A data processing method and system for online collaboration and conflict detection of multi-specialty models, and an electronic device

Abstract

The embodiment of the application relates to a kind of data processing method, system and electronic equipment for multi-professional model online cooperation and conflict detection, belong to the technical field of electronic data processing, comprising: constructing multidimensional parameter space and mapping as discrete coordinate point and storing in distributed hash table, calculating the normalized Euclidean distance between coordinate points obtains parameter difference degree, mark as conflict association pair and construct conflict association graph below threshold value, identify the model group to be synchronized by graph decomposition. Obtain the local update direction vector of model in group, calculate global direction vector and update coordinate point, iterative cycle until no conflict association pair. The embodiment of the application realizes the unified representation of heterogeneous model by parameter space mapping, decouples global conflict into local sub-problems using graph decomposition, and combines consensus protocol to drive collaborative convergence. Under the premise of guaranteeing model autonomy, parameter conflicts are eliminated, and the efficiency of multi-professional model online cooperation is improved.

Description

Technical Field

[0001] This application relates to a data processing method, system, and electronic device for online collaboration and conflict detection of multi-disciplinary models, belonging to the technical field of electronic data processing. Background Technology

[0002] Multi-disciplinary online model collaboration refers to the process by which multiple subsystems from different professional fields (such as structural analysis models, fluid simulation models, thermodynamic models, etc.) run together and exchange parameters in the same engineering project. In the design of large-scale equipment or the simulation of complex engineering projects, each professional model is often maintained independently by different teams, each with its own set of quality of service parameters (such as mesh accuracy, time step, convergence threshold, etc.). In actual engineering, designers often encounter the following problem: increasing the mesh density of the structural analysis model requires a corresponding reduction in the computation time step of the fluid model; otherwise, non-physical oscillations will appear in the pressure data at the coupled boundaries. These constraints between parameters are difficult to enumerate in advance in documentation.

[0003] In existing collaborative systems, a central database is typically used to manage parameters. Each specialized model periodically reads parameter values ​​from the database from other models, and adjustments are made through manual meetings if inconsistencies are found. In a joint simulation project for engine blades, the aerodynamic and structural models of the turbine blades repeatedly had to stop for adjustments due to parameter conflicts, taking more than four weeks to complete the calibration cycle. Some systems introduced rule-based conflict detection modules, but the construction of the rule base relied on domain experts manually entering each rule. When the number of models exceeded five, the number of rule combinations increased exponentially, leading to a sharp rise in maintenance costs. More importantly, existing collaborative methods lack a mechanism to automatically identify groups of conflicting parameters and guide multi-model collaborative convergence. Each model often adjusts its own parameters independently, resulting in resolving a conflict with one model but triggering a new conflict with another, making the overall convergence process difficult to control.

[0004] Existing technologies lack solutions to the above problems and can no longer meet people's needs, so they urgently need to be improved. Summary of the Invention

[0005] The main objective of this application is to provide a data processing method, system, and electronic device for online collaboration and conflict detection of multi-disciplinary models, so as to solve the defects existing in related technologies.

[0006] The embodiments of this application are implemented using the following technical solutions: According to one aspect of the embodiments of this application, a data processing method for online collaboration and conflict detection of multi-professional models is provided, comprising: reading the initial model parameter sequence of each professional model; constructing a multi-dimensional parameter space based on the initial model parameter sequence; mapping each professional model to discrete coordinate points in the multi-dimensional parameter space, wherein the coordinate values ​​are composed of the current values ​​of service quality parameters in the professional model; generating a unique identifier for each discrete coordinate point using a hash function; and storing the unique identifier and the corresponding model parameter sequence in a distributed hash table in the form of key-value pairs; calculating the normalized Euclidean distance between any two discrete coordinate points in the multi-dimensional parameter space to obtain the parameter difference degree between each professional model; and when the parameter difference degree is lower than a preset difference threshold, linking the corresponding two professional models... Models are labeled as conflicting pairs; a conflicting association graph is constructed based on all conflicting pairs, and a graph decomposition operation is performed on the conflicting association graph to identify specialized models located in the same decomposed subgraph as a group of models to be synchronized; the local update direction vector of each specialized model in the group to be synchronized is obtained, and the local update direction vectors are exchanged in the group to be synchronized through a distributed consensus protocol, and the global update direction vector within the group is calculated. Each specialized model in the group to be synchronized is driven to move synchronously by one unit step along the global update direction vector to update its own discrete coordinate points, and the labeling of conflicting pairs and the identification of the group to be synchronized are re-executed based on the updated discrete coordinate points to form an iterative loop until there are no conflicting pairs in the multidimensional parameter space with parameter differences lower than a preset difference threshold.

[0007] According to at least one specific embodiment of the present application, when performing a graph decomposition operation on the conflict association graph, the graph decomposition operation is specifically a connected component decomposition; the connected component decomposition traverses all nodes of the conflict association graph using a depth-first search algorithm, and divides nodes that are connected by paths to each other into the same connected component.

[0008] According to at least one specific embodiment of the present application, the initial value of the unit step size is set to one-tenth of the minimum Euclidean distance between any two discrete coordinate points in the multidimensional parameter space. In subsequent iterations, when no new model group to be synchronized is identified in three consecutive iterations, the unit step size is adjusted to 90% of the current value.

[0009] According to at least one specific embodiment of the present application, the local update direction vector is determined by the first derivative of the optimization function of the service quality parameters of each professional model at the current discrete coordinate point. When the angle between the local update direction vector of any professional model and the global update direction vector within the group exceeds a preset angle threshold, the current professional model is marked as a direction abnormal model, and the weight of the current professional model in the vector weighted average calculation is reduced in subsequent iterations.

