Three-dimensional modeling data management method and system and storage medium

Abstract

The invention relates to a three-dimensional modeling data management method and system and a storage medium, and belongs to the technical field of data management. The method comprises the following steps: inputting original construction data and environment attribute data into a data fusion algorithm, and generating fusion data of each sub-module of three-dimensional modeling according to data categories; according to the operation data, the periodic access frequency of the user to the three-dimensional modeling sub-modules is obtained, and the access priority of each sub-module in comprehensive three-dimensional modeling is judged; judging the storage priority of each sub-module in the comprehensive three-dimensional modeling in combination with the current access frequency of the sub-module; and generating a migration decision of the sub-module through a non-cooperative game model according to the storage priority. According to the method, the access complexity is reduced through fusion reforming of the three-dimensional modeling sub-modules, and accurate resource allocation and real-time response optimization are realized in combination with a dynamic priority decision; a priority layering and high-speed caching mechanism is adopted, and user experience and hardware resource consumption are balanced through a dynamic strategy.

Description

Technical Field

[0001] The present invention belongs to the technical field of data management, and particularly relates to a three-dimensional modeling data management method, system and storage medium. Background Art

[0002] With the wide application of three-dimensional modeling technology, the demand for data management in fields such as engineering, urban management, and intelligent manufacturing is increasing day by day.

[0003] However, the existing three-dimensional modeling data management still has deficiencies in aspects such as data access, storage efficiency, and real-time update. For example, the storage area is divided based on fixed rules (such as data size, type), lacking dynamic response to the real-time access needs of users, resulting in the inability to quickly call frequently accessed data, while low-frequency data occupies high-performance storage resources for a long time. Using a fixed time interval or a simple threshold to trigger data update is difficult to balance the real-time requirement and the consumption of computing resources, and is prone to system overload or data update lag. Therefore, the existing technology is difficult to meet the efficient and accurate data management requirements in complex scenarios, and there are problems such as rigid resource allocation and imbalance between real-time performance and resource efficiency. Summary of the Invention

[0004] To solve the above problems existing in the prior art, the present invention provides a three-dimensional modeling data management method, system and storage medium, which reduces the access complexity through the fusion and reorganization of three-dimensional modeling sub-modules, and realizes precise resource allocation and real-time response optimization by combining dynamic priority decision-making; adopts a priority hierarchy and cache mechanism, and balances the user experience and hardware resource consumption through dynamic strategies.

[0005] The object of the present invention can be achieved by the following technical solutions: The first aspect of the present disclosure provides a three-dimensional modeling data management method, including the steps of: Data collection: Obtain the original construction data of three-dimensional modeling, the environmental attribute data within the target area, and the operation data generated by the user interaction platform; Module reorganization: Input the original construction data and environmental attribute data into a data fusion algorithm, and generate the fusion data of each sub-module of three-dimensional modeling according to the data category; Priority decision-making: Obtain the periodic access frequency of the user to the three-dimensional modeling sub-modules according to the operation data, and determine the access priority of each sub-module in the comprehensive three-dimensional modeling; then combine the current access frequency of the sub-module to determine the storage priority of each sub-module in the comprehensive three-dimensional modeling; Data migration: Divide the storage area of the system into a cache area and a low-speed cache area, and generate the migration decision of the sub-module through a non-cooperative game model according to the storage priority; Module update decision-making: Combine the access priority and storage priority of each sub-module, and dynamically adjust the update frequency through a non-linear weighted model; Parameter Tuning: Based on historical operation data, a reinforcement learning model is constructed to achieve parameter adaptive tuning for dynamically adjusting the update frequency.

[0006] Further, the original construction data includes geometric structure data, texture mapping data, and material property data; the environmental property data includes light intensity, temperature distribution, and humidity distribution; the operation data includes data update frequency settings and user call instructions.

