A web-based infrastructure programmatic modeling system

Abstract

The present invention discloses a web-based infrastructure programmatic modeling system, which relates to the technical field of data transmission; obtaining a data file on the web side for programmatic modeling to obtain a target file; performing congestion detection on the transmission path, and if the transmission path is determined to be congested, performing path optimization to obtain a transmission strategy; performing spatial compression on the target file to obtain target point cloud data, converting the target point cloud data into structured data, and performing security detection on the structured data; if the structured data is determined to be secure, performing data transmission according to the transmission strategy. Performing security detection on the structured data, and only the data determined to be secure will be transmitted according to the optimized transmission strategy, thereby realizing efficient and secure transmission of data between the web side and the cloud side, and ensuring the integrity and transmission efficiency of the data.

Description

Technical Field

[0001] The present invention belongs to the technical field of data transmission, and particularly relates to a web-based infrastructure programmed modeling system. Background Art

[0002] With the rapid development of Internet technology, web-based modeling systems have gradually become important tools for digital transformation in various industries. In the early days, web technology mainly consisted of static pages. However, with the continuous evolution of technologies such as HTML, CSS, and JavaScript, web applications have gradually developed towards dynamic and interactive directions. In recent years, the rise of Web3D technology has further promoted the development of web modeling systems. Through technologies such as WebGL and WebGPU, users can directly render and interact with 3D models in the browser without additional plugins. In addition, the integration of AI technology has also brought the possibility of automation and intelligence to web modeling, enabling faster model generation and optimization. In the field of systems engineering, the model-based systems engineering (MBSE) method is widely used, and web modeling systems provide more convenient tool support for MBSE, making modeling more efficient and collaborative.

[0003] In the prior art, web-based infrastructure faces difficulties in standardizing heterogeneous data formats, low efficiency due to congested transmission paths, and insufficient security threatening data integrity during the process of data transmission to the cloud. Summary of the Invention

[0004] The purpose of the present invention is to solve the problems that web-based infrastructure faces difficulties in standardizing heterogeneous data formats, low efficiency due to congested transmission paths, and insufficient security threatening data integrity during the process of data transmission to the cloud, and to propose a web-based infrastructure programmed modeling system.

[0005] In the first aspect of the implementation of the present invention, a web-based infrastructure programmed modeling system is first proposed. The system includes: a data modeling module, a path analysis module, a data transmission module, and a task execution module:

[0006] The data modeling module is used to obtain data files on the web side and perform programmed modeling on the data files to obtain target files;

[0007] The path analysis module is used to detect congestion in the transmission path for the current cycle. If the transmission path is determined to be congested, path optimization is performed to obtain a transmission strategy;

[0008] The data transmission module is used to perform spatial compression on the target file to obtain target point cloud data, convert the target point cloud data into structured data, and perform security detection on the structured data;

[0009] The task execution module is configured to perform data transmission according to the transmission policy if the structured data is determined to be secure.

[0010] Optionally, the path analysis module includes: a data detection module, a delay calculation module, and a weight calculation module:

[0011] The data detection module is configured to send detection data packets to the web side and the cloud side respectively to obtain a first detection time and a second detection time, and the web side and the cloud side receive the detection data packets and send the detection data packets to each other to obtain a first verification time and a second verification time;

[0012] The delay calculation module is configured to calculate a link delay according to the first detection time, the second detection time, the first verification time, and the second verification time.

[0013] The weight calculation module is configured to obtain the minimum remaining bandwidth of the target path, calculate a path delay according to the link delays of the links in the target path, and calculate a target path weight according to the minimum remaining bandwidth and the path delay.

[0014] Screen the transmission paths according to the target path weights to obtain a set of transmission paths, and determine the transmission policy according to the set of transmission paths.