[0010] According to at least one specific embodiment of the present application, the orientation anomaly model after weight reduction moves along the vertical direction of the global update orientation vector within the group by a perturbation step, the perturbation step being proportional to the number of times the orientation anomaly model is continuously labeled.

[0011] According to at least one specific implementation of the embodiments of this application, after driving each professional model to synchronously move and update discrete coordinate points, the historical unique identifier stored in the distributed hash table is compared and verified with the unique identifier regenerated at the current discrete coordinate point; if the verification is inconsistent, a backtracking mechanism is triggered to restore the discrete coordinate points of all professional models in the model group to be synchronized to the coordinate points that passed the last verification, and a backtracking event is recorded.

[0012] According to at least one specific implementation of the embodiments of this application, when the number of consecutive backtracking events reaches a preset backtracking threshold, a local perturbation operation is performed. The local perturbation operation specifically involves generating a random perturbation vector for each professional model in the model group to be synchronized, the professional model moving independently along its own random perturbation vector, and then re-executing the conflict association pair marking step.

[0013] According to at least one specific embodiment of the present application, during the formation of the iterative loop, the trend of the size change of the model group to be synchronized is monitored. When the size of the model group to be synchronized continues to expand during the iteration, it is determined to be a conflict diffusion state, the unit step size is adjusted to half of the current step size, and the preset difference threshold is adjusted to 1.5 times the current threshold.

[0014] According to another aspect of the embodiments of this application, a data processing system for online collaboration and conflict detection of multi-professional models is provided, which implements the data processing method for online collaboration and conflict detection of multi-professional models, including: a unique identifier and model parameter sequence generation module, which reads the initial model parameter sequence of each professional model, constructs a multi-dimensional parameter space based on the initial model parameter sequence, maps each professional model to discrete coordinate points in the multi-dimensional parameter space, the coordinate values ​​are composed of the current values ​​of service quality parameters in the professional model, generates a unique identifier for each discrete coordinate point using a hash function, and stores the unique identifier and the corresponding model parameter sequence in a distributed hash table in the form of key-value pairs; a model group identification module, which calculates the normalized Euclidean distance between any two discrete coordinate points in the multi-dimensional parameter space to obtain the parameter difference degree between each professional model; when the... When the parameter difference is lower than a preset difference threshold, the two corresponding professional models are marked as conflict association pairs. A conflict association graph is constructed based on all conflict association pairs, and a graph decomposition operation is performed on the conflict association graph to identify professional models located in the same decomposed subgraph as a group of models to be synchronized. The parameter difference and threshold comparison module obtains the local update direction vector of each professional model in the group of models to be synchronized, exchanges the local update direction vectors in the group of models to be synchronized through a distributed consensus protocol, and calculates the global update direction vector in the group. It drives each professional model in the group of models to be synchronized to move synchronously by one unit step along the global update direction vector to update its own discrete coordinate points. Based on the updated discrete coordinate points, the marking of conflict association pairs and the identification of the group of models to be synchronized are re-executed to form an iterative loop until there are no conflict association pairs with parameter differences lower than the preset difference threshold in the multidimensional parameter space.

[0015] According to another aspect of the embodiments of this application, an electronic device is provided, including: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method.

[0016] The beneficial technical effects of the embodiments of this application are: This application provides a method and system for online collaboration and conflict detection of multi-disciplinary models. Through techniques such as parameter space mapping, conflict graph decomposition, and consensus-based collaborative updates, heterogeneous model parameters are uniformly mapped to a multi-dimensional space and hash identifiers are generated, achieving distributed, efficient management and rapid retrieval of model states. Furthermore, normalized Euclidean distance is used to accurately quantify parameter differences between models and construct a conflict association graph. A graph decomposition algorithm is used to decouple the global conflict problem into local synchronization sub-problems, reducing collaboration complexity. Based on this, a distributed consensus protocol aggregates local update directions to form a globally consistent strategy, driving synchronous iterative convergence of models within the conflict group. This achieves the goal of eliminating parameter conflicts while ensuring the autonomy of each model, realizing the technical effects of efficient online collaboration and conflict self-healing of multi-disciplinary models. It solves the technical problems of communication bottlenecks, poor scalability, and low collaboration efficiency caused by centralized coordination in existing technologies. Attached Figure Description

[0017] To more clearly illustrate the specific implementation methods of the embodiments of this application or the technical solutions in the prior art, the drawings used in the description of the specific implementation methods or the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a data processing method for online collaboration and conflict detection of multi-disciplinary models.

[0019] Figure 2 This is an architecture diagram of a data processing system used for online collaboration and conflict detection of multi-disciplinary models.

[0020] Figure 3 This is a schematic diagram of the electronic device. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the specific implementation methods in the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of the embodiments of this application.

[0022] like Figure 1 The data processing method shown for online collaboration and conflict detection of multi-disciplinary models includes: Step S1: Read the initial model parameter sequence of each professional model, construct a multi-dimensional parameter space based on the initial model parameter sequence, map each professional model to discrete coordinate points in the multi-dimensional parameter space, the coordinate values ​​are composed of the current values ​​of the service quality parameters in the professional model, generate a unique identifier for each discrete coordinate point using a hash function, and store the unique identifier and the corresponding model parameter sequence in a distributed hash table in the form of key-value pairs.

[0023] Step S2: Calculate the normalized Euclidean distance between any two discrete coordinate points in the multidimensional parameter space to obtain the parameter difference degree between each professional model; when the parameter difference degree is lower than the preset difference threshold, mark the corresponding two professional models as conflict association pairs; construct a conflict association graph based on all conflict association pairs, perform graph decomposition operation on the conflict association graph, and identify professional models located in the same decomposition subgraph as a group of models to be synchronized.