[0007] Further, the calculation formula of the data fusion algorithm is: ; In the formula, represents the weighted sum of geometric structure, texture mapping, material property, light intensity, temperature distribution, and humidity distribution, represents the type of geometric structure data, represents the type of texture mapping data, represents the type of material property data, represents the light intensity, represents the temperature distribution, represents the humidity distribution, , , respectively represent the weight coefficients of the type of geometric structure data, the type of texture mapping data, and the type of material property data, , , respectively represent the weights of light intensity, temperature distribution, and humidity distribution.

[0008] Further, the priority decision includes the following steps: Calculate the access heat value of each sub-module in 3D modeling according to the obtained periodic access frequency; Combine the data update frequency settings in the operation data to construct a data update frequency weight coefficient, and perform weighted processing on the access heat value of each sub-module to obtain the access priority of each sub-module; Calculate the storage priority of each sub-module according to the access priority and current access frequency of the sub-module.

[0009] Further, the calculation formula of the access heat value is: ; In the formula, represents the access heat value of the th sub-module, represents the number of times the user accesses the sub-module within the time period, represents the total number of time periods; The calculation formula for the access priority is: ; where, represents the access priority of the th sub-module, represents the data update frequency weight corresponding to the th sub-module; The calculation formula for the storage priority is: ; where, represents the storage priority of the th sub-module, represents the current access frequency, represents the weight coefficient.

[0010] Furthermore, the generation of the migration decision includes the steps of: Regarding the set of sub-modules to be migrated as the set of participants, and regarding the storage priority, access latency, storage cost, and migration overhead of each sub-module as the information set; Regarding each sub-module as an independent decision-making entity, and maximizing its own benefit by balancing the access latency, storage cost, and migration overhead of each sub-module; Solving the Nash equilibrium strategy combination, and determining the final migration decision of each sub-module according to the Nash equilibrium strategy combination; The solving of the Nash equilibrium strategy combination includes the steps of: Setting the current system status as the root node, and setting the number of simulations as N; Starting from the root node, recursively generating child nodes, balancing exploration and exploitation through the UCB formula, randomly performing a complete migration of the sub-module from the current node, and performing the migration operation according to the strategy combination, and calculating the benefit of each sub-module and the total system benefit; Backpropagating the simulation result to the parent node, and updating the average benefit and access times of the node; After N iterations, outputting the strategy combination with the most access times as the Nash equilibrium strategy combination; where, the UCB formula is: ; In the formula, is the average benefit, is the exploration times.

[0011] Furthermore, the calculation formula for the non-linear weighted model is: In the formula, is the update frequency of the th sub-module, is the balance coefficient, and , and are the sensitivity parameters of the access priority and the sensitivity parameter of the storage priority respectively, and and are both greater than 0; If , the influence of the access priority on the update frequency shows exponential growth; if , the influence of the access priority on the update frequency tends to level off; When the storage location of the sub-module is in the cache area, is set to 2; when the storage location of the sub-module is in the low-speed cache area, is set to 0.5; At the same time, a real-time update threshold is set for the update frequency of the sub-module. If the update frequency of the sub-module is greater than the real-time update threshold, an active synchronization strategy is implemented.

[0012] Furthermore, the parameter tuning includes the following steps: Define the state space, action space, and reward function of reinforcement learning according to the access priority and storage priority; the state space includes the access heat value, the current access frequency, and the data update frequency weight, expressed as ; the state space depicts the dynamic characteristics of the sub-module through multi-dimensional information and provides a decision-making basis for the reinforcement learning model; The action space includes the balance coefficient and the sensitivity parameter of the access priority, expressed as ; the action space defines the range of policy adjustments that the system can execute; The calculation formula of the reward function is: In the formula, is the cache hit rate, is the data freshness, is the consumption of computing resources, , and are the target weight coefficients.