[0015] Optionally, the path analysis module further includes a policy optimization module; the policy optimization module is configured to obtain an expert data set and a historical log set, and input the path weights, link bandwidths, and historical log set into a preset model to obtain a transmission policy; the policy optimization module is configured to execute the following steps, and the specific steps include:

[0016] Step 1: Divide the expert data set according to a preset rule to obtain expert data subsets, and determine expert policies according to the expert data subsets;

[0017] Step 2: Initialize the policy to obtain a target policy, and take the target policy as the optimal solution of the reward function;

[0018] Step 3: Save the dynamic distribution of the target policy , and update the reward function through an online projection gradient method;

[0019] Step 4: Determine the corresponding historical log according to the target policy, calculate a service quality evaluation value according to the link bandwidth, packet loss rate, and historical log, and determine the transmission policy according to the service quality evaluation value.

[0020] Optionally, the policy optimization module further includes: an evaluation value calculation module and a policy determination module:

[0021] The evaluation value calculation module is used to calculate the quality of service evaluation value according to the historical log, cache packet transmission speed, and path weight, and select the transmission path according to the quality of service evaluation value; the historical log includes: average delay and maximum packet loss rate;

[0022] The policy determination module is used to, if the quality of service evaluation value is greater than the preset threshold, take the transmission path as the transmission policy;

[0023] Quality of service calculation formula:

[0024]

[0025] where, represents the transmission quality, represents the standard cache packet transmission speed, represents the cache packet data transmission speed, represents the packet loss rate, represents the average delay, T represents the unit time, and W represents the path weight.

[0026] Optionally, the data transmission module further includes: a file cutting module and a data compression module:

[0027] The file cutting module is used to determine the Euclidean space of the target file, and divide the Euclidean space into multiple data blocks according to a preset length;

[0028] The data compression module is used to, if any data block intersects with the Euclidean space, set marker points in the data block through a preset rule, and combine all the marker points to obtain the target point cloud data.

[0029] Optionally, the data compression module further includes: a data conversion module, a model training module, and a security detection module:

[0030] The data conversion module is used to convert the target point cloud data into structured data and obtain a historical data set; the historical data set includes: a historical training data set and a historical verification data set;

[0031] The model training module is used to input the historical training data set into a preset model for training to obtain target model parameters, and update the model parameters of the preset model according to the target model parameters to obtain a target model;

[0032] The security detection module is used to input the historical verification data set into the target model for verification to obtain a pass rate. If the pass rate > the preset pass rate, it is determined that the model training is completed, and the structured data is input into the target model for security detection. Otherwise, the target model is retrained.

[0033] Optionally, the task execution module is further configured to, if the structured data is determined to be dangerous, perform a risk assessment on the structured data to obtain a feedback report, and report the vulnerabilities in the feedback report.

[0034] Advantages of the present invention:

[0035] The present invention provides a web-based infrastructure programmatic modeling system. By programmatically modeling the data files on the web side, a standardized target file is generated, and spatial compression is performed to convert it into structured data. At the same time, the congestion of the transmission path is detected and the path is optimized to ensure the efficiency and reliability of data transmission. The structured data is subjected to security detection, and only the data determined to be safe will be transmitted according to the optimized transmission strategy, thereby realizing the efficient and secure transmission of data between the web side and the cloud side, and ensuring the integrity and transmission efficiency of the data. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The present invention will be further described below with reference to the accompanying drawings.

[0037] Figure 1 FIG. is a flowchart of a web-based infrastructure programmatic modeling system provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0038] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The term "and / or" in this article is only a description of the associated relationship of the associated objects, indicating that there can be three relationships. For example, A and B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the descriptions such as "first" and "second" in the present invention are only for the purpose of description, and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the fact that those skilled in the art can implement them. When the combination of the technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0039] All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the protection scope of the present invention.

