defects will still occur: when customer complaints or quality disputes occur, engineers need to manually review records of multiple processes, nodes, and batches, making it difficult to quickly answer the core question of which minor fluctuation evolved into the final defect through which path. Therefore, this embodiment aggregates on-chain and off-chain information through product identification to construct a time-ordered and traceable traceability tree. Specifically, the traceability management module obtains the product identifier of the target product; this identifier can be a single cell serial number, module code, or a unique identifier bound to RFID; in this example, cell unit C17 is taken as the target object; the system uses C17 as the search key to find the corresponding circulation log data and deviation compensation action information or status maintenance mark in the blockchain network; For ease of explanation, the search results can be set as follows: Time T1, C17 enters N1, drying temperature 84; Time T2, C17 leaves N1 and enters N2, winding tension 10.5; Time T3, C17 enters N3, liquid injection volume 5.05; Time T4, the system calculates risk index 0.62; Time T5, the system generates compensation instruction, target node N4; Time T6, N4 executes packaging pressure 1.23, holding pressure for 2.3 seconds, corresponding to the on-chain evidence as compensation action summary and proof result; Time T7, C17 enters N5 for final inspection, capacity 98.4, judged as qualified; For another product C18, it may not have a compensation action at T5, but instead generates a status maintenance mark and uploads it to the chain; Among them, T1 to T7 are event timestamps arranged in chronological order, corresponding to key event times such as flow entry, risk calculation, instruction generation, on-site execution, and final inspection completion; During tree construction, the system links nodes according to time sequence; the root is the product identifier C17; the first layer can be nodes N1 to N5 of each process; the second layer can be attached to the status summary of each process, such as temperature, tension, liquid injection volume, packaging parameters, and final inspection indicators; the third layer establishes the deviation evolution path, such as high drying time → change in residual moisture → risk of decreased liquid injection uniformity → risk of final inspection capacity fluctuation. The fourth layer connects compensation action information or status retention markers, including action type, execution time, evidence summary and verification status; at the visualization level, this structure can also be represented as a directional tree link; if a node has no compensation, the records on its branches are retained; if there are multiple compensations, multiple compensation child nodes can be connected in series. Furthermore, if subsequent customer feedback on failures occurs during the usage phase, such as an abnormal capacity decay in a module after six months, the system can trace the cause-effect tree of its internal unit C17 through the product identifier, thereby quickly seeing that it had been compensated in N4 and that the upstream anomaly originated from the joint fluctuation of N1 and N2; in this way, not only can the technical cause be located, but the boundary of responsibility can also be clarified, such as determining whether it is upstream equipment drift, downstream improper execution or material batch abnormality; As an exception handling mechanism, if the retrieved on-chain record is incomplete, such as the delay in uploading the process log to the chain, the system can first build a temporary branch based on the off-chain cache record, and then replace it with the official record after the subsequent on-chain completion; if there are conflicting records for the same product identifier, such as two different devices writing the same state at the same time, the system can select the main record based on the timestamp signature, device trust level, or consensus confirmation height, while retaining the conflicting branch for auditing; if the product is split or merged midway, such as multiple cells being assembled into a module, the traceability tree can add a combination node at the upper layer to record the assembly relationship between parent and child product identifiers; For example, after a customer complains about a module, the after-sales personnel enter the module code. The system first locates the relevant cell C17, and then expands its entire life cycle cause-effect traceability tree. The interface can directly display: N1 drying temperature is too high, N2 winding tension is too high, T4 risk index reaches 0.62, T5 triggers compensation, T6 performs packaging fine-tuning, and T7 final inspection is qualified. If other cells in the same batch also share similar paths, they can be further clustered according to similar branches of the tree to determine whether it is a batch equipment problem. The purpose of this step is to restore records scattered across different stages, times, and entities into a single-product readable causal evolution structure, so as to achieve second-level location of quality problems, clear division of responsibilities, and credible auditing of compensation history.

[0025] Furthermore, the system also includes a resilience assessment module, which is configured to: statistically analyze the non-negative triggering ratio of smart contract-triggered process compensation instructions and the corresponding compensation success rate within a preset time period; wherein the triggering ratio is the proportion of the number of product units triggering process compensation instructions within the preset time period to the total number of product units within that period; if the triggering ratio is greater than zero, the ratio of the compensation success rate to the triggering ratio is used as the overall anti-interference resilience index of the production line; if the triggering ratio is equal to zero, the overall anti-interference resilience index of the production line is set to a preset full-score benchmark value; the preset full-score benchmark value is used to characterize that the production line has not been subjected to interference exceeding the preset tolerance within the preset time period; and the overall anti-interference resilience index of the production line is encapsulated into a resilience assessment data package and output.

