Industrial production line distributed collaborative processing and control system based on hon gming

Abstract

The application provides a distributed collaborative processing and control system based on a Hongmeng for an industrial production line, relates to the technical field of production line design, and effectively overcomes the defects of the state lag of the offline planning and dynamic reality of the prior art through the construction of a real-time closed-loop feedback mechanism of deep fusion of the physical domain and the digital domain, and fills the semantic gap between the optimal solution of the algorithm and the feasible solution of the engineering through the introduction of a diagnosis decision model based on empirical data; the system generates a diagnosis signal containing a clear fault mode and a causal correlation confidence score based on the characteristic library of the known fault mode by analyzing the causal correlation between the physical stress and the digital communication behavior in real time, and then executes a hierarchical adaptive treatment matched with the diagnosis confidence, thereby ensuring the high synchronization of the control strategy and the physical reality, and having the theoretical optimality and the practical robustness, and significantly improving the production line efficiency and the flexible response capability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of production line design, in particular to a distributed collaborative processing and control system for industrial production lines based on the Hongmeng system. BACKGROUND

[0002] The industrial production line collaborative processing and control system is a comprehensive technical system integrating intelligent perception, data analysis and control execution. It connects the originally isolated equipment, processes and material flows on the production line into an organic whole through communication networks and collaborative algorithms, realizes real-time sharing and intelligent analysis of state information in the whole production line.

[0003] In the prior art, the patent number CN118364724B, named island assembly flexible production line design method and system. The invention includes: obtaining a grid map by gridizing a digital automotive assembly workshop, obtaining all node coordinates according to the grid map, wherein the node coordinates are the coordinates of the intersection points of the grid; selecting key nodes from all node coordinates, obtaining a weighted undirected graph according to the key nodes; obtaining the key nodes through which the path passes, generating a multi-generation initial assembly island path population, wherein the initial assembly island path population includes an initial assembly island path, and the initial assembly island path includes the node coordinates of the key nodes through which the initial assembly island path passes; obtaining a path set according to the weighted undirected graph and the initial assembly island path population; and arranging the assembly island according to the path set.

[0004] However, in the prior art, especially in the application of automobile assembly, the industrial production line collaborative processing and control system often has the following technical defects:

[0005] In the prior art, the layout design method for island assembly flexible production line usually uses offline optimization algorithm to plan the static topological model of the workshop. For example, by abstracting the workshop environment into a fixed grid map and a weighted undirected graph, and using heuristic search techniques such as genetic algorithm, an optimal or suboptimal path layout under certain constraints is obtained. However, this method has significant technical defects: the planning process is based on a static, idealized snapshot of the physical environment, and does not establish a closed-loop feedback mechanism between the actual running state of the production line. In the real industrial production environment, such as the appearance of temporary obstacles, changes in dynamic material flow, and real-time adjustment of production plans, uncertainty disturbances are normal. Therefore, this one-off offline planning scheme will quickly lose its "optimality" after deployment as the production line dynamic environment evolves, and cannot realize adaptive adjustment and real-time reconstruction of the production line layout, resulting in "state step-out" between the planning model and the physical reality, and further affecting the overall operation efficiency and flexible response capability of the production line.

[0006] Further, the multi-objective optimization algorithm adopted in the prior art usually outputs a final result which is a set of Pareto optimal frontiers containing multiple non-dominated solutions. This provides the decision maker with a variety of choices in theory, but introduces new technical problems in practice. First, the mathematical model on which the algorithm is based when performing optimization often cannot fully represent the complex and difficult-to-quantify operational constraints and implicit knowledge in the real world, such as the potential failure risk of equipment, the vibration impact of a specific area, or the convenience of human operation. This leads to a huge semantic gap between the "optimal solution" identified by the algorithm and the "best feasible solution" in engineering practice. Second, the mechanism (such as crowded distance calculation) adopted by the algorithm to ensure the diversity of solutions is limited to the objective function space, and does not directly evaluate the robustness or anti-disturbance ability of the solution, which may produce "fragile solutions" that perform well under ideal conditions but deteriorate rapidly under minor disturbances. Therefore, how to bridge the gap between the optimal solution of the algorithm and the operational feasibility, and ensure that the layout scheme has both theoretical optimality and real robustness, is a technical problem that needs to be solved in this field. SUMMARY

[0007] The purpose of the present application is to provide a distributed collaborative processing and control system for industrial production lines based on the Hongmeng system to solve the problems raised in the background art.

[0008] To achieve the above purpose, the present application provides the following technical solutions:

[0009] The distributed collaborative processing and control system for industrial production lines based on the Hongmeng system specifically comprises:

[0010] A first data acquisition module for acquiring first data representing the physical and mechanical stress state of a target industrial device in a production line within a preset time window;

[0011] A second data acquisition module for acquiring second data representing communication behavior in a Hongmeng distributed system of the target industrial device within the preset time window;

[0012] A state anomaly correlation analysis module for performing time-series causal correlation analysis on the first data and the second data to determine whether there is a preset causal relationship between the abnormality of the physical and mechanical stress state and the abnormality of the communication behavior;

[0013] A diagnostic signal generation module for generating a high-confidence diagnostic signal representing a causal resonance state based on the analysis results of the time-series causal correlation analysis when it is determined that there is a preset causal relationship;

[0014] A collaborative control execution module for receiving the high-confidence diagnostic signal and performing multiple preset control operations in linkage based on the high-confidence diagnostic signal.