[0040] An embodiment of the present invention provides a web-based infrastructure programmatic modeling system. Refer to Figure 1 , Figure 1 which is a flowchart of a web-based infrastructure programmatic modeling system provided by an embodiment of the present invention. The system includes the following modules: a data modeling module, a path analysis module, a data transmission module, and a task execution module:

[0041] The data modeling module is used to obtain the data file on the web side and perform programmatic modeling on the data file to obtain the target file;

[0042] The path analysis module is used to detect congestion in the transmission path for the current cycle. If the transmission path is determined to be congested, path optimization is performed to obtain the transmission strategy;

[0043] The data transmission module is used to perform spatial compression on the target file to obtain the target point cloud data, convert the target point cloud data into structured data, and perform security detection on the structured data;

[0044] The task execution module is used to perform data transmission according to the transmission strategy if the structured data is determined to be secure.

[0045] Based on the web-based infrastructure programmatic modeling system provided by an embodiment of the present invention, by performing programmatic modeling on the web-side data file, a standardized target file is generated, and spatial compression is performed to convert it into structured data. At the same time, congestion in the transmission path is detected and the path is optimized to ensure the efficiency and reliability of data transmission. Security detection is performed on the structured data, and only the data determined to be secure will be transmitted according to the optimized transmission strategy, thereby realizing efficient and secure transmission of data between the web side and the cloud side, and ensuring the integrity and transmission efficiency of the data.

[0046] In one implementation, by performing programmatic modeling on the data file obtained from the web side, the original data can be converted into a target file. The technical effect of this process is to realize the normalization and standardization processing of the data. Programmatic modeling can automatically extract the key features of the data and generate a formatted file suitable for subsequent processing, thereby improving the efficiency and accuracy of data processing. At the same time, this process can reduce manual intervention, ensure the consistency and integrity of the data, and provide a high-quality data basis for subsequent transmission and analysis.

[0047] In one implementation, congestion detection is performed on the transmission path for the current cycle. If the path is determined to be congested, path optimization is carried out to obtain a transmission strategy. The technical effect of this process is to dynamically adjust the transmission path to ensure the efficiency and reliability of data transmission. By monitoring the congestion situation of the path in real time, potential bottlenecks can be detected in a timely manner, and a better transmission path can be selected through an optimization algorithm. This not only reduces transmission latency but also improves the utilization rate of network resources, enhances the adaptive ability of the system, and thus maintains stable transmission performance in a complex network environment.

[0048] In one implementation, spatial compression is performed on the target file to obtain target point cloud data, which is then converted into structured data. The technical effect of this process is to significantly reduce the storage and transmission costs of data while retaining the core information of the data. Spatial compression can remove redundant data and improve the transmission efficiency of data, while the structured conversion of point cloud data provides a suitable format for subsequent security detection. This processing method not only optimizes the organization form of data but also facilitates subsequent analysis and processing, ensuring that the data can be efficiently processed in security detection.

[0049] In one implementation, security detection is performed on the structured data. The technical effect of this process is to ensure the security and integrity of the data during transmission. Through security detection, anomalies or potential threats in the data, such as data tampering and malicious code injection, can be detected in a timely manner. The implementation of security detection provides the last line of defense for data transmission, preventing sensitive information from being leaked or maliciously exploited. Only when the data is determined to be secure will it enter the next transmission process, thus effectively ensuring the security of the data.

[0050] In one implementation, if the structured data is determined to be secure, data transmission is performed according to the optimized transmission strategy. The technical effect of this process is to achieve efficient, secure, and reliable data transmission. The implementation of the transmission strategy is based on the result of path optimization to ensure that the data is transmitted through the optimal path. At the same time, combined with the guarantee of security detection, threats to the data during transmission are avoided. This method that comprehensively considers transmission efficiency and security can significantly improve the overall performance of the system and ensure stable transmission of data between the Web side and the cloud side.

[0051] In one embodiment, the path analysis module includes: a data detection module, a delay calculation module, and a weight calculation module:

[0052] The data detection module is used to send detection data packets to the web side and the cloud side respectively to obtain the first detection time and the second detection time. The web side and the cloud side receive the detection data packets and send detection data packets to each other to obtain the first verification time and the second verification time;

[0053] A delay calculation module, which is used to calculate the link delay according to the first detection time, the second detection time, the first verification time, and the second verification time.