[0024] Step S3: Obtain the local update direction vector of each professional model in the model group to be synchronized. Exchange the local update direction vectors in the model group to be synchronized through a distributed consensus protocol and calculate the global update direction vector in the group. Drive each professional model in the model group to be synchronized to move synchronously by one unit step along the global update direction vector to update its own discrete coordinate points. Based on the updated discrete coordinate points, re-execute the conflict association pair marking and identification with the model group to be synchronized to form an iterative loop until there are no conflict association pairs in the multidimensional parameter space with parameter differences lower than a preset difference threshold.

[0025] In the technical solutions of steps S1 to S3 provided in this application embodiment, step S1 maps each professional model to a unified multi-dimensional parameter space and generates a hash identifier code. Step S1 establishes a traceable and tamper-proof data foundation for the parameter difference calculation in step S2. The hash identifier code of step S1 can also be used for consistency verification after coordinate point update in step S3. That is to say, the storage structure of step S1 directly supports the backtracking verification of step S3, and a collaborative relationship of positioning, detection, and verification is formed among the three steps.

[0026] In step S2, the normalized Euclidean distance is calculated based on the discrete coordinate points generated in step S1 to quantify the parameter differences. This is then used to construct a conflict correlation graph and perform connected component decomposition, transforming the global conflict detection problem into a local subgraph identification problem. The model group to be synchronized output in step S2 can be directly used as the operation object in step S3. Furthermore, the updated coordinate points in step S3 are fed back to step S2 to re-trigger conflict detection. This feedback mechanism between steps S2 and S3 makes the conflict resolution process adaptive, avoiding the technical defect of repeated oscillations between different conflict pairs when each model is adjusted independently.

[0027] Step S3 uses a distributed consensus protocol to exchange local update direction vectors and calculate the global direction within the model group to be synchronized, driving the entire model group to move synchronously and achieving collaborative convergence within the conflicting group. The model group size identified in step S2 can also serve as the basis for dynamically adjusting the step size and threshold in step S3. In other words, steps S2 and S3 form a linkage mechanism of conflict scale perception and parameter adaptive adjustment.

[0028] In summary, through the progressive and feedback relationship between steps S1 and S3, the technical effect of automatically detecting and collaboratively resolving parameter conflicts of multiple professional models without the need for a preset rule base or manual intervention is achieved.

[0029] Preferably, in step S2, when performing a graph decomposition operation on the conflict-affected graph, the graph decomposition operation is specifically a connected component decomposition. The connected component decomposition uses a depth-first search algorithm to traverse all nodes of the conflict-affected graph, classifying nodes that are connected by paths into the same connected component. A connected component refers to the largest connected subgraph in an undirected graph, where any two nodes are connected by paths, and no other nodes can be added without disrupting connectivity. Graph decomposition refers to the process of splitting a complete graph into several subgraphs according to specific rules; in this embodiment, it specifically refers to splitting into connected components based on connectivity. Connected component decomposition refers to a specific algorithm that uses a depth-first or breadth-first search to traverse the graph nodes and classify mutually reachable nodes into the same connected component.

[0030] Preferably, in step S2, the initial value of the unit step size is set to one-tenth of the minimum Euclidean distance between any two discrete coordinate points in the multidimensional parameter space. In subsequent iterations, if no new model group to be synchronized is identified in three consecutive iterations, the unit step size is adjusted to 90% of the current value.

[0031] Preferably, in step S2, the local update direction vector is determined by the first derivative of the optimization function of the service quality parameters of each professional model at the current discrete coordinate point. When the angle between the local update direction vector of any professional model and the global update direction vector within the group exceeds a preset angle threshold, the current professional model is marked as a direction abnormal model, and the weight of the current professional model in the vector weighted average calculation is reduced in subsequent iterations.

[0032] As an example, the recommended preset angle threshold is 30° to 60°, with 45° being the optimal value.

[0033] The rationale for setting the value to 30° (lower limit) is as follows: When the angle between two vectors is less than 30°, the cosine value is greater than 0.866, and the overlap of their projections in space exceeds 86.6%. This difference can usually be attributed to random noise or individual preferences and should not be considered anomaly. If the threshold is lower than 30°, it will cause a large number of normal models to be misjudged as anomalies, reducing the effective participation rate of the consensus protocol.

[0034] The reason for setting the value to 60° (the upper limit) is that when the included angle exceeds 60°, the cosine value is less than 0.5, and the directional consistency between the two is less than 50%. At this point, the difference between the local and global directions of the model is significantly greater than the degree of consistency, and continuing to include it in the weighted average calculation will severely distort the global direction. If the threshold is higher than 60°, the negative impact of the abnormal model will spread to the entire model group.

[0035] The reason for choosing 45° is that the cosine value of 45° is 0.707, which means that the overlap of the projections of the two vectors on the unit circle is 70.7%. 45° can provide an engineering balance point. An angle less than 45° is considered an acceptable deviation, while an angle greater than 45° is considered an anomaly that requires intervention. Therefore, the optimal angle value has both vector consistency and universality.

[0036] As an example, when reducing the weight of the current professional model in the vector weighted average calculation in subsequent iterations, the orientation anomaly model after reducing the weight moves by a perturbation step along the vertical direction of the global update orientation vector within the group, and the perturbation step is proportional to the number of times the orientation anomaly model is continuously labeled.

[0037] Driving force of the orientation anomaly model: The directional anomaly model is driven by the arbitration logic in the distributed consensus protocol. The arbitration logic can be encapsulated into a software module and run on the computing nodes deployed by each professional model. It is usually executed by the initiator of the global update of the direction vector within the group (which can be the master node generated by the election mechanism in the group of models to be synchronized) or an independent coordination service (such as the corresponding functional component of the distributed coordinator ZooKeeper).