[0015] Further, the first data acquisition module acquires original vibration time-domain signals reflecting the running state of the target industrial equipment through vibration sensors deployed in the target industrial equipment and its adjacent physical environment.

[0016] Further, the first data acquisition module acquires original vibration time-domain signals reflecting the running state of the target industrial equipment through vibration sensors deployed in the target industrial equipment and its adjacent physical environment.

[0017] Further, the specific process of performing spectrum analysis on the original vibration time-domain signals in the first data acquisition module includes:

[0018] Before performing spectrum analysis, a preset band-pass filter is first applied to the original vibration time-domain signals to filter out background noise bands irrelevant to known mechanical failure modes of the target industrial equipment.

[0019] Further, the specific process of performing spectrum analysis on the original vibration time-domain signals in the first data acquisition module includes:

[0020] Further, the second data acquisition module non-invasively monitors the communication soft bus of the Hongmeng distributed system to capture state synchronization messages and collaborative task message streams related to the target industrial equipment.

[0021] Further, the specific process of performing spectrum analysis on the original vibration time-domain signals in the first data acquisition module includes:

[0022] Further, the specific process of performing spectrum analysis on the original vibration time-domain signals in the first data acquisition module includes:

[0023] Further, the state anomaly correlation analysis module includes:

[0024] Further, the specific process of performing spectrum analysis on the original vibration time-domain signals in the first data acquisition module includes:

[0025] Further, a time sequence difference between the first time point and the second time point is calculated, and the time sequence difference is taken as a core characteristic parameter representing the state anormaly correlation between the physical domain and the digital domain; the core characteristic parameter is compared with a preset fault characteristic library in which a plurality of known fault modes and corresponding time sequence difference characteristic intervals are stored, to determine whether a preset causal relationship exists, and to output a causal correlation confidence score.

[0026] Further, the construction and updating process of the fault characteristic library in the state anormaly correlation analysis module is specifically as follows:

[0027] The time sequence of the first data and the second data under different fault modes is obtained by offline mining analysis of historical fault data or by fault injection experiment on the target industrial equipment in a controlled environment, and the statistical distribution of the core characteristic parameter is extracted therefrom, to establish or optimize the time sequence difference characteristic interval.

[0028] Further, the diagnostic signal generation module is configured to receive the causal correlation confidence score representing the certainty of the causal relationship determined by the state anormaly correlation analysis module and the dominant fault mode determined.

[0029] The causal correlation confidence score is mapped to discrete risk levels with clear treatment direction by comparing it with at least two sequentially increasing preset confidence thresholds; when the causal correlation confidence score exceeds the highest level of confidence threshold, a high-confidence diagnostic signal containing the dominant fault mode identifier and marked as the highest risk level is generated, to trigger subsequent linkage control operation.

[0030] Further, when the diagnostic signal generation module generates the high-confidence diagnostic signal, it encapsulates the key evidence information leading to the determination, including:

[0031] The original value of the causal correlation confidence score, the digital physical delay duration, the communication abnormality starting time point of the triggering event, and the physical stress peak time point.

[0032] Further, the collaborative control execution module is configured to, after receiving the high-confidence diagnostic signal, retrieve and match a corresponding linkage control instruction sequence from a preset control strategy matrix according to the dominant fault mode identifier contained in the high-confidence diagnostic signal.

[0033] Based on the original value of the causal correlation confidence score in the high-confidence diagnostic signal, the key control parameters in the linkage control instruction sequence are dynamically adjusted to realize adaptive disposal response matched with fault diagnosis confidence and hierarchical, and finally the adjusted linkage control instruction sequence is issued through the communication soft bus of the Hongmeng distributed system to the related equipment including the target industrial equipment for execution.

[0034] Further, the linkage control instruction sequence in the cooperative control execution module includes three types of parallel operations:

[0035] One is the local safety avoidance operation for the target industrial equipment, including reducing the running speed or switching to a safe working mode;

[0036] The second is the cooperative suspension and material rerouting instruction sent to the upstream and downstream associated equipment in the production line to prevent fault propagation or secondary effects;

[0037] The third is the alarm and maintenance request containing the complete diagnostic evidence chain pushed to the manufacturing execution system and the equipment maintenance work order system.

[0038] Compared with the prior art, the beneficial effects of the present application are:

[0039] The first data acquisition module and the second data acquisition module of the present application uninterruptedly collect the first data reflecting the physical and mechanical stress state of the equipment and the second data representing the communication behavior in the Hongmeng distributed system, and the state anomaly correlation analysis module performs real-time time series causal correlation analysis on the two data, thereby establishing an online diagnosis and control closed loop closely coupled with the actual running state of the production line. This mechanism replaces the offline planning relying on static topological model, can instantly perceive and respond to the cooperative anomaly between the physical domain and the digital domain caused by dynamic disturbance, and then execute adaptive disposal through the cooperative control execution module, thereby fundamentally ensuring the high synchronization of the control strategy and the physical reality, and significantly improving the overall running efficiency and flexible response ability of the production line.