[0054] A weight calculation module, which is used to obtain the minimum remaining bandwidth of the target path, calculate the path delay according to the link delays of each link in the target path, and calculate the target path weight according to the minimum remaining bandwidth and the path delay.

[0055] In one implementation, by sending detection data packets to the Web side and the cloud side respectively and obtaining the first detection time and the second detection time, and at the same time allowing the Web side and the cloud side to send detection data packets to each other to obtain the first verification time and the second verification time, the two-way transmission time of the network link can be accurately measured. The technical effect of this process is that the delay data of the network link can be obtained in real time, providing accurate timing information for subsequent link delay calculation, thereby ensuring the reliability of link performance evaluation.

[0056] In one implementation, calculating the link delay based on the first detection time, the second detection time, and the verification time can accurately quantify the network delay between the Web side and the cloud side. The technical effect of this is to provide key performance indicators for path selection, help identify potential bottlenecks in the network, thereby optimizing the data transmission path and reducing the transmission delay. For example: the first detection time and the second detection time , the first verification time and the second verification time , link delay: , where LY represents the link delay, represents the first detection time, represents the second detection time, represents the first verification time, represents the second verification time.

[0057] In one implementation, obtaining the minimum remaining bandwidth of the target path and calculating the path delay according to the delays of each link can comprehensively evaluate the transmission capacity and performance of the path. The minimum remaining bandwidth reflects the available bandwidth resources of the path, while the path delay synthesizes the delay conditions of all links. The technical effect of this is to provide comprehensive performance parameters for the calculation of the path weight, ensuring that the path selection not only considers the delay but also takes into account the availability of bandwidth resources. For example: path delay: , where JY represents the path delay, LY represents the link delay, and j represents there are j links; path weight: , where W represents the path weight, DK represents the minimum remaining bandwidth, and JY represents the path delay.

[0058] In one implementation, the target path weight is calculated by combining the minimum remaining bandwidth and the path delay, which can comprehensively evaluate the transmission performance of the path. The path weight reflects the comprehensive advantages of the path in terms of delay and bandwidth. The higher the weight, the better the transmission performance of the path. This technical effect provides a quantitative standard for the selection of the transmission path, making the path screening process more scientific and reasonable.

[0059] In one implementation, the transmission path is screened according to the target path weight to obtain a set of transmission paths, and finally the transmission strategy is determined. The technical effect of this process is that the optimal transmission path can be selected from multiple candidate paths, so as to achieve efficient and reliable data transmission. By screening the path set, it can be ensured that the most suitable transmission path for the current network state is found in the multi-path selection, improving the overall transmission efficiency and reliability.

[0060] In one embodiment, the path analysis module further includes a policy optimization module; the policy optimization module is used to obtain an expert data set and a historical log set, and input the path weight, link bandwidth, and historical log set into a preset model to obtain a transmission strategy; the policy optimization module is used to execute the following steps, and the specific steps include:

[0061] Step 1, divide the expert data set according to a preset rule to obtain an expert data subset, and determine an expert strategy according to the expert data subset;

[0062] Step 2, initialize the strategy to obtain a target strategy, and set the target strategy as the optimal solution of the reward function;

[0063] Step 3, save the dynamic distribution of the target strategy , and update the reward function by the online projection gradient method;

[0064] Step 4, determine the corresponding historical log according to the target strategy, calculate the quality of service evaluation value according to the link bandwidth, packet loss rate, and historical log, and determine the transmission strategy according to the quality of service evaluation value.