[0013] The second aspect of the present disclosure provides a three-dimensional modeling data management system that executes a three-dimensional modeling data management method as described above, including a data collection module, a data reorganization module, a priority management module, a storage scheduling module, an update control module, and a parameter optimization module; The data collection module is used to obtain original construction data, environmental attribute data, and user operation data from different sources and implement standardized preprocessing of the data; The data reorganization module is used to integrate the original construction data and the environmental attribute data through a data fusion algorithm to generate sub-module fusion data for 3D modeling; The priority management module dynamically calculates the access priority of the sub-modules based on the user operation data, and at the same time combines the current access frequency and the access priority to generate a storage priority ranking of the sub-modules; The storage scheduling module is used to divide the storage area into a cache area and a low-speed cache area, manage the capacity and performance balance of the two types of storage, and optimize the data migration strategy through a non-cooperative game model based on the storage priority; The update control module is used to dynamically adjust the update frequency of the sub-modules by using a non-linear weighting model to balance real-time performance and resource efficiency; The parameter optimization module constructs a reinforcement learning model based on historical operation data, dynamically tunes the update frequency and the storage migration threshold, and realizes the adaptive optimization of the system.

[0014] The third aspect of the present disclosure provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method are run.

[0015] The beneficial effects of the present invention are as follows: The present invention reorganizes the sub-modules of the 3D modeling data, integrates multi-source heterogeneous data according to the weight setting to complete the integration of sub-modules of different data types, avoids the limitations caused by single data to the modeling access, and at the same time reduces the access frequency and interleaving complexity of each sub-module from the bottom layer through the fusion of sub-modules, providing clear operation data for priority decision-making; on the basis of doing necessary sub-module integration, makes access priority and storage priority decisions for the 3D modeling sub-modules according to the operation data, realizes precise resource allocation and real-time response optimization through dynamic priority management, and conducts global efficiency optimization through game-driven storage migration based on storage priority; then dynamically adjusts the update frequency according to the priority through a non-linear weighting model, achieving a balance between resource efficiency and real-time performance; finally, introduces a reinforcement learning model to realize adaptive parameter tuning, further optimizing the problem of reduced cache hit rate caused by priority decision-making. The present invention ensures the rapid response of key data through priority stratification and cache mechanism, and at the same time is compatible with large-scale data storage; the dynamic update strategy and game model find the optimal solution between user interaction experience and hardware resource consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] For the convenience of those skilled in the art to understand, the present invention will be further described below with reference to the accompanying drawings.

[0017] Figure 1 Schematic diagram of the steps of a 3D modeling data management method provided by an embodiment of the present invention; Figure 2 Schematic diagram of the structure of a 3D modeling data management system provided by an embodiment of the present invention. Detailed implementation manners

[0018] To further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation manners, structures, features and their effects according to the present invention.

[0019] This embodiment provides a 3D modeling data management method, as Figure 1 shown, including the steps: S1. Data collection: Obtain the original construction data of 3D modeling, the environmental attribute data in the target area, and the operation data generated by the user interaction platform.

[0020] The original construction data includes geometric structure data, texture mapping data, and material attribute data; the environmental attribute data includes light intensity, temperature distribution, and humidity distribution; the operation data includes data update frequency setting and user call instructions.

[0021] S2. Module reorganization: Input the original construction data and environmental attribute data into a data fusion algorithm, and generate fusion data for each sub-module of 3D modeling according to the data categories; wherein, the calculation formula of the data fusion algorithm is: ; In the formula, represents the weighted sum of geometric structure, texture mapping, material attribute, light intensity, temperature distribution, and humidity distribution, represents the th type of geometric structure data, represents the th type of texture mapping data, represents the th type of material attribute data, represents the light intensity, represents the temperature distribution, represents the humidity distribution, , , respectively represent the weight coefficients of the th type of geometric structure data, the th type of texture mapping data, and the th type of material attribute data, , , Weight coefficients representing light intensity, temperature distribution, and humidity distribution respectively.

[0022] In the acquisition of the fusion data of each sub-module, each weight coefficient is adjusted according to the access situation of the sub-modules accessed in the 3D modeling task. For example, in geometric structure modeling, the main module is the geometric structure data sub-module, and the sub-modules are the texture mapping sub-module, the material property sub-module, and so on. By continuously correcting the weight coefficient of data fusion during the access process of the data sub-module, the access interleaving complexity of the sub-module can be reduced, and the storage allocation and data update rate of subsequent data management can be improved.