[0040] The application successfully bridges the semantic gap between the optimal solution of the algorithm and the engineering feasible solution in the prior art, and ensures the real robustness of the control scheme; instead of relying on abstract mathematical models that cannot fully characterize complex reality for optimization, the state anormaly correlation analysis module uses a fault feature library that is constructed based on historical data and fault injection experiments and stores a plurality of known fault modes and their corresponding time difference characteristic intervals as a decision basis, thereby internalizing the operational constraints and implicit knowledge that are difficult to quantify in the diagnostic model; further, the diagnostic signal generation module outputs not a single Pareto solution set, but a high-confidence diagnostic signal containing the dominant fault mode identifier and the causal correlation confidence score, and the cooperative control execution module executes a graded adaptive response that matches the diagnostic confidence, which ensures that the strength of the system response is proportional to the certainty of the risk, avoids the "fragile solution" that deteriorates sharply in performance under a small disturbance, and realizes the high unification of theoretical optimality and real robustness. BRIEF DESCRIPTION OF DRAWINGS

[0041] Fig. 1 is a schematic diagram of the system framework structure of the application;

[0042] Fig. 2 is a flowchart of the state anormaly correlation analysis module and the diagnostic signal generation module;

[0043] Fig. 3 is a specific flowchart of the cooperative control execution module. DETAILED DESCRIPTION

[0044] In order to make the above-mentioned purposes, features and advantages of the application more obvious and easy to understand, the specific embodiments of the application will be described in detail below with reference to the accompanying drawings.

[0045] In the following description, many specific details are set forth in order to provide a thorough understanding of the application, but the application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the application, therefore the application is not limited by the specific embodiments disclosed below.

[0046] Embodiment one:

[0047] Please refer to Figs. 1 to 3 The application provides a technical solution: a distributed cooperative processing and control system for industrial production lines based on the Hongmeng, specifically comprising:

[0048] A first data acquisition module is used to acquire first data representing the physical and mechanical stress state of the target industrial equipment in the production line within a preset time window;

[0049] a second data acquisition module configured to acquire second data representing communication behavior in the distributed system within the preset time window;

[0050] a state anomaly correlation analysis module configured to perform time-causal correlation analysis on the first data and the second data to determine whether a preset causal relationship exists between the abnormality of the physical mechanical stress state and the abnormality of the communication behavior;

[0051] a diagnostic signal generation module configured to generate a high-confidence diagnostic signal representing a causal resonance state based on the analysis result of the time-causal correlation analysis when it is determined that the preset causal relationship exists;

[0052] a collaborative control execution module configured to receive the high-confidence diagnostic signal and perform a plurality of preset control operations in linkage based on the high-confidence diagnostic signal.

[0053] In this embodiment, the distributed system is an existing operating system product developed by Huawei and widely used in the market. The core of the distributed system is to fuse multiple independent physical devices into a super terminal in a logical sense at the system bottom layer based on the "distributed soft bus" technology, so as to realize hardware resource sharing and task cross-device collaboration. In the second data acquisition module, the distributed system is a microkernel-based distributed operating system for all scenarios. Through the built-in distributed communication, data management and task scheduling services, the distributed system provides standardized cross-device collaboration capabilities for upper-layer applications, which is used as a technical platform for distributed data acquisition and control in this technical solution.

[0054] Embodiment Two:

[0055] The first data acquisition module collects original vibration time-domain signals reflecting the running state of the target industrial equipment through vibration sensors deployed in the target industrial equipment and its adjacent physical environment.

[0056] In addition, frequency spectrum analysis is performed on the original vibration time-domain signals to generate a multi-dimensional frequency spectrum data set containing frequency, amplitude and phase information, and the multi-dimensional frequency spectrum data set is output as first data.

[0057] In the first data acquisition module, the specific process of performing frequency spectrum analysis on the original vibration time-domain signals includes:

[0058] Before performing the frequency spectrum analysis, a preset band-pass filtering process is first applied to the original vibration time-domain signals to filter out background noise frequency bands irrelevant to known mechanical fault modes of the target industrial equipment.

[0059] And, after generating the multi-dimensional spectrum dataset, a normalization process is performed on the amplitude of each frequency component in the multi-dimensional spectrum dataset to eliminate measurement bias introduced by the range or sensitivity difference of different vibration sensors.

[0060] The second data acquisition module captures state synchronization messages and collaborative task message streams related to the target industrial equipment by non-invasively monitoring the communication software bus of the distributed system.

[0061] Based on the captured state synchronization messages and collaborative task message streams, a plurality of quantitative indicators are calculated to represent the communication behavior, including a time sequence deviation degree indicator reflecting the stability of device state reporting, and an information complexity indicator reflecting the smoothness of multi-device collaboration; and the time sequence deviation degree indicator and the information complexity indicator are output as the second data.