[0065] In one implementation, the policy is initialized to obtain the target policy, and the target policy is the optimal solution of the reward function; that is, the policy is initialized to the optimal solution of the reward function, providing a reasonable starting point for subsequent optimization. This step can accelerate the convergence speed of the algorithm and avoid inefficient training caused by starting from a random policy. Policy: In reinforcement learning, a policy refers to the rule for selecting actions in a given state. A policy can be deterministic (fixed action) or stochastic (action probability distribution). Reward function: Used to measure the immediate reward for taking a specific action in a specific state. The reward function is the core of reinforcement learning and is used to guide the behavior of the agent. The expert dataset refers to high-quality data generated by experienced operators or optimization algorithms and is used to guide the learning algorithm. In Web and cloud data transmission, the expert dataset includes historical transmission paths, link performance metrics (such as latency, packet loss rate, bandwidth utilization), and corresponding transmission quality evaluation results. These data reflect the optimal transmission policies under different network conditions. The expert dataset consists of the union of expert data subsets , where represents the expert dataset, represents the expert data subset; the reward function is , and the dynamic distribution is , and the reward function is updated by for updating, where represents the expert policy.

[0066] In one implementation, save the dynamic distribution: Save the dynamic distribution under the current policy for comparison and adjustment with the expert policy in subsequent steps. This step ensures that the algorithm can track the changes in the policy in real time and provides a basis for updating the reward function. Glossary of technical terms: Dynamic distribution: Represents the state distribution of the system under the current policy. In reinforcement learning, the dynamic distribution reflects the state transition probability during the execution of the policy. Policy is the rule for selecting actions in a given state. In Web and cloud data transmission, the policy can be a rule for selecting the optimal transmission path based on the current network state (such as link latency, bandwidth utilization). Through imitation learning, the algorithm continuously optimizes the policy so that it can dynamically adjust the transmission path according to the real-time network state.

[0067] In one implementation, the reward function is updated by the online projected gradient method to make it closer to the expert policy. This step is the core of the algorithm. By continuously adjusting the reward function, the policy is optimized to gradually approach the optimal solution. The online projected gradient method is an optimization algorithm that dynamically adjusts parameters by calculating the gradient and projecting it into the feasible region. In imitation learning, it is used to update the reward function to approximate the expert policy.

[0068] In one implementation, the quality of service evaluation value is calculated based on parameters such as link bandwidth and packet loss rate to evaluate the performance of the current policy. This step provides a quantitative metric for subsequent policy updates, ensuring that the optimization direction meets the actual requirements. Quality of service: A metric for measuring the transmission performance of a network path, including latency, packet loss rate, jitter, throughput, etc. Link bandwidth: The transmission capacity of a network link, usually measured in bits per second (bps). Bandwidth directly affects the data transmission speed. Packet loss rate: The proportion of data packets lost during network transmission, which is a key metric for measuring network reliability.

[0069] In one implementation, the policy is gradually optimized through multiple iterations to achieve better performance in multi-objective optimization problems. The iterative process ensures that the algorithm can gradually approach the optimal solution while adapting to the dynamically changing environment. Multi-objective optimization: An optimization problem that simultaneously optimizes multiple objectives, usually requiring trade-offs between different objectives. Number of iterations: The number of loops executed by the algorithm, which determines the optimization accuracy and computational cost. Based on the quality of service evaluation value, the transmission policy is determined. Through multi-objective optimization (such as minimizing latency, maximizing bandwidth utilization, and reducing packet loss rate), the algorithm can find a set of optimal paths that meet the multi-objective requirements in a dynamic network environment. These paths can not only improve the efficiency and reliability of data transmission but also adapt to changes in the network state.

[0070] In one embodiment, the policy optimization module further includes: an evaluation value calculation module and a policy determination module:

[0071] The evaluation value calculation module is used to calculate the quality of service evaluation value based on historical logs, cache packet transmission speed, and path weight, and select the transmission path according to the quality of service evaluation value; the historical logs include: average latency and maximum packet loss rate;

[0072] The policy determination module is used to use the transmission path as the transmission policy if the quality of service evaluation value is greater than the preset threshold;

[0073] Quality of service calculation formula:

[0074]

[0075] Where, represents the transmission quality, represents the standard cache packet transmission speed, represents the cache packet data transmission speed, represents the packet loss rate, represents the average latency, T represents the unit time, and W represents the path weight.