[0023] It should be noted that each sub-module of 3D modeling is set according to the data type. For example, in the geometric structure module, in this embodiment, through data fusion, the data of each sub-module is weighted and fused to meet the access requirements for data in the 3D modeling process. For example, in the modeling of geometric structures, the most basic is the structure data, but at the same time, the auxiliary judgment of material properties and the like is required. At this time, there is no need to repeatedly access the material property data, thereby reducing the access frequency and interleaving complexity of each sub-module. In the initial fusion processing of data, the access habits of users to sub-modules are affected, making the subsequent operation data collected by priority decision-making clearer, and improving the accuracy of decision-making data storage and update.

[0024] It can be understood that the data fusion algorithm avoids the limitations caused by single data to modeling access by fusing multi-dimensional data, and then generates a more comprehensive data sub-module. The role of data fusion calculation is to directly access the fusion value to reduce module calls, and use dynamic weights to guide high-speed / low-speed storage division to reduce the broadband occupancy rate; as an overall update, it avoids scattered updates of multiple sub-modules; through the access of the main module, the necessary sub-module information is implied, reducing the operation complexity. Therefore, the system has achieved qualitative improvement in terms of efficiency (response speed, resource utilization), user experience (operation simplification, friendly interface), adaptability (dynamic parameter adjustment), etc., and the associated storage of original data provides flexibility for in-depth operations. By forming a hierarchical cooperation mechanism with the original data through associated storage with the original data, in high-frequency operations, it quickly responds according to the fusion value, and in low-frequency in-depth operations, it loads the original data from low-speed storage as needed. By associating the fusion value M with the metadata for storage, a metadata association, hierarchical storage, and dynamic weight management mechanism is established to achieve rapid decision-making in data management and in-depth operations on metadata, while ensuring the efficiency of 3D modeling, taking into account the flexibility and interpretability of data.

[0025] S3. Priority decision: According to the operation data, obtain the periodic access frequency of the user to the 3D modeling sub-module, and judge the access priority of each sub-module in the comprehensive 3D modeling; then combine the current access frequency of the sub-module to judge the storage priority of each sub-module in the comprehensive 3D modeling, including: Calculate the access heat value of each sub-module in 3D modeling according to the obtained periodic access frequency. The calculation formula for the access heat value is as follows: ; In the formula, represents the access heat value of the -th sub-module, represents the number of accesses by the user to this sub-module during the -th time period, represents the total number of time periods.

[0026] Combined with the data update frequency setting in the operation data, construct a data update frequency weight coefficient, and perform weighted processing on the access heat value of each sub-module to obtain the access priority of each sub-module. The calculation formula for the access priority is as follows: ; Among them, represents the access priority of the -th sub-module, represents the data update frequency weight corresponding to the -th sub-module.

[0027] It should be noted that the access heat value is a representation of the actual access frequency of each sub-module. By adding the data update frequency setting to the calculation of the access priority and incorporating the data update requirement setting, the efficiency, consistency, and resource utilization rate of the system can be significantly improved; by designing the update frequency weight coefficient of the sub-module through the data update frequency setting, the access priority calculation meets the goal of "the higher the access frequency and the higher the update frequency, the higher the priority score of the data", thus meeting the strong real-time data requirements, collaboration consistency requirements, and dynamic data-driven loop efficiency in the 3D modeling process.

[0028] Calculate the storage priority of each sub-module according to the access priority of the sub-module and the current access frequency. The calculation formula for the storage priority is as follows: ; Among them, represents the storage priority of the -th sub-module, represents the current access frequency, represents the weight coefficient.

[0029] It should be noted that the periodic access frequency represents the access status of each sub-module within a certain period of time, while the current access frequency is the access status within a short period of time in case of emergencies. By combining the two, the obtained storage priority can handle data that may be frequently accessed and unpredictable access surges, thus maximizing the cache efficiency.

[0030] It is understandable that both in the priority decision-making and subsequent steps, it is carried out based on the access situation of the data fusion sub-module obtained through module reorganization, so that the association between the fusion value and the access situation is implicitly reflected in the sub-module.