[0062] The monitoring function in the second data acquisition module is a software probe embedded in the distributed communication management service of the operating system.

[0063] In this embodiment, the first data acquisition module aims to provide high-fidelity physical state representation data; the specific implementation process is as follows:

[0064] First, after collecting the original time domain signal through the vibration sensor, a pre-set band-pass filter processing is applied to match the known fault characteristic frequency, the purpose of which is to maximize the suppression of background noise irrelevant to the diagnostic target, and to preliminarily improve the signal-to-noise ratio; then, after performing spectrum analysis, a normalization process is performed on the amplitude of each frequency component in the obtained spectrum data set, the purpose of which is to eliminate systematic measurement bias introduced by individual differences of sensors, and to ensure the comparability and consistency of multi-source data.

[0065] To objectively evaluate the effectiveness of the output first data, the first data acquisition module is embedded with a "physical feature prominence score" (hereinafter referred to as ) calculation model, the specific construction and definition process of which is as follows:

[0066] First, a "feature signal-to-noise ratio" (hereinafter referred to as ) is defined. This indicator is obtained by calculating the logarithmic ratio between the total signal energy (hereinafter referred to as ) in the feature frequency band determined by the band-pass filter and the total signal energy (hereinafter referred to as ) in the adjacent noise frequency band. This definition ensures that can sensitively reflect the prominence of the fault feature signal relative to the background noise.

[0067] Subsequently, the calculated As the independent variable, input is a predefined, monotonically increasing Sigmoid function. This function can have a wide range of... The value is non-linearly mapped to the open interval (0, 1), and its output is the final value. This model construction method makes it possible for when When the value approaches 1, it clearly indicates that the current data has a high signal-to-noise ratio, significant features, and reliable quality; when it approaches 0, it clearly indicates that the data is overwhelmed by noise, the features are blurred, and it is not suitable for accurate diagnosis.

[0068] Furthermore, the second data acquisition module aims to provide a quantitative assessment of the communication behavior of the HarmonyOS distributed system. Its implementation includes:

[0069] By using a non-intrusive software probe deployed within the operating system, the communication soft bus is monitored, and two mutually orthogonal core quantitative indicators are calculated: one is the "timing deviation indicator" (denoted as...). The first is the "information complexity index" (denoted as ), used to characterize the jitter and deviation of the status message timestamp, reflecting the stability of the time dimension; the second is the "information complexity index" (denoted as ). (), used to characterize the degree of disorder in message types and order in a message flow, reflecting the orderliness of the logical dimension.

[0070] To comprehensively assess the overall health status of the digital domain, a "Digital Domain Communication Health Score" (denoted as ) is constructed. The computational model of ) is constructed and defined as follows:

[0071] The computational model is mathematically a weighted arithmetic mean model; it undergoes reverse normalization (to ensure that a higher index represents a better state). and As input, through a set of preset weights (respectively...) and A linear combination of the two (where the sum of the two is 1) will ultimately output a value located in the closed interval [0, 1]. ;

[0072] The weight and These are pre-configured adjustable parameters based on the business characteristics of specific industrial application scenarios, including increasing the time synchronization threshold in scenarios with stringent time synchronization requirements. The value of is determined; however, in scenarios with high requirements for multi-device collaborative logic, the value is increased. The value of ; this weight definition process enables this evaluation model to have scenario adaptability, and can be flexibly adjusted according to the diagnostic focus; the model output When the value approaches 1, it represents a healthy and stable communication; when the value approaches 0, it represents a significant time sequence or logic anomaly.

[0073] Embodiment three:

[0074] The state anomaly correlation analysis module comprises:

[0075] The first data and the second data are received and high-precision timestamp alignment processing is performed; in the aligned data stream, a fault feature frequency component of the first data is identified, a first time point at which a significant energy transition occurs is identified, and a second time point at which the second data first exceeds a preset dynamic baseline is identified;

[0076] A time sequence difference value between the first time point and the second time point is calculated, and the time sequence difference value is taken as a core feature parameter representing the correlation between the physical domain and the digital domain state anomaly; the core feature parameter is compared with a preset fault feature library in which a plurality of known fault modes and corresponding time sequence difference value feature intervals are stored, to determine whether a preset causal relationship exists, and a causal correlation confidence score is output.

[0077] In the state anomaly correlation analysis module, the construction and updating process of the fault feature library is specifically as follows:

[0078] The time series of the first data and the second data under different fault modes are obtained by offline mining analysis on historical fault data or by fault injection experiments on target industrial equipment in a controlled environment, and the statistical distribution of the core feature parameter is extracted therefrom, to establish or optimize the time sequence difference value feature interval.

[0079] In this embodiment, the quantification model of the causal correlation confidence score and the definition process thereof are specifically as follows:

[0080] To accurately quantify the causal relationship between the physical domain and the digital domain state anomaly, this embodiment defines a confidence calculation method based on a statistical probability model, aiming to compare the digital physical delay time (denoted as ) measured in real time as the core feature parameter, with the “time sequence difference value feature interval” in the fault feature library for the fault mode , and output a standardized “single-mode causal correlation confidence” (denoted as ).