[0038] The specific control logic of the driving method is as follows: Once a particular model is marked as an abnormal orientation model, its behavior in the current iteration round will be executed according to specific logic. The following pseudocode example illustrates this: if (angle > preset angle threshold): Step 1: Reduce weight weight = original weight * decay factor (e.g., 0.3) Step 2: Not participating in the weighted average calculation of the global direction vector Remove the model from the list of participants in the weighted average calculation. Step 3: Calculate the vertical movement vector v_projection = the component of the local direction vector projected onto the global direction. v_perpendicular = local direction vector - v_projection Step 4: Calculate the perturbation step size step_perp = baseline step size * (1 + number of consecutive marks * step size growth factor, e.g., 0.2) Step 5: Perform vertical movement New coordinates = Current coordinates + normalize(v_perpendicular) * step_perp Not involved in global direction calculation: The local update direction vector given by the model is not input into the vector weighted average formula, and the global update direction vector within the group is calculated only by the other normal models in the group.

[0039] Determining the vertical direction: Project the local update direction vector onto the global direction, and subtract the projected vector from the original vector to obtain the component perpendicular to the global direction.

[0040] Calculation of perturbation step size: The base step size is usually taken as the unit step size of the current iteration (i.e. the step size used for normal model synchronous movement), multiplied by a coefficient that is proportional to the number of consecutive markings (e.g., 1+0.2×n, where n is the number of consecutive rounds marked as abnormal), so that the deviation amplitude gradually increases as the abnormality continues, avoiding the model from repeatedly oscillating in the abnormal state and being unable to get out.

[0041] The driving entity: The driving instructions are sent from the master node or coordination service of the distributed consensus protocol to the computing node where the directional anomaly model is located via RPC (Remote Procedure Call) or message queue. After receiving the instructions, the computing node performs the coordinate point update operation.

[0042] Preferably, in step S2, after driving each professional model to synchronously move and update the discrete coordinate points, the historical unique identifier stored in the distributed hash table is compared and verified with the unique identifier regenerated for the current discrete coordinate point; if the verification is inconsistent, a backtracking mechanism is triggered to restore the discrete coordinate points of all professional models in the model group to be synchronized to the coordinate points that passed the last verification, and a backtracking event is recorded.

[0043] As an example, after the backtracking mechanism is triggered, when the number of consecutive backtracking events reaches a preset backtracking threshold, a local perturbation operation is performed. The local perturbation operation specifically involves generating a random perturbation vector for each professional model in the model group to be synchronized, allowing the professional models to move independently along their respective random perturbation vectors, and then re-executing the conflict association pair marking step.

[0044] In the preferred technical solution of step S2, by specifically limiting the graph decomposition operation to connected component decomposition and employing a depth-first search algorithm, efficient traversal and partitioning of the conflict-related graph are achieved in step S2. The preferred solution of step S2 works synergistically with the distributed hash table storage structure constructed in step S1: firstly, the required node access states can be marked using key-value pairs in the hash table, avoiding the memory overhead of maintaining an additional access array. Simultaneously, the model group to be synchronized output by connected component decomposition serves as the operation object of the distributed consensus protocol in step S3, limiting the scope of subsequent synchronization movements to model groups with transitive conflict relationships, rather than the entire model set, thereby reducing ineffective consensus communication overhead.

[0045] In the preferred scheme for step size control and local update direction vector in step S2, the initial value of the unit step size is set to one-tenth of the minimum Euclidean distance, and it decays to 90% of the current value after three consecutive failures to identify new models to be synchronized. This preferred scheme is linked with the iterative loop formed in step S3: when the size of the model group to be synchronized no longer changes, the decay step size allows for more refined subsequent coordinate point updates, avoiding crossing the optimal convergence region due to excessively large step sizes. Simultaneously, the determination of abnormal direction models is introduced—when the angle between the local update direction vector and the global direction vector exceeds a preset angle threshold, the model's weighting is reduced, solving the problem of a few models within the group interfering with the overall convergence direction. The marked abnormal direction models move along the direction perpendicular to the global direction, and the movement step size is proportional to the number of consecutive markings. It can be seen that through the optimization and improvement of step S2, while maintaining a certain exploration capability, the deviation amplitude is gradually increased to seek escape from the local conflict state.

[0046] In the example scheme related to verification and perturbation in step S2, the historical unique identifier in the distributed hash table is compared with the regenerated identifier of the current coordinate point to ensure data consistency for coordinate updates in step S3. When the verification is inconsistent, a backtracking mechanism is triggered to restore the entire set of coordinates to the state of the last successful verification, avoiding the propagation of data inconsistency caused by concurrent modifications. After continuous backtracking reaches a threshold, synchronous movement is paused and random perturbation is performed instead, complementing the step size decay mechanism—step size decay is suitable for the fine-grained convergence stage, while random perturbation is suitable for the deadlock escape stage. The two play their respective roles in different convergence states. By monitoring the trend of the size change of the model group to be synchronized and dynamically adjusting the step size and threshold when the size change continues to expand, the algorithm can adaptively distinguish between "size fluctuations" and "uncontrolled conflict propagation" in the normal convergence process, avoiding the problem of unstable performance of fixed parameters on different scale problems.

[0047] Preferably, during the formation of the iterative loop in step S3, the trend of the size change of the model group to be synchronized is monitored. When the size of the model group to be synchronized continues to expand during the iteration, it is determined to be a conflict diffusion state. The unit step size is adjusted to half of the current step size, and the preset difference threshold is adjusted to 1.5 times the current threshold.