[0026] This embodiment provides a mechanism for assessing the overall anti-interference resilience of a production line. Specifically, after the aforementioned system has the ability to compensate and trace individual products, if a higher-level production line resilience assessment is lacking, there will still be a blind spot for management: it can only see whether a single compensation is successful, but it is difficult to determine whether the entire production line's ability to absorb disturbances is enhanced or weakened over a period of time. Therefore, this embodiment forms an overall anti-interference resilience index of the production line based on the triggering ratio of smart contracts and the compensation success rate. Specifically, the resilience assessment module collects two types of data within a preset time period, such as by shift, by day, or by week: one is the triggering rate of smart contract-triggered process compensation instructions, and the other is the corresponding compensation success rate. The triggering rate can be understood as the proportion of product units that are determined to require compensation within a unit period to the total number of product units in that period. The compensation success rate can be understood as the proportion of compensated products that pass subsequent inspections or whose risk index is effectively reduced to below the target threshold. To ensure that the ratio of compensation success rate to trigger ratio has a stable and comparable dimension, this embodiment preferably normalizes the trigger ratio according to the total number of products, total number of work orders, or standardized output cycle within the statistical period before participating in the calculation; when the total number of products is used as the normalization base, the trigger ratio represents the proportion of the number of triggers to the total number of product units in that period. For ease of explanation, let's take the night shift of the aforementioned power battery module assembly line as an example. Assuming that 1,000 battery cells are produced in a shift, and compensation commands are triggered 50 times, the trigger ratio after normalization based on the total number of product cells can be recorded as 0.05. If 45 of these 50 compensations result in risk relief or indicator recovery in the subsequent final inspection, the compensation success rate can be recorded as 45 / 50, or 0.9. According to the definition in this embodiment, the ratio of 0.9 to 0.05 can be used as the base value of the resilience index, for example, 18. For easier management and display, it can also be further mapped to a scoring range of 0 to 100. For example, after combining the historical best value, product model baseline, or preset normalization coefficient, a score of 72 can be obtained. Therefore, although the trigger ratio can be described as 50 times in the text description, when participating in the actual calculation of the resilience index, the normalized frequency value of 0.05 corresponding to the statistical caliber is uniformly adopted to avoid mixing the number value and the ratio value. In another time period, if the production line runs smoothly without any compensation triggers, the toughness evaluation cannot be simply recorded as 0. This is because a trigger ratio of zero may mean that the production line itself has very little interference, the process window is stable, and the yield can be maintained without additional compensation. Therefore, when the trigger ratio is equal to zero, this embodiment directly sets the toughness index to a preset full score benchmark value, such as 100 points or 1.0, to reflect that the system is in the best stable state in this period. Furthermore, the resilience assessment data package may include a statistical period identifier, product type, number of triggers, number of successful compensations, compensation success rate, calculated resilience index, and a summary of the main sources of disturbance during the same period. For example, the data package for a day shift may record: Period 2025-01-18 day shift, product model A12, 12 compensation triggers, 11 successful, success rate 91.7%, resilience index 76, and the main sources of disturbance are drying oven temperature drift and slight fluctuations in winding tension. This data package can be output to the manufacturing execution dashboard, factory scheduling platform, or operation and maintenance analysis system. As an anomaly handling mechanism, if the trigger ratio is greater than zero within a certain period, but the compensated products have not yet all completed final inspection, resulting in the compensation success rate being temporarily unquantifiable, the system can first output a phased resilience index and append a field of numbers for samples awaiting final inspection to the results. If there are multiple criteria for determining whether compensation is successful, such as considering both the decline in the risk index and the final inspection results, then multiple conditions can be used to define success, or a comprehensive success rate can be calculated based on weights. If the statistical period is too short, resulting in excessive randomness in the trigger ratio, such as using a 5-minute window, which may cause significant fluctuations in the index, then a moving average processing method can be added to the resilience index, or at least a minimum sample size should be required before outputting a formal score. For example, during the production week of batch P202501, the Monday night shift experienced increased fluctuations in N1 and N2 due to rising ambient humidity. The system triggered compensation 50 times, with 45 of those compensations being successful. Based on this, the management platform generated a resilience assessment data package for this shift and indicated that the resilience index for this shift was lower than the average of the previous week. Further investigation by the process engineer revealed that although the compensation system effectively absorbed most of the risks, the disturbance frequency of the upstream drying equipment had increased significantly, indicating that the overall anti-interference capability of the production line was declining, and equipment calibration and maintenance needed to be arranged. The purpose of this step is to elevate single-time compensation behavior into a periodic production line health evaluation indicator, so as to achieve quantitative monitoring of the production system's anti-interference capability and provide a basis for equipment maintenance, process optimization and production scheduling.