[0081] Specifically, the construction and definition process of the quantification model is as follows:

[0082] The model is based on a normalized Gaussian probability density function. The logical basis for this choice is that, for a specific physical fault process, the “digital physical delay time” The multiple measurement values of the physical process usually present a normal distribution around a stable central value. Therefore, the Gaussian model can best reflect the inherent statistical characteristics of the physical process.

[0083] For any fault mode to be diagnosed , the calculation of its confidence depends on the absolute deviation between the real-time measurement and the central value of the feature interval of the mode retrieved from the fault feature library . The deviation is used as the independent variable of an exponential decay function, ensuring that the value of the confidence is strictly monotonically decreasing with the deviation. The distribution width of the feature interval is then used as a parameter to control the decay rate of the exponential function, the smaller the value, the more concentrated the timing characteristics of the fault mode, the lower the tolerance of the model to the deviation, and the more stringent the matching requirements.

[0084] After the above calculation and normalization, the output is a dimensionless value in the closed interval [0, 1]. When the value tends to 1, it clearly indicates that the current measurement is highly consistent with the typical timing characteristics of the fault mode , and the likelihood of the causal relationship is very high; when the value tends to 0, it clearly indicates that there is a significant deviation between the two, and the causal relationship can be excluded with high confidence. Finally, the module outputs the maximum value of the confidence calculated for all known fault modes as the final "causal correlation confidence score" .

[0085] Further, the construction and definition process of the fault feature library and its parameters are as follows:

[0086] The fault feature library is the cornerstone of the accurate calculation of the above quantitative model, which provides a highly reliable reference benchmark for each known fault mode , namely the central value of the feature interval and the distribution width . The construction and definition of these parameters follow a rigorous offline processing procedure that includes data collection, statistical modeling, and verification optimization:

[0087] First, the training data is systematically collected through two approaches. One is to mine the large amount of historical operation data and fault records of the equipment offline, and to filter out the effective data segments with clearly labeled fault types. The other is to conduct active fault injection experiments on the target industrial equipment in a controlled experimental environment to simulate various typical fault modes. In both approaches, the time series of the first data and the second data are collected synchronously to build a sample database for subsequent analysis.

[0088] For each fault mode in the database , the "digital physical delay time" values of all samples are extracted to form a sample set. Then, statistical analysis is performed on the sample set, and the sample mean is defined as the center value of the "time sequence difference feature interval" of the fault mode , and the sample standard deviation is defined as the distribution width of the interval . This definition ensures that the constructed parameters objectively and quantitatively reflect the central tendency and dispersion of the inherent time sequence characteristics of the fault mode.

[0089] To ensure the accuracy and robustness of the defined parameters, verification and optimization steps need to be performed. The collected data is divided into a training set and a test set. The and parameters calculated on the training set are used to classify the samples in the test set and evaluate their classification accuracy, recall rate, and other performance indicators. If the performance indicators do not meet the pre-set engineering requirements, return to the data collection and modeling step, and iteratively optimize the parameters by removing abnormal samples, increasing the number of samples, or optimizing data preprocessing methods, until the verification results meet the requirements. Finally, this set of well-verified and optimized parameters are stored in the fault feature library as the final "time sequence fingerprint" of the fault mode.

[0090] The diagnostic signal generation module is configured to receive the causal correlation confidence score representing the certainty of the causal relationship determination and the dominant fault mode determined by the state anoma lysis correlation analysis module.

[0091] The causal correlation confidence score is compared with at least two sequentially increasing pre-set confidence thresholds to map it into discrete risk levels with clear disposition directionality. When the causal correlation confidence score exceeds the highest level of confidence threshold, a high-confidence diagnostic signal containing the dominant fault mode identifier and marked as the highest risk level is generated to trigger subsequent linkage control operations.

[0092] When generating the high-confidence diagnostic signal, the diagnostic signal generation module encapsulates the key evidence information that leads to this determination, which includes:

[0093] The original numerical value of the causal correlation confidence score, the numerical physical delay duration, the communication anomaly starting time point of the triggering event, and the physical stress peak time point.

[0094] In this embodiment, the input "causal correlation confidence score" is identified as ; the input "dominant failure mode code" is identified as ; the internal preset "alarm threshold" is identified as ; the internal preset "early warning threshold" is identified as The calculated "diagnostic risk level" is identified as ;

[0095] In the diagnostic signal generation module, if the value of is less than or equal to , then is assigned to the first monitoring state;

[0096] If the value of is greater than and less than or equal to , then is assigned to the second early warning state;

[0097] If the value of is greater than , then is assigned to the third alarm state;

[0098] In the logic chain of , the only dynamic input is ; and and are static, pre-set decision boundaries; and is the only output. The entire logic is a feedforward comparison and decision process; this logic structure ensures that the increase of input results in the output remaining unchanged or jumping to a higher level, and there is a strict monotonic non-decreasing relationship between the two, i.e., a stepwise proportional relationship, which guarantees the logical consistency of the system response;