[0048] In the preferred technical solution of step S3, the state discrimination capability of the iterative process is established by monitoring the trend of the size change of the model group to be synchronized. During the iterative loop of step S3, the size of the model group to be synchronized does not change monotonically—after each professional model moves synchronously along the global update direction vector, models that were originally in conflict may escape the conflict state, or new conflict associations may be triggered due to parameter adjustments, leading to an expansion of the model group size. The preferred technical solution of step S3 distinguishes the iterative process into a normal convergence state and a conflict diffusion state by judging whether the size continues to expand, providing a state basis for subsequent parameter adjustments. It does not require the introduction of additional external prior knowledge and is entirely based on the data of the model group to be synchronized generated during the iteration process, achieving self-supervised state awareness.

[0049] like Figure 2 The data processing system shown is for online collaboration and conflict detection of multi-disciplinary models, used to implement the data processing method for online collaboration and conflict detection of multi-disciplinary models described in any specific embodiment of this application, including: The unique identifier and model parameter sequence generation module reads the initial model parameter sequence of each professional model, constructs a multi-dimensional parameter space based on the initial model parameter sequence, maps each professional model to discrete coordinate points in the multi-dimensional parameter space, the coordinate values ​​are composed of the current values ​​of the service quality parameters in the professional model, generates a unique identifier for each discrete coordinate point using a hash function, and stores the unique identifier and the corresponding model parameter sequence in a distributed hash table in the form of key-value pairs; The module for identifying the model group to be synchronized calculates the normalized Euclidean distance between any two discrete coordinate points in the multidimensional parameter space to obtain the parameter difference degree between each professional model; when the parameter difference degree is lower than a preset difference threshold, the corresponding two professional models are marked as conflict association pairs; a conflict association graph is constructed based on all conflict association pairs, and a graph decomposition operation is performed on the conflict association graph to identify professional models located in the same decomposition subgraph as the model group to be synchronized. The parameter difference and threshold comparison module obtains the local update direction vector of each professional model in the model group to be synchronized, exchanges the local update direction vectors in the model group to be synchronized through a distributed consensus protocol, and calculates the global update direction vector in the group. It then drives each professional model in the model group to be synchronized to move synchronously by one unit step along the global update direction vector to update its own discrete coordinate points. Based on the updated discrete coordinate points, it re-executes the conflict association pair marking and the identification of the model group to be synchronized, forming an iterative loop until there are no conflict association pairs in the multidimensional parameter space with parameter difference lower than the preset difference threshold.

[0050] The technical solutions provided in steps S1 to S3 are described below through implementation methods in specific application scenarios based on embodiments of this application: An industrial design team needs to perform co-simulation of turbine blades, involving three specialized models: a structural analysis model (M1) to calculate the stress distribution of the blade under aerodynamic loads, a fluid simulation model (M2) to simulate the flow of high-temperature combustion gases on the blade surface, and a thermodynamic model (M3) to calculate the temperature field distribution inside the blade. There are parameter couplings between the three models: the mesh density parameter (p1) of M1 affects the boundary layer mesh generation of M2; the time step parameter (p2) of M2 is constrained by the thermal diffusion time constant of M3; and the material thermal conductivity parameter (p3) of M3 interacts with the stress calculation accuracy of M1. Different initial parameter values ​​for each model lead to collaboration conflicts. All service quality parameters have been normalized, with values ​​ranging from [0,1].

[0051] Step S1, Parameter space construction and hash storage: Each specialized model registers with the collaboration service via a RESTful interface, submitting its initial sequence of model parameters. M1 submits {p1=0.85}, M2 submits {p2=0.60}, and M3 submits {p3=0.90}. The collaboration service constructs a three-dimensional parameter space based on the total number of parameters M=3 and the number of specialized models N=3. In this three-dimensional space, each specialized model is mapped to a complete discrete coordinate point: for service quality parameters not involved in that model, the value is taken as the global default value of 0.5. Therefore, the coordinates of M1 are (0.85, 0.50, 0.50), the coordinates of M2 are (0.50, 0.60, 0.50), and the coordinates of M3 are (0.50, 0.50, 0.90).

[0052] Perform a SHA-256 hash operation on each discrete coordinate point. Taking the coordinate point (0.85, 0.50, 0.50) of M1 as an example, serialize it into the string "0.85, 0.50, 0.50", input it into the SHA-256 function to obtain the unique identifier "3f5b2c8e9a1d4f7b6c2e8a9d4f7b6c2e". Store this identifier and the model parameter sequence of M1 in a distributed hash table (using a consistent hashing ring, with 150 virtual nodes) as key-value pairs, with the identifier as the key and {p1=0.85} as the value. This storage method allows subsequent steps to quickly retrieve and verify whether the parameter sequence has been tampered with using the identifier.

[0053] Step S2, Conflict Detection and Identification of Model Groups to be Synchronized: The collaborative service calculates the normalized Euclidean distance between any two discrete coordinate points in a three-dimensional parameter space. Since all dimensions of the parameter space have the same dimensions (and are all normalized), the Euclidean distance formula is used directly: d(M1,M2) = sqrt[(0.85-0.50) 2 + (0.50-0.60) 2 + (0.50-0.50) 2 ] = sqrt(0.1225 + 0.01 + 0) = sqrt(0.1325) = 0.364 d(M1,M3) = sqrt[(0.85-0.50) 2 + (0.50-0.50) 2 + (0.50-0.90) 2 ] = sqrt(0.1225 + 0 + 0.16) = sqrt(0.2825) = 0.531 d(M2,M3) = sqrt[(0.50-0.50) 2 + (0.60-0.50) 2 + (0.50-0.90) 2 ] = sqrt(0 +0.01 + 0.16) = sqrt(0.17) = 0.412 The preset difference threshold ε is set to 0.45. Comparison results: d(M1,M2)=0.364 < 0.45, (M1,M2) is labeled as a conflicting association pair; d(M1,M3)=0.531 > 0.45, no labeling; d(M2,M3)=0.412 < 0.45, (M2,M3) is labeled as a conflicting association pair.