[0031] S4. Data Migration: Divide the storage area of the system into a cache area and a low-speed cache area, and generate the migration decision of the sub-module through a non-cooperative game model according to the storage priority, so as to avoid the sub-module maliciously preempting the cache resources and causing cache thrashing. The generation of the migration decision includes the steps of: Input Data Definition: Take the set of sub-modules to be migrated as the set of participants, and take the storage priority, access latency, storage cost, and migration overhead of each sub-module as the information set; Non-Cooperative Game Model Construction: Take each sub-module as an independent decision-making entity, and maximize its own benefit by balancing the access latency, storage cost, and migration overhead of each sub-module; Solve the Nash equilibrium strategy combination so that no sub-module can unilaterally improve its own benefit by changing the strategy; Determine the final migration decision of each sub-module according to the Nash equilibrium strategy combination; The solution of the Nash equilibrium strategy combination includes the steps of: Initialization: Set the current system status as the root node, and set the number of simulation times as N; Selection: Recursively select child nodes from the root node, and balance exploration and exploitation through the UCB (Upper Confidence Bound) formula, so that the algorithm can efficiently cover the strategy space under limited computing resources, thereby avoiding local optimality; Simulation: Randomly perform a complete migration of the sub-module from the current node, and execute the migration operation according to the strategy combination, and calculate the benefits of each sub-module and the total system benefit; Backpropagation: Propagate the simulation results backward to the parent node, and update the average benefit and access times of the node; Output: After N iterations, output the strategy combination with the most access times as the Nash equilibrium strategy combination; The UCB formula is: ; Where is the average benefit, is the exploration times.

[0032] It is understandable that the fusion sub-module has a higher priority than the benefit function and policy selection. Through the non-cooperative game framework, self-optimization and stable allocation of data migration are achieved, thereby completing the data migration of the sub-module between the cache area and the low-speed cache area. If the data migration of the sub-module between the cache area and the low-speed cache area is completed through one-way migration, the storage threshold can be set to quickly make a decision on data storage according to the storage priority.

[0033] S5. Module update decision: Combining the access priority and storage priority of each sub-module, the update frequency is dynamically adjusted through a non-linear weighted model to balance the relationship between real-time performance and resource efficiency, and avoid excessive consumption of computing resources (such as frequent synchronization of low-priority data), including: The calculation formula of the non-linear weighted model is: Where, is the update frequency of the th sub-module, which is used to adjust the update rhythm of the sub-module and balance the data freshness and system resource consumption. is the balance coefficient, and , and are the sensitivity parameters of the access priority and the storage priority respectively, which are used to control the non-linear influence of the access priority and the storage priority on the update frequency, and and are both greater than 0; If , the influence of the access priority on the update frequency shows exponential growth. For example, the update frequency of the sub-module with high-frequency access will increase sharply; if , the influence of the access priority on the update frequency tends to be gentle. For example, the update frequency of the sub-module with low-frequency access will decrease slowly; When the storage location of the sub-module is in the cache area, is set to 2; when the storage location of the sub-module is in the low-speed cache area, is set to 0.5; At the same time, a real-time update threshold is set for the update frequency of the sub-module. If the update frequency of the sub-module is greater than the real-time update threshold, an active synchronization strategy is implemented to ensure that it can be applied to scenarios with high-delay parameter requirements.