[0099] The value range of the input is [0, 1], and the value range of the output is a discrete set {0, 1, 2, 3};

[0100] When the output is the first monitoring state, it indicates that The system determines that the current existing abnormal correlation is weak or within the normal range of fluctuations, and no active intervention is required; at this time, the system maintains normal operation, but increases the monitoring frequency of related data;

[0101] When the output is the second early warning state, it means has been exceeded , indicating that the system has detected a potential failure sign with medium confidence; trigger non-interruptive early warning operations, including sending inspection notifications to operation and maintenance personnel, automatically recording detailed context data logs, or starting an intensive monitoring mode for specific components, which creates a valuable time window for early detection and preventive maintenance of failures;

[0102] When the output is the third alarm state, it means has been exceeded , indicating that the system determines that a failure is about to occur or has occurred with high confidence; at this time, a high-confidence diagnosis signal is immediately generated, triggering downstream collaborative control execution modules to execute emergency handling operations to prevent the situation from expanding and ensure safety;

[0103] In the diagnosis signal generation module, the following steps are used to determine and :

[0104] 1) Based on historical cases and equipment failure mechanisms, the confidence levels of "worth paying attention to", "need to prepare", and "must act" are conceptually defined; at the same time, through statistical analysis of historical data, a relationship model between the value and the actual failure probability is established using ROC curve analysis, quantifying the diagnostic accuracy and recall rate at different confidence levels;

[0105] 2) Develop preliminary division criteria and verify: combined with expert experience and model evaluation, develop preliminary division criteria, such as setting to 0.60, to 0.85; use an independent validation data set to backtest this standard, specifically measuring the early warning lead time, alarm false negative rate, and overall false alarm cost of the system (including production interruption losses and unnecessary maintenance overhead) at these threshold values.

[0106] 3) Optimize and determine the final division criteria: based on the verification results, fine-tune the threshold values to minimize the overall risk of the system; if the verification finds that a large number of failures have caused irreversible damage when reaches 0.80, then ​Down from 0.85 to 0.80 or below. Through this closed-loop optimization process, a set of thresholds that can achieve the best balance between sensitivity, specificity and economy of the system is finally determined and solidified as system parameters.

[0107] The cooperative control execution module is configured to, after receiving the high-confidence diagnosis signal, retrieve and match to a corresponding set of linked control instruction sequences from a preset control strategy matrix according to the dominant fault mode identifier contained in the high-confidence diagnosis signal.

[0108] Based on the original value of the causal association confidence score in the high-confidence diagnosis signal, the key control parameters in the linked control instruction sequences are dynamically adjusted to achieve adaptive response to the fault diagnosis confidence that matches and is graded, and finally the adjusted linked control instruction sequences are issued through the communication soft bus of the Hongmeng distributed system to the related devices including the target industrial device for execution.

[0109] Further, the "control strategy matrix" is a pre-constructed expert knowledge base, and its core principle is to quickly map the diagnosed specific fault mode to a set of accurate and executable linked control actions. Its construction process follows four key steps: first, enumerate and assign a unique identifier to each fault through fault mode and effect analysis; second, for each fault identifier, design a set of linked control strategies including target devices, operation instructions and execution sequence; third, parameterize the key control values in the instructions to reserve interfaces for subsequent dynamic adjustment; finally, solidify the mapping relationship from "fault identifier" to "parameterized instruction sequence" as a structured data that can be quickly retrieved by computer, so as to realize automated decision-making from diagnosis to response.

[0110] The linked control instruction sequences in the cooperative control execution module include three types of parallel operations:

[0111] The first is the local safety avoidance operation for the target industrial device, including reducing the running speed or switching to a safe working mode;

[0112] The second is the cooperative suspension and material rerouting instructions sent to the upstream and downstream associated devices in the production line to prevent fault propagation or secondary effects;

[0113] The third is the alarm and maintenance request containing the complete diagnosis evidence chain pushed to the manufacturing execution system and device maintenance work order system.

[0114] In this embodiment, the retrieved "basic control template" is identified as ; the "maximum safety intervention amplitude" defined in the template is identified as ; and the calculated "control strength adjustment factor" is identified as ; the final generated "target device speed-down ratio" is identified as ;

[0115] Further, is equal to the input minus , and the obtained difference is divided by the difference between 1 and ; in the calculation formula of , the only independent variable is , while is a fixed reference benchmark; the formula structure ensures that the change of is linearly transferred to , and the function trend is monotonically increasing, i.e. increases linearly with the increase of , and they are strictly in direct proportion, which constitutes the basis of adaptive control;

[0116] When the value range of is [0, 1], the output value range of is limited within the interval [0, 1].