[0027] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A smart production line control platform system based on blockchain and industrial big data, characterized in that: include: The data acquisition module is configured to acquire time-series status data and flow log data of each node in a multi-node production line process through an industrial IoT interface; The causal analysis module is configured to process the time-series state data and the flow log data based on a preset causal coupling model, calculate the transmission influence factor of the state fluctuation of the upstream node on the quality of the downstream node, and quantify the cross-node deviation cumulative effect based on the transmission influence factor to generate a dynamic defect risk index. The smart contract execution module is configured to input the dynamic defect risk index into a smart contract deployed on the blockchain network through a blockchain remote call interface, and compare the dynamic defect risk index with a preset risk threshold through the smart contract. If the dynamic defect risk index is greater than or equal to the preset risk threshold, then the process compensation instruction for the downstream node is generated through the smart contract and sent to the corresponding downstream node through the programmable logic controller or device gateway to update the node parameter status. At the same time, the deviation compensation action information corresponding to the parameter status update is stored on the blockchain. If the dynamic defect risk index is less than the preset risk threshold, a state preservation flag is generated to control the downstream node to maintain the current parameter state, and the state preservation flag is stored on the blockchain for evidence. The traceability management module is configured to generate a product lifecycle causal traceability tree based on the flow log data and the deviation compensation action information or the state maintenance marker stored on the blockchain.

2. The intelligent production line control platform system based on blockchain and industrial big data as described in claim 1, characterized in that, The causal coupling model includes Bayesian networks and physical mechanism knowledge graphs; the causal analysis module is specifically configured as follows when calculating the transmission influence factor of the state fluctuations of upstream nodes on the quality of downstream nodes: Based on the physical mechanism knowledge graph, cross-node variable relationships are extracted; The cross-node variable associations and the time-series state data are input into the Bayesian network to calculate the transmission influence factor.

3. The intelligent production line control platform system based on blockchain and industrial big data according to claim 2, characterized in that, When the causal analysis module quantifies the cumulative effect of cross-node deviation based on the transmission influencing factors and generates a dynamic defect risk index, the specific configuration is as follows: Calculate the difference between the timing state data of the current node in each node and the preset baseline state data, and use it as the parameter deviation value; The single-node risk value is obtained by multiplying the parameter deviation value with the transmission influence factor. Based on the node order determined by the flow log data, the risk values ​​of the single nodes are aggregated using an exponential decay function that assigns weights in descending order of the node order, thereby generating the dynamic defect risk index.

4. The intelligent production line control platform system based on blockchain and industrial big data according to claim 1, characterized in that, When the smart contract execution module generates process compensation instructions for downstream nodes through the smart contract, it is specifically configured as follows: Extract the target downstream node corresponding to the dynamic defect risk index; The system uses a built-in database to match the range of available adjustable parameters corresponding to the target downstream node from a preset process knowledge base. Using the minimization of the dynamic defect risk index as the objective function, the optimal process parameters are calculated by iteratively solving the problem within the available adjustable parameter range using a preset optimization algorithm. The optimal process parameters are encapsulated into the process compensation instruction.

5. The intelligent production line control platform system based on blockchain and industrial big data according to claim 1, characterized in that, The system also includes a privacy computing module, which is configured as follows: Before uploading the deviation compensation action information or the state maintenance flag to the blockchain for evidence storage, the parameters in the process compensation instruction or the state maintenance flag are encrypted based on the zero-knowledge proof algorithm to generate quality compliance proof data. The quality compliance certification data is broadcast to the blockchain network as part of the deviation compensation action information or the state maintenance marker for distributed consensus and storage.

6. The intelligent production line control platform system based on blockchain and industrial big data according to claim 2, characterized in that, The traceability management module is specifically configured as follows when generating a causal traceability tree for the entire product lifecycle: Obtain the product identifier of the target product; Based on the product identifier, retrieve the corresponding circulation log data in the blockchain network, as well as the deviation compensation action information or the state maintenance marker; and based on the circulation log data and the cross-node variable correlation obtained by the causal analysis module, parse out the node state of the corresponding node and the deviation evolution path across nodes. By linking the flow log data, deviation compensation action information, or state retention markers according to the time series, a causal traceability tree for the entire product lifecycle is constructed, which includes node states, deviation evolution paths, and deviation compensation action information or state retention markers.

7. The intelligent production line control platform system based on blockchain and industrial big data according to claim 1, characterized in that, The system also includes a resilience assessment module, which is configured as follows: The non-negative triggering ratio of the smart contract triggering the process compensation instruction and the corresponding compensation success rate are statistically analyzed within a preset time period; wherein the triggering ratio is the proportion of the number of product units that trigger the process compensation instruction within the preset time period to the total number of product units within that period. If the trigger ratio is greater than zero, the ratio of the compensation success rate to the trigger ratio is used as the overall anti-interference resilience index of the production line. If the trigger ratio is equal to zero, the overall anti-interference resilience index of the production line is set to a preset full score benchmark value; the preset full score benchmark value is used to characterize that the production line is not subjected to interference exceeding the preset tolerance within the preset time period; The overall anti-interference toughness index of the production line is encapsulated into a toughness evaluation data package and output.

Images