[0034] S6. Parameter tuning: Based on historical operation data, a reinforcement learning model is constructed to achieve parameter adaptive tuning for dynamically adjusting the update frequency, including: According to the access priority and storage priority, the state space, action space and reward function of reinforcement learning are defined; the state space includes the access heat value, the current access frequency and the data update frequency weight. ; The state space characterizes the dynamic features of the sub-module through multi-dimensional information (historical behavior, real-time demand, user preference), providing a decision-making basis for the reinforcement learning model; The action space includes sensitivity parameters of the balance coefficient and access priority; ; The action space defines the range of policy adjustments that the system can execute. For example, if the current update frequency is too high, resulting in resource waste, the influence of the access priority can be reduced to decrease the update frequency; if real-time performance needs to be enhanced, the sensitivity of the access priority to the update frequency can be amplified; If the real-time performance needs to be enhanced, the sensitivity of the access priority to the update frequency can be amplified; Amplify the sensitivity of the access priority to the update frequency; The calculation formula of the reward function is: Where, is the cache hit rate, which is used to measure the data access efficiency, is the data freshness, which is used to reflect the synchronization degree between the model and the real state, is the consumption of computing resources, 、 and are target weight coefficients, which are used to balance the priorities of different targets; The reward function guides the reinforcement learning model to optimize the parameter adjustment direction by quantifying the benefits and costs of the policy. For example: when it is necessary to maximize the cache hit rate, increase , reduce redundant data transmission, and improve the system response speed; when it is necessary to minimize the consumption of computing resources, increase , avoid resource waste caused by excessive adjustment; when it is necessary to balance the data freshness, increase , ensure the synchronization between the model and the dynamic environment.

[0035] It should be noted that in the design of the access priority in this embodiment, following the goal of "the higher the access frequency and the higher the update frequency, the higher the data priority score", there is a low hit rate, frequent failures leading to a high backend load. Therefore, by constructing a reinforcement learning model, the balance coefficient and sensitivity parameters are optimized to optimize the dynamic adjustment of the update frequency, so as to address the problem of low cache hit rate.

[0036] This embodiment also provides a three-dimensional modeling data management system, as Figure 2 shown, including a data collection module, a data reorganization module, a priority management module, a storage scheduling module, an update control module, and a parameter optimization module; The data collection module is used to obtain raw construction data, environmental attribute data, and user operation data from different sources to implement data standardization preprocessing.

[0037] It is understandable that the original construction data includes geometric models, material parameters, etc.; the environmental attribute data includes the terrain, lighting, physical properties of the target area, etc.; and the user operation data comes from the real-time operation records of the interaction platform.

[0038] The data reorganization module integrates the original construction data and the environmental attribute data through a data fusion algorithm to generate sub-module fusion data for 3D modeling, classifies and processes it according to data categories (such as geometry, texture, physical properties), ensures the logical consistency of the sub-module data, and provides a structured input for the subsequent modules.

[0039] The priority management module dynamically calculates the access priority of the sub-modules based on the user operation data (such as the periodic access frequency of the sub-modules), and at the same time combines the current access frequency (real-time load) and the access priority to generate a storage priority ranking of the sub-modules, guiding the storage resource allocation.

[0040] The storage scheduling module is used to divide the storage area into a cache area (high-frequency access data) and a low-speed cache area (low-frequency access data), manage the capacity and performance balance of the two types of storage; based on the storage priority, optimize the data migration strategy through a non-cooperative game model (such as Nash equilibrium) to ensure the fast response of high-priority sub-modules and reduce the storage cost.

[0041] The update control module is used to dynamically adjust the update frequency of the sub-modules by using a non-linear weighting model, and the weight factors include access priority, storage priority, and resource occupancy rate; balance the real-time performance (such as user interaction requirements) and resource efficiency (such as computing and storage consumption), and avoid performance bottlenecks caused by excessive updates.

[0042] The parameter optimization module constructs a reinforcement learning model (such as Q-learning or deep reinforcement learning) based on historical operation data, dynamically tunes key parameters such as the update frequency and storage migration threshold, realizes the adaptive optimization of the system, improves the long-term operation efficiency, and reduces the need for manual intervention.

[0043] It should be noted that the parameter optimization module iteratively improves the strategies of other modules through historical data to form a dynamic tuning closed loop. The cache area gives priority to serving high-priority sub-modules, and combines the dynamic update strategy to ensure the smoothness of user interaction. Solve the resource competition problem of multiple sub-modules through a non-cooperative game model to optimize the global storage efficiency; adopt non-linear update control to avoid resource waste caused by traditional linear weighting and accurately match real-time requirements; realize parameter self-adaptability through reinforcement learning tuning to adapt to complex and changing user operation modes. Through modular design, this system effectively solves the conflict problems of massive data management, real-time response, and resource efficiency in 3D modeling.