[0076] In one implementation, is the cache packet transmission speed, which reflects the congestion situation of the transmission path. When the cache packet transmission speed is faster, the congestion situation is lighter; is the average latency in the historical transmission log. The lower the average latency, the better the quality of service; is the maximum packet loss rate in the historical transmission log. The lower the packet loss rate, the better the quality of service; W is the weight of the transmission path. The higher the weight, the better the quality of service, thus achieving efficient and reliable data transmission.

[0077] In one embodiment, the data transmission module further includes: a file cutting module and a data compression module:

[0078] The file cutting module is used to determine the Euclidean space of the target file and divide the Euclidean space into multiple data blocks according to a preset length;

[0079] The data compression module is used to set marker points in any data block through a preset rule if there is an intersection between the data block and the Euclidean space, and combine all the marker points to obtain the target point cloud data.

[0080] In one implementation, the m-dimensional space where the point cloud is located is divided into data blocks (cubes) with side length α. The purpose is to discretize the continuous point cloud data for subsequent processing. This division method simplifies the complex spatial data into a regular grid structure, so that each point in the point cloud is assigned to a specific data block. The space is divided into regular lattice points through the equivalence relation x ∼ y. This division method is similar to "slicing" the space, and each "slice" is a data block. Mathematical expression: , where the value of i is: i = 1, 2,... n. If the integer parts of two points after dividing by α in each dimension are the same, they are considered to belong to the same data block.

[0081] In one implementation, a representative point x∗ is selected from each hypercube C that intersects with the point cloud X, thereby reducing the scale of the point cloud. This step is the key to data simplification. By selecting representative points, the topological features of the original point cloud are retained while redundant information is removed. If at least one point in a certain data block C belongs to the original point cloud X, it is said that C intersects with X, that is, the interior of this hypercube contains points in the point cloud. Marker points are set in the data block through a preset rule. The determination method of the marker point x: Determine the position of the data block C, which is represented by integer coordinates ( ), where , and the coordinates of the marker point x are , and this point is usually the lower left corner vertex (or a certain fixed position) of the data block and does not necessarily belong to the original point cloud X.

[0082] In one implementation, all the marker points are combined to obtain the target point cloud data, and all the selected representative points x are combined into a new point cloud , and this new point cloud It retains the topological features of the original point cloud but is smaller in scale. When α increases, the volume of the data block increases, and the spatial range covered by each hypercube becomes wider. Therefore, more points will be mapped into the same hypercube, and the number of representative points finally selected decreases. As a result, the scale of the new point cloud decreases as α increases.

[0083] In one implementation, reducing the scale of the point cloud can reduce the storage and computing costs and improve the efficiency of subsequent processing. Although the scale of the point cloud is reduced, the topological structure of the original point cloud is retained by selecting representative points. By adjusting the size of α, the simplification degree can be flexibly controlled according to actual needs.

[0084] In one embodiment, the data compression module further includes: a data conversion module, a model training module, and a security detection module:

[0085] The data conversion module is used to convert the target point cloud data into structured data and obtain a historical data set; the historical data set includes: a historical training data set and a historical verification data set;

[0086] The model training module is used to input the historical training data set into a preset model for training to obtain target model parameters, and update the model parameters of the preset model according to the target model parameters to obtain a target model;

[0087] The security detection module is used to input the historical verification data set into the target model for verification to obtain a pass rate. If the pass rate > the preset pass rate, it is determined that the model training is completed, and the structured data is input into the target model for security detection. Otherwise, the target model is retrained.