[0054] A conflict association graph is constructed based on conflict association pairs: nodes are {M1, M2, M3}, and edges are (M1-M2) and (M2-M3). Connected component decomposition is performed: starting from M1, a depth-first search is performed, visiting M2, and then M2 visits M3, resulting in the connected component {M1, M2, M3}. This component represents the group of models to be synchronized, indicating that there is a transitive conflict among the three models—M1 conflicts with M2, and M2 conflicts with M3. Although the direct distance between M1 and M3 does not exceed a threshold, they form an indirect conflict relationship through M2, requiring overall coordinated adjustment.

[0055] Step S3, Coordinated Adjustment and Iterative Convergence: Obtain the local update direction vector for each specialty model within the model group to be synchronized. The optimization function for each specialty model is defined as a quadratic form considering parameter coupling, and the function is differentiable everywhere in its domain.

[0056] The optimization function of M1 (structural analysis model) is defined as: f1(p1,p2,p3) = -[ (p1-0.85) 2 + 0.3·(p2-0.60) 2 + 0.2·(p3-0.90) 2 ] The coefficients 0.3 and 0.2 represent the degree of influence of p2 and p3 on target M1. Taking the partial derivatives with respect to p1, p2, and p3: ∂f1 / ∂p1 = -2(p1-0.85) ∂f1 / ∂p2 = -0.6(p2-0.60) ∂f1 / ∂p3 = -0.4(p3-0.90) Calculate the gradient vector at the initial coordinate point (0.85, 0.50, 0.50): ∂f1 / ∂p1 = -2(0.85-0.85)=0 ∂f1 / ∂p2 = -0.6(0.50-0.60)=0.06 ∂f1 / ∂p3 = -0.4(0.50-0.90)=0.16 The gradient vector is (0, 0.06, 0.16), which after normalization is approximately (0, 0.35, 0.94). The optimization function of M2 (fluid simulation model) is defined as: f2(p1,p2,p3) = -[ 0.2·(p1-0.85) 2 + (p2-0.60) 2 + 0.3·(p3-0.90) 2 ] Calculate the gradient at the initial coordinate point (0.50, 0.60, 0.50): ∂f2 / ∂p1 = -0.4(0.50-0.85)=0.14 ∂f² / ∂p² = -2(0.60-0.60)=0 ∂f² / ∂p³ = -0.6(0.50-0.90)=0.24 The gradient vector is (0.14, 0, 0.24), which is approximately (0.50, 0, 0.86) after normalization. The optimization function of M3 (thermodynamic model) is defined as: f3(p1,p2,p3) = -[ 0.1·(p1-0.85) 2 + 0.1·(p2-0.60) 2 + (p3-0.90) 2 ] Calculate the gradient at the initial coordinate point (0.50, 0.50, 0.90): ∂f3 / ∂p1 = -0.2(0.50-0.85)=0.07 ∂f³ / ∂p² = -0.2(0.50-0.60)=0.02 ∂f3 / ∂p3 = -2(0.90-0.90)=0 The gradient vector is (0.07, 0.02, 0), which after normalization is approximately (0.96, 0.27, 0). The local update direction vectors are exchanged within the model group to be synchronized using a distributed consensus protocol (employing the Raft algorithm, with M1 elected as the master node). After the master node collects the three vectors, it performs a weighted average, with each vector initially weighted at 1 / 3. V_global = (1 / 3)×(0,0.35,0.94) + (1 / 3)×(0.50,0,0.86) + (1 / 3)×(0.96,0.27,0) = (0.487, 0.207, 0.600) After normalization, V_global ≈ (0.61, 0.26, 0.75) Calculate the angle between each local direction and the global direction. Taking M1 as an example: the dot product is (0,0.35,0.94)·(0.61,0.26,0.75)=0×0.61+0.35×0.26+0.94×0.75=0.091+0.705=0.796, the modulus is 1, the cosine value is 0.796, and the angle θ=arccos(0.796)≈37.2°. In this embodiment, the preset angle threshold is set to 45°. 37.2° does not exceed the threshold, therefore M1 is not marked as an orientation anomaly model.

[0057] Calculate the angle between M2 and the global direction: Dot product = (0.50, 0, 0.86) · (0.61, 0.26, 0.75) = 0.305 + 0 + 0.645 = 0.950, angle ≈ 18.2°, not exceeding the threshold. Calculate the angle between M3 and the global direction: Dot product = (0.96, 0.27, 0) · (0.61, 0.26, 0.75) = 0.586 + 0.070 + 0 = 0.656, angle ≈ 49.0°, exceeding the 45° threshold, therefore M3 is marked as an orientation anomaly model.