[0117] When the output is equal to 0: this is due to the input just exceeding the alarm threshold ; at this time, the certainty of the diagnostic conclusion just reaches the minimum standard that requires action to be taken; the system generates a very small control strength accordingly, which means that the intervention to the device is extremely gentle; while ensuring the response to potential risks, the production is maximally avoided from being interrupted due to the low confidence of the diagnostic conclusion;

[0118] When the output is equal to 1, it indicates that the input is equal to 1; at this time, the diagnostic conclusion has extremely high certainty, indicating that the device has a specific fault; the system generates a maximum control strength accordingly, which means that the intervention to the device will reach the preset safety upper limit; in the case of extremely high fault certainty, decisive measures are taken to provide the strongest protection, giving priority to personnel and equipment safety;

[0119] Further, Rde is equal to the product of the retrieved from and the calculated ; in the calculation process, represents the "maximum handling authority" for a specific fault , while This refers to the "percentage of permissions used" under the current diagnostic reliability. The logic of multiplying these two ensures that the final control output considers both the physical characteristics of the fault itself and the real-time determinism of the current diagnosis; where the physical characteristics of the fault itself are determined by... This is reflected in the fact that the real-time certainty of current diagnoses is determined by... reflect;

[0120] The algorithm of the collaborative control execution module only applies to the received diagnostic signals. Greater than It is activated at that time. Furthermore, the control strategy knowledge base must contain information related to... The corresponding entry will be selected; otherwise, the module will execute the default security operation.

[0121] Furthermore, to accommodate the limited precision of the digital processing system and improve computational efficiency, the system also includes a processing step that calculates the "control intensity adjustment factor". With a preset "minimum effective adjustment threshold" Compare;

[0122] like Less than Then in subsequent calculations In the steps, the The value of is forcibly set to 0; this processing method is a common technique used by those skilled in the art when implementing numerical algorithms. It aims to effectively process parameters that theoretically approach the target value in engineering, thereby ignoring the minimum values ​​that have no actual impact on the final result.

[0123] Key thresholds in this module The determination of [the value] follows these rigorous steps:

[0124] Step 1: Based on the failure mode and effects analysis results, assess the risk priority of different failures and initially define the confidence level ranges for "acceptable," "requiring attention," and "requiring intervention." Simultaneously, using a historical failure database, train a logistic regression classification model to analyze different failure scenarios. The probability distribution of actual failures occurring at the horizontal level;

[0125] Step Two: Based on expert opinions and model analysis, initially set... The threshold was set at 0.85. Subsequently, a set of independent historical datasets with real fault labels were used for backtesting to verify the results. The false alarm rate and false alarm rate of the system at this threshold were evaluated, and their impact on simulated production efficiency was calculated.

[0126] Step 3: Based on the verification results, if the false negative rate is higher than the safety baseline, then reduce the threshold. ; if the false alarm rate is too high, resulting in a loss of production efficiency beyond an acceptable range, then increase ; through iterative optimization, finally determine a value that achieves the best balance between safety and economy as the final standard, and solidify it into the system.

[0127] It should be noted that all the calculation formulas in this application file use regression analysis including but not limited to machine learning algorithms to deeply analyze the collected relevant parameters and identify their natural trends and mutual relationships. Professional software such as Python's Scikit-learn library or R language is used to automatically generate mathematical models that match the data. Then, through methods such as cross-validation, the performance of the model is objectively evaluated, and combined with continuous feedback and optimization, to ensure that the created formula truly reflects the inherent law of the data, thereby ensuring its effectiveness and accuracy. In all the calculation formulas in this application, the parameters in each formula are processed by consistent range of dimensionless to ensure that different physical quantities are compared on the same scale; the dimensionless technique includes but is not limited to Min-Max-Normalization, Z-Score standardization;

[0128] The technical solutions of the present application can be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), FLASH, hard disk or optical disk, etc., including a number of instructions to make a computer device (which can be a personal computer, server, or network device, etc.) execute the methods of various embodiments of the present application.

[0129] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered a list of executable instructions for implementing logic functions, which can be specifically embodied in any computer-readable medium for use by or in conjunction with an instruction execution system, device or apparatus, such as a computer-based system, a system including a processor or other system that can fetch and execute instructions from an instruction execution system, device or apparatus. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate or transport a program for use by or in conjunction with an instruction execution system, device or apparatus, or in conjunction with these instruction execution systems, devices or apparatus.