[0088] In one implementation, converting the target point cloud data into structured data is a key step to achieve efficient processing and analysis. Point cloud data usually consists of a large number of unordered three-dimensional coordinate points and is difficult to be directly used as the input of a machine learning model. To convert it into structured data, the following methods can be adopted: Voxelization: Divide the point cloud data into a three-dimensional grid, and each grid cell (voxel) contains point information within a certain range, thus converting the unordered point cloud into a regular three-dimensional array. Feature extraction: Extract the geometric features (such as curvature, normal) and statistical features (such as density, local variance) of the point cloud, and organize these features into a table form. Deep learning method: Use a three-dimensional convolutional neural network (3D CNN) or a point cloud processing network (such as PointNet) to encode the point cloud to generate a structured feature vector. Through these methods, the point cloud data is converted into structured data, which can better adapt to the subsequent model training and security detection processes.

[0089] In one implementation, the preset model is a large language model: The preset model adopts a large language model (LLM). The above model has powerful natural language processing capabilities and context understanding capabilities, and can process complex text data and structured data. By inputting structured data into the large language model, its pre-trained parameters and powerful generalization capabilities can be utilized to quickly adapt to specific tasks, such as security detection. The effects brought by using the large language model include: Efficient feature learning: The large language model can automatically extract key features in the data, reducing the workload of manual feature engineering. Powerful context understanding: It can understand the complex relationships between data, thereby performing security detection more accurately. Quick adaptation to new tasks: Through fine-tuning or prompt engineering, the large language model can quickly adapt to new security detection tasks without having to train from scratch.

[0090] In one implementation, model training and validation: Input the historical training data set into the preset model for training to obtain the target model parameters, and update the model parameters to obtain the target model. Use the historical validation data set to validate the target model and calculate the pass rate. Model optimization: Through the training of historical data, the model can learn the patterns and rules in the data, thereby optimizing its own parameters and improving the prediction ability for new data. Performance evaluation: Through the test of the validation data set, the performance of the model can be objectively evaluated to ensure its reliability and accuracy in actual applications. Iterative improvement: If the pass rate does not meet the preset standard, perform secondary training to further optimize the model to ensure that its performance meets the actual requirements.

[0091] In one implementation, security detection of three-dimensional data transmission through a large language model Input the structured data into the target model for security detection. Utilizing the powerful capabilities of the large language model, efficient detection of security risks during the three-dimensional data transmission process can be achieved. The benefits it brings include: Precise identification of abnormal behaviors: The large language model can understand the context and patterns of the data and precisely identify abnormal behaviors during the transmission process, such as data tampering, data leakage, etc. Real-time security warning: By analyzing the transmitted data in real time, potential security threats can be detected in a timely manner and warnings can be issued to reduce the occurrence of security incidents. Adaptive ability: The large language model can dynamically adjust the detection strategy according to new data to adapt to the changing network environment and security threats. Reduction of false alarm rate: Utilizing its powerful context understanding and feature extraction capabilities, false alarms can be reduced and the accuracy of security detection can be improved.

[0092] In one embodiment, the task execution module is further configured to, if the structured data is determined to be dangerous, perform a risk assessment on the structured data to obtain a feedback report, and report the vulnerabilities in the feedback report.

[0093] In one implementation, quickly evaluate the detected insecure data to determine the data type, scope of impact, and potential risks; isolate the insecure data to prevent its further spread or impact on other systems, and block relevant network connections or access permissions; repair the exploited security vulnerabilities according to the feedback report, and update security policies and tools, including but not limited to firewalls and intrusion detection systems; strengthen the real-time monitoring and auditing of data flow, record all access and operation behaviors for tracking and analyzing potential security threats; verify the integrity of data backups and recover data from backups when necessary to reduce the losses caused by insecure data; continuously optimize data security policies and protection measures according to the event analysis results to prevent similar problems from occurring again.

[0094] The above has described a specific embodiment of the present invention in detail, but the content is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the application of the present invention should still fall within the scope covered by the patent of the present invention.