[0044] The present invention reorganizes sub-modules for three-dimensional modeling data, integrates multi-source heterogeneous data according to weight settings to complete the integration of sub-modules of different data types, avoids the limitations caused by single data for modeling access, and at the same time reduces the access frequency and interleaving complexity of each sub-module from the bottom layer through the fusion of sub-modules, providing clear operation data for priority decision-making; on the basis of completing necessary sub-module integration, makes decisions on access priority and storage priority for three-dimensional modeling sub-modules according to operation data, realizes precise resource allocation and real-time response optimization through dynamic priority management, and conducts global efficiency optimization through storage migration driven by game based on storage priority; then dynamically adjusts the update frequency according to the priority through a non-linear weighting model, achieving a balance between resource efficiency and real-time performance; finally, introduces a reinforcement learning model to realize parameter self-adaptation tuning, further optimizing the problem of reduced cache hit rate caused by priority decision-making. The present invention ensures fast response of key data through priority stratification and cache mechanism, while being compatible with large-scale data storage; the dynamic update strategy and game model find the optimal solution between user interaction experience and hardware resource consumption.

[0045] The above are only the preferred embodiments of the present invention, and do not impose any form of limitation on the present invention. Although the present invention has been disclosed above with preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to the above-disclosed technical content to form equivalent embodiments with equivalent changes, but as long as the technical content of the present invention is not departed from, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A three-dimensional modeling data management method, characterized in that: Including the steps: Data collection: Obtain the original construction data for 3D modeling, environmental attribute data within the target area, and operation data generated by the user interaction platform; Module reorganization: Input the original construction data and environmental attribute data into the data fusion algorithm, and generate fusion data for each sub-module of the 3D modeling according to the data category; Priority decision-making: Obtain the periodic access frequency of the user to the 3D modeling sub-modules according to the operation data, and determine the access priority of each sub-module in the comprehensive 3D modeling; then combine the current access frequency of the sub-module to determine the storage priority of each sub-module in the comprehensive 3D modeling; Data migration: Divide the storage area of the system into a cache area and a low-speed cache area, and generate a migration decision for the sub-module through a non-cooperative game model according to the storage priority; Module update decision-making: Combine the access priority and storage priority of each sub-module, and dynamically adjust the update frequency through a non-linear weighted model; Parameter tuning: Build a reinforcement learning model based on historical operation data to achieve parameter adaptive tuning for dynamically adjusting the update frequency.

2. The three-dimensional modeling data management method according to claim 1, wherein: The original construction data includes geometric structure data, texture mapping data, and material attribute data; the environmental attribute data includes light intensity, temperature distribution, and humidity distribution; the operation data includes data update frequency settings and user call instructions.

3. A 3D modeling data management method according to claim 1, characterized in that: The calculation formula of the data fusion algorithm is: ； In the formula, represents the weighted sum of geometric structure, texture map, material property, light intensity, temperature distribution, and humidity distribution, represents the th type of geometric structure data, represents the th type of texture map data, represents the th type of material property data, represents the light intensity, represents the temperature distribution, represents the humidity distribution, , , respectively represent the weight coefficients of the th type of geometric structure data, the th type of texture map data, and the th type of material property data, , , respectively represent the weights of light intensity, temperature distribution, and humidity distribution.

4. A three-dimensional modeling data management method according to claim 1, characterized in that: The priority decision-making includes the following steps: Calculate the access popularity value of each sub-module in the 3D modeling according to the obtained periodic access frequency; Combine the data update frequency setting in the operation data to construct a data update frequency weight coefficient, and perform weighted processing on the access popularity value of each sub-module to obtain the access priority of each sub-module; Calculate the storage priority of each sub-module according to the access priority and current access frequency of the sub-module.