[0130] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A distributed collaborative processing and control system for industrial production lines based on HarmonyOS, characterized in that: Specifically, it includes: The first data acquisition module is used to acquire first data representing the physical and mechanical stress state of the target industrial equipment in the production line within a preset time window. The second data acquisition module is used to acquire second data representing the communication behavior of the target industrial equipment in the HarmonyOS distributed system within the preset time window; specifically, it captures status synchronization messages and collaborative task message streams related to the target industrial equipment by non-intrusively monitoring the communication soft bus of the HarmonyOS distributed system. Based on the captured status synchronization messages and collaborative task message streams, various quantitative indicators to characterize the communication behavior are calculated, including a timing deviation indicator to reflect the stability of device status reporting and an information complexity indicator to reflect the smoothness of multi-device collaborative execution; and the timing deviation indicator and information complexity indicator are used as the second data output; wherein, the monitoring function in the second data acquisition module is a software probe embedded in the distributed communication management service of the operating system; The state deviation correlation analysis module is used to perform time-series causal correlation analysis on the first data and the second data to determine whether there is a preset causal relationship between the anomaly of physical mechanical stress state and the anomaly of communication behavior. This includes receiving the first data and the second data and performing high-precision timestamp alignment processing; identifying the fault characteristic frequency components of the first data in the aligned data stream, identifying the first time point where a significant energy transition occurs; and identifying the second time point where the second data first exceeds a preset dynamic baseline; then calculating the time difference between the first and second time points, and using this time difference as a core feature parameter characterizing the correlation between state deviation in the physical and digital domains; comparing the core feature parameter with a preset fault feature library that stores multiple known fault modes and their corresponding time difference feature intervals to determine whether a preset causal relationship exists, and outputting a causal correlation confidence score. The diagnostic signal generation module generates a high-confidence diagnostic signal to characterize the causal resonance state based on the analysis results of time-series causal correlation analysis when a preset causal relationship is determined. The collaborative control execution module is used to receive the high-confidence diagnostic signal and execute multiple preset control operations in a coordinated manner based on the high-confidence diagnostic signal.

2. The distributed collaborative processing and control system for industrial production lines based on HarmonyOS according to claim 1, characterized in that: The first data acquisition module collects raw vibration time-domain signals reflecting the operating status of the target industrial equipment by means of vibration sensors deployed in the target industrial equipment and its adjacent physical environment; Furthermore, spectral analysis is performed on the original vibration time-domain signal to generate a multidimensional spectral dataset containing frequency, amplitude, and phase information, and the multidimensional spectral dataset is output as the first data.

3. The distributed collaborative processing and control system for industrial production lines based on HarmonyOS according to claim 2, characterized in that: The specific process of performing spectral analysis on the original vibration time-domain signal in the first data acquisition module includes: Before performing spectrum analysis, a preset bandpass filter is first applied to the original vibration time-domain signal to filter out background noise frequency bands that are not related to the known mechanical fault modes of the target industrial equipment. Furthermore, after generating the multidimensional spectrum dataset, the amplitude of each frequency component in the multidimensional spectrum dataset is normalized to eliminate measurement biases caused by differences in the range or sensitivity of different vibration sensors.

4. The distributed collaborative processing and control system for industrial production lines based on HarmonyOS according to claim 3, characterized in that: In the state deviation correlation analysis module, the process of constructing and updating the fault feature database is as follows: By performing offline mining and analysis on historical fault data, or by conducting fault injection experiments on target industrial equipment in a controlled environment, the time series of first and second data under different fault modes can be obtained, and the statistical distribution of the core feature parameters can be extracted from them to establish or optimize the time series difference feature interval.

5. The distributed collaborative processing and control system for industrial production lines based on HarmonyOS according to claim 4, characterized in that: The diagnostic signal generation module is used to receive the causal association confidence score, which characterizes the determinacy of causal relationship, and the determined dominant failure mode, output by the state deviation correlation analysis module. The causal association confidence score is mapped to a discrete risk level with a clear treatment direction by comparing it with at least two successively increasing pre-set confidence thresholds; When the causal correlation confidence score exceeds the highest level confidence threshold, a high-confidence diagnostic signal containing the dominant failure mode identifier and marked as the highest risk level is generated to trigger subsequent linkage control operations.

6. The distributed collaborative processing and control system for industrial production lines based on HarmonyOS according to claim 5, characterized in that: When generating the high-confidence diagnostic signal, the diagnostic signal generation module encapsulates the key evidence information that led to this determination. This key evidence information includes: The raw values ​​of the causal association confidence score, the duration of digital physical delay, the start time of the communication anomaly that triggered the event, and the peak time of physical stress.

7. The distributed collaborative processing and control system for industrial production lines based on HarmonyOS according to claim 6, characterized in that: The collaborative control execution module is used to, upon receiving the high-confidence diagnostic signal, retrieve and match a set of corresponding linkage control command sequences from a preset control strategy matrix based on the dominant fault mode identifier contained in the high-confidence diagnostic signal. Based on the original values ​​of the causal correlation confidence scores in the high-confidence diagnostic signals, the key control parameters in the linkage control command sequence are dynamically adjusted to achieve an adaptive handling response that matches and is graded according to the fault diagnosis confidence. Finally, the adjusted linkage control command sequence is sent to relevant equipment, including the target industrial equipment, for execution via the communication soft bus of the HarmonyOS distributed system.

8. The distributed collaborative processing and control system for industrial production lines based on HarmonyOS according to claim 7, characterized in that: The coordinated control execution module includes a sequence of control instructions for three types of parallel operations: One is local safety avoidance operations for the target industrial equipment, including reducing the operating speed or switching to a safe operating mode; Secondly, coordinated pause and material rerouting instructions are sent to upstream and downstream related equipment within the production line to prevent the spread of the fault or secondary impacts. Thirdly, alarms and maintenance requests containing a complete chain of diagnostic evidence are pushed to the manufacturing execution system and equipment maintenance work order system.

Images