[0058] According to the preferred scheme, the directional anomaly model M3 does not participate in the global direction vector calculation in this round. Normal models M1 and M2 recalculate the global direction: V_global_new = (1 / 2)×(0,0.35,0.94) + (1 / 2)×(0.50,0,0.86) =(0.25, 0.175, 0.90), normalized to (0.26, 0.18, 0.95). The vertical movement vector for M3 is calculated: first, the local direction (0.96,0.27,0) of M3 is projected onto the global direction (0.26,0.18,0.95). The dot product u·v = 0.96×0.26 + 0.27×0.18 + 0×0.95 = 0.2496 + 0.0486 = 0.2982 v·v = 0.26 2 +0.18 2 +0.95 2 = 0.0676 + 0.0324 + 0.9025 = 1.0025 ≈ 1 proj_v(u) = 0.2982 × (0.26,0.18,0.95) = (0.0775, 0.0537, 0.2833) Vertical component = u - proj_v(u) = (0.96-0.0775, 0.27-0.0537, 0-0.2833) =(0.8825, 0.2163, -0.2833) After normalization, the vertical direction is approximately (0.94, 0.23, -0.30). The initial unit step size is set to one-tenth of the minimum Euclidean distance between any two discrete coordinate points in the parameter space. The initial minimum Euclidean distance in the parameter space is d(M1,M2)=0.364, and one-tenth of it is 0.0364. The perturbation step size is proportional to the number of times the orientation anomaly model is continuously marked. Assuming a proportionality coefficient of 0.2, and M3 being marked for the first time (number of consecutive markings = 1), the perturbation step size is 0.0364×(1+0.2×1)=0.0437. After M3 moves vertically, its coordinates are updated to: original coordinates (0.50,0.50,0.90) + 0.0437×(0.94,0.23,-0.30) = (0.541, 0.510, 0.887). M1 and M2 move synchronously along the global direction (0.26, 0.18, 0.95) by a unit step of 0.0364: M1's new coordinates = (0.85, 0.50, 0.50) + 0.0364 × (0.26, 0.18, 0.95) = (0.859, 0.507, 0.535); M2's new coordinates = (0.50, 0.60, 0.50) + 0.0364 × (0.26, 0.18, 0.95) = (0.509, 0.607, 0.535). After the coordinates are updated, a distributed hash table is used for verification: the SHA-256 identifier of each coordinate point is recalculated and compared with the historical identifier stored in the hash table. If the local parameters of M1 are modified concurrently by other processes, causing inconsistency, backtracking is triggered—the coordinates of the entire group of M1, M2, and M3 are restored to the coordinates that passed the last verification, and the backtracking event is recorded. If backtracking occurs 3 times consecutively (preset backtracking threshold), synchronous movement is paused, and random perturbation vectors are generated for each model in each group (each component is evenly distributed in the interval [-0.02, 0.02]), and collision detection is re-executed after independent movement.

[0059] Based on the updated coordinates, step S2 is re-executed to identify the conflict association markers and the model group to be synchronized, forming an iterative loop. The size changes of the model group to be synchronized are monitored: if the three models remain in the same connected component and their size does not decrease after multiple iterations, it is determined to be in a conflict diffusion state. At this time, the unit step size is adjusted to half of the current size (0.0182), and the preset difference threshold is adjusted to 1.5 times the current size (0.675) to make the convergence process smoother.

[0060] The iteration continues, and the parameter differences between all pairs of models are recalculated after each update. When the gradient magnitude of the optimization function of M2 drops below 0.01 (set as the convergence threshold in this embodiment), the coordinates of M2 are basically stable. At this point, the parameter differences between each model are calculated. If all three are greater than the current threshold, there are no conflicting pairs, the conflict graph is empty, and the iteration terminates. If some differences are still below the threshold, but the parameter changes of all models in five consecutive iterations are less than 10, the iteration terminates. -6 If the iteration terminates, the remaining conflict pairs are marked as tolerable conflicts (i.e., minor differences that are acceptable in engineering). The final output parameters are configured as M1(0.67,0.52,0.50), M2(0.50,0.52,0.50), and M3(0.50,0.52,0.71), and the cooperative conflict resolution is complete.

[0061] like Figure 3 As shown, this application embodiment, in addition to providing a data processing method and system for online collaboration and conflict detection of multi-disciplinary models, also provides a corresponding electronic device: An electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described in any specific embodiment of the present application.

[0062] Figure 3 This is a schematic diagram of a computer device provided in an embodiment of this application. The device 800 includes a processor 801, a memory 802, a communication interface 803, and a bus 804. The processor 801, memory 802, and communication interface 803 communicate via the bus 804, or via other means such as wireless transmission. The memory 802 stores instructions, and the processor 801 executes the instructions stored in the memory 802. The memory 802 stores program code 8021, and the processor 801 can call the program code 8021 stored in the memory 802 to execute the steps of a data processing method for online collaboration and conflict detection of multi-disciplinary models.

[0063] It should be understood that in the embodiments of this application, processor 801 may be a CPU, or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. General-purpose processors may be microprocessors or any conventional processors, etc.

[0064] The memory 802 may include read-only memory (ROM) and random access memory (RAM), and provides instructions and data to the processor 801. The memory 802 may also include non-volatile random access memory. The memory 802 may be volatile memory, non-volatile memory, or both. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM).

[0065] In addition to the data bus, bus 804 may also include a power bus, a control bus, and a status signal bus. However, for clarity, all buses are labeled as bus 804 in the diagram.

[0066] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive (SSD).

[0067] In the description of the embodiments of this application, the reference to terms such as "an embodiment," "example," "specific example," etc., means that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the embodiments of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0068] Furthermore, the technical solutions of the various implementation methods in this application can be combined with each other, but only if they are based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the embodiments of this application.

[0069] All features disclosed in the embodiments of this application, or all steps in the disclosed methods or processes, may be combined in any way, except for mutually exclusive features and / or steps. Any feature disclosed in the specification of the embodiments of this application, unless specifically stated otherwise, may be replaced by other equivalent or similar alternative features. That is, unless specifically stated otherwise, each feature is merely one example of a series of equivalent or similar features. Throughout the specification, the same reference numerals indicate the same elements.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of this application, and are not intended to limit them. Although the embodiments of this application have been described in detail with reference to the foregoing specific embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein, and such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the specific embodiments of this application.