5. A 3D modeling data management method according to claim 4, characterized in that: The calculation formula of the access popularity value is: ； In the formula, represents the access heat value of the th sub-module, represents the total number of time periods; The calculation formula of the access priority is: ； Among them, represents the access priority of the th sub-module, represents the data update frequency weight corresponding to the th sub-module; The calculation formula of the storage priority is: ； Among them, represents the storage priority of the th sub-module, represents the current access frequency, represents the weight coefficient.

6. A three-dimensional modeling data management method according to claim 1, characterized in that: The generation of the migration decision includes the steps: Take the set of sub-modules to be migrated as the set of participants, and take the storage priority, access latency, storage cost, and migration overhead of each sub-module as the information set; Take each sub-module as an independent decision-making entity, and maximize its own benefit by balancing the access latency, storage cost, and migration overhead of each sub-module; Solve the Nash equilibrium strategy combination, and determine the final migration decision of each sub-module according to the Nash equilibrium strategy combination; The solution of the Nash equilibrium strategy combination includes the steps: Set the current system status as the root node and set the number of simulations as N; Recursively generate child nodes from the root node, balance exploration and exploitation through the UCB formula, randomly perform a complete migration of the sub-module from the current node, and execute the migration operation according to the strategy combination, and calculate the benefit of each sub-module and the total system benefit; Backpropagate the simulation results to the parent node, and update the average benefit and access times of the node; After N iterations, output the strategy combination with the most access times as the Nash equilibrium strategy combination; Among them, the UCB formula is: ； In the formula, is the average return, is the number of explorations.

7. A three-dimensional modeling data management method according to claim 5, characterized in that: The calculation formula of the non-linear weighted model is: Wherein, is the update frequency of the th sub-module, is the balance coefficient, and , and are the sensitivity parameters of the access priority and the storage priority respectively, and and are both greater than 0; If , the influence of access priority on the update frequency increases exponentially; if , the influence of access priority on the update frequency tends to level off; When the storage location of the sub-module is in the cache area, set it to 2; when the storage location of the sub-module is in the low-speed cache area, set it to 0.5; At the same time, a real-time update threshold is set for the update frequency of the sub-module. If the update frequency of the sub-module is greater than the real-time update threshold, an active synchronization strategy is implemented.

8. A three-dimensional modeling data management method according to claim 1, characterized in that: The parameter tuning includes the following steps: Define the state space, action space, and reward function of reinforcement learning according to the access priority and storage priority; the state space includes the access popularity value, the current access frequency, and the data update frequency weight, expressed as ; the state space characterizes the dynamic features of the sub-module through multi-dimensional information, providing a decision-making basis for the reinforcement learning model; The action space includes a balance coefficient and a sensitivity parameter of access priority, expressed as ; the action space defines the range of policy adjustments executable by the system; The calculation formula of the reward function is: Wherein, is the cache hit rate, is the data freshness, is the consumption of computing resources, , and are the target weight coefficients.

9. A three-dimensional modeling data management system that executes a three-dimensional modeling data management method according to any one of claims 1-8, characterized in that: It includes a data collection module, a data reorganization module, a priority management module, a storage scheduling module, an update control module, and a parameter optimization module; The data collection module is used to obtain raw construction data, environmental attribute data, and user operation data from different sources, and implement data standardization preprocessing; The data reorganization module is used to integrate the raw construction data and environmental attribute data through a data fusion algorithm to generate sub-module fusion data for 3D modeling; The priority management module dynamically calculates the access priority of the sub-module based on the user operation data, and at the same time combines the current access frequency and access priority to generate a storage priority ranking of the sub-module; The storage scheduling module is used to divide the storage area into a cache area and a low-speed cache area, manage the capacity and performance balance of the two types of storage, and optimize the data migration strategy through a non-cooperative game model based on the storage priority; The update control module is used to dynamically adjust the update frequency of the sub-module by using a non-linear weighting model to balance real-time performance and resource efficiency; The parameter optimization module constructs a reinforcement learning model based on historical operation data, dynamically tunes the update frequency and storage migration threshold, and realizes system adaptive optimization.

10. A storage medium, on which a computer program is stored, characterized in that, When the computer program is executed by a processor, it runs the steps in the method according to any one of claims 1 to 8.

Images