Claims

1. A data processing method for multi-specialty model online collaboration and conflict detection, characterized in that, include: Read the initial model parameter sequence of each professional model, construct a multi-dimensional parameter space based on the initial model parameter sequence, map each professional model to discrete coordinate points in the multi-dimensional parameter space, the coordinate values ​​are composed of the current values ​​of the service quality parameters in the professional model, generate a unique identifier for each discrete coordinate point using a hash function, and store the unique identifier and the corresponding model parameter sequence in a distributed hash table in the form of key-value pairs; Calculate the normalized Euclidean distance between any two discrete coordinate points in the multidimensional parameter space to obtain the parameter difference degree between each professional model; when the parameter difference degree is lower than a preset difference threshold, mark the corresponding two professional models as conflict association pairs; construct a conflict association graph based on all conflict association pairs, perform graph decomposition operation on the conflict association graph, and identify professional models located in the same decomposition subgraph as a group of models to be synchronized. The local update direction vector of each professional model in the model group to be synchronized is obtained. The local update direction vectors are exchanged in the model group to be synchronized through a distributed consensus protocol, and the global update direction vector in the group is calculated. The model group to be synchronized is driven to move one unit step along the global update direction vector to update its discrete coordinate points. The conflict association pair marking and identification with the model group to be synchronized are re-executed based on the updated discrete coordinate points to form an iterative loop until there are no conflict association pairs in the multidimensional parameter space with parameter differences lower than a preset difference threshold.

2. The data processing method for multi-specialty model online collaboration and conflict detection according to claim 1, characterized in that, When performing a graph decomposition operation on the conflict association graph, the graph decomposition operation is specifically a connected component decomposition; the connected component decomposition uses a depth-first search algorithm to traverse all nodes of the conflict association graph and divides nodes that are connected by paths into the same connected component.

3. The data processing method for multi-specialty model online collaboration and conflict detection according to claim 1, characterized in that, The initial value of the unit step size is set to one-tenth of the minimum Euclidean distance between any two discrete coordinate points in the multidimensional parameter space. In subsequent iterations, if no new model group to be synchronized is identified in three consecutive iterations, the unit step size is adjusted to 90% of the current value.

4. The data processing method for multi-specialty model online collaboration and conflict detection according to claim 1, characterized in that, The local update direction vector is determined by the first derivative of the optimization function of the service quality parameters of each professional model at the current discrete coordinate point. When the angle between the local update direction vector of any professional model and the global update direction vector within the group exceeds a preset angle threshold, the current professional model is marked as an abnormal direction model, and the weight of the current professional model in the vector weighted average calculation is reduced in subsequent iterations.

5. The data processing method for online collaboration and conflict detection of multi-disciplinary models according to claim 4, characterized in that, The orientation anomaly model, after weight reduction, moves along the vertical direction of the global update orientation vector within the group by a perturbation step, the perturbation step being proportional to the number of times the orientation anomaly model is continuously labeled.

6. The data processing method for multi-specialty model online collaboration and conflict detection according to claim 1, characterized in that, After driving the various professional models to synchronously move and update the discrete coordinate points, the historical unique identifier stored in the distributed hash table is compared and verified with the regenerated unique identifier of the current discrete coordinate point. If the verification fails, a backtracking mechanism is triggered to restore the discrete coordinate points of all professional models in the model group to be synchronized to the coordinate points that passed the last verification, and a backtracking event is recorded.

7. The data processing method for multi-specialty model online collaboration and conflict detection according to claim 6, characterized in that, When the number of consecutive backtracking events reaches a preset backtracking threshold, a local perturbation operation is performed. Specifically, the local perturbation operation is as follows: a random perturbation vector is generated for each professional model in the model group to be synchronized, the professional model moves independently along its own random perturbation vector, and then the conflict association pair marking step is re-executed.

8. The data processing method for multi-specialty model online collaboration and conflict detection according to claim 1, characterized in that, During the formation of the iterative cycle, the trend of the size change of the model group to be synchronized is monitored. When the size of the model group to be synchronized continues to expand during the iteration, it is determined to be a conflict diffusion state. The unit step size is adjusted to half of the current step size, and the preset difference threshold is adjusted to 1.5 times the current threshold.

9. A data processing system for multi-specialty model online collaboration and conflict detection, configured to implement the data processing method for multi-specialty model online collaboration and conflict detection according to any one of claims 1 to 8, characterized in that, include: The unique identifier and model parameter sequence generation module reads the initial model parameter sequence of each professional model, constructs a multi-dimensional parameter space based on the initial model parameter sequence, maps each professional model to discrete coordinate points in the multi-dimensional parameter space, the coordinate values ​​are composed of the current values ​​of the service quality parameters in the professional model, generates a unique identifier for each discrete coordinate point using a hash function, and stores the unique identifier and the corresponding model parameter sequence in a distributed hash table in the form of key-value pairs; The module for identifying the model group to be synchronized calculates the normalized Euclidean distance between any two discrete coordinate points in the multidimensional parameter space to obtain the parameter difference degree between each professional model; when the parameter difference degree is lower than a preset difference threshold, the corresponding two professional models are marked as conflict association pairs; a conflict association graph is constructed based on all conflict association pairs, and a graph decomposition operation is performed on the conflict association graph to identify professional models located in the same decomposition subgraph as the model group to be synchronized. The parameter difference and threshold comparison module obtains the local update direction vector of each professional model in the model group to be synchronized, exchanges the local update direction vectors in the model group to be synchronized through a distributed consensus protocol, and calculates the global update direction vector in the group. It then drives each professional model in the model group to be synchronized to move synchronously by one unit step along the global update direction vector to update its own discrete coordinate points. Based on the updated discrete coordinate points, it re-executes the conflict association pair marking and the identification of the model group to be synchronized, forming an iterative loop until there are no conflict association pairs in the multidimensional parameter space with parameter difference lower than the preset difference threshold.

10. An electronic device, comprising: include: The system includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; the memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the method according to any one of claims 1 to 8.

Images