Claims

1. A web-based infrastructure programmatic modeling system, characterized in that: The system includes: a data modeling module, a path analysis module, a data transmission module and a task execution module: The data modeling module is used to obtain the data file on the web end, and perform programmatic modeling on the data file to obtain the target file; The path analysis module is used to perform congestion detection on the transmission path of the current cycle, and if the transmission path is determined to be congested, perform path optimization to obtain a transmission strategy; The data transmission module is used to perform spatial compression on the target file to obtain target point cloud data, convert the target point cloud data into structured data, and perform security detection on the structured data; The task execution module is used to perform data transmission according to the transmission strategy if the structured data is determined to be safe.

2. A web-based infrastructure programmatic modeling system according to claim 1, characterized in that: The path analysis module includes: a data detection module, a delay calculation module and a weight calculation module: The data detection module is used to send a detection data packet to the web end and the cloud end respectively to obtain a first detection time and a second detection time, and the web end and the cloud end receive the detection data packet and send the detection data packet to each other to obtain a first verification time and a second verification time; The delay calculation module is used to calculate the link delay according to the first detection time, the second detection time, the first verification time and the second verification time, The weight calculation module is used to obtain the minimum remaining bandwidth of the target path, calculate the path delay according to the link delay of each link in the target path, and calculate the target path weight according to the minimum remaining bandwidth and the path delay; The transmission paths are screened according to the target path weights to obtain a transmission path set, and a transmission strategy is determined according to the transmission path set.

3. A web-based infrastructure programmatic modeling system according to claim 2, characterized in that: The path analysis module also includes a strategy optimization module; The strategy optimization module is used to obtain an expert data set and a historical log set, and input the path weight, link bandwidth, and historical log set into a preset model to obtain a transmission strategy; The strategy optimization module is used to perform the following steps, which specifically include: Step 1: dividing the expert data set according to preset rules to obtain expert data subsets, and determining the expert strategy according to the expert data subsets; Step 2: Initialize the strategy to obtain the target strategy, and set the target strategy as the optimal solution of the reward function; Step 3: Save the dynamic distribution P of the target strategy π , update the reward function by online projected gradient method; Step 4: determine the corresponding historical log according to the target strategy, calculate the service quality evaluation value according to the link bandwidth, packet loss rate and historical log, and determine the transmission strategy according to the service quality evaluation value.

4. A web-based infrastructure programmatic modeling system according to claim 3, characterized in that: The strategy optimization module also includes: an evaluation value calculation module and a strategy determination module: The evaluation value calculation module is used to calculate the service quality evaluation value according to the historical log, the cache packet transmission speed and the path weight, and select the transmission path according to the service quality evaluation value; the historical log includes: average delay and maximum packet loss rate; The strategy determination module is used to use the transmission path as the transmission strategy if the service quality evaluation value is greater than a preset threshold; Service quality calculation formula: , Among them, f qos represents the transmission quality, f s1 represents the standard cache packet transmission speed, f s Represents the cache packet data transmission speed, f d represents the packet loss rate, f y represents the average delay, T represents the unit time, and W represents the path weight.

5. The web-based infrastructure programmatic modeling system according to claim 1, characterized in that: The data transmission module also includes: a file cutting module and a data compression module: The file cutting module is used to determine the Euclidean space of the target file, and divide the Euclidean space into multiple data blocks according to a preset length; The data compression module is used to set marking points in any data block according to preset rules if there is an intersection between any data block and the Euclidean space, and combine all marking points to obtain target point cloud data.

6. A web-based infrastructure programmatic modeling system according to claim 5, characterized in that: The data compression module also includes: a data conversion module, a model training module and a security detection module: The data conversion module is used to convert the target point cloud data into structured data and obtain a historical data set; the historical data set includes: a historical training data set and a historical verification data set; The model training module is used to input the historical training data set into a preset model for training to obtain target model parameters, and update the model parameters of the preset model according to the target model parameters to obtain the target model; The safety detection module is used to input the historical verification data set into the target model for verification to obtain the pass rate. If the pass rate is greater than the preset pass rate, the model training is determined to be completed, and the structured data is input into the target model for safety detection. Otherwise, the target model is trained again.

7. A web-based infrastructure programmatic modeling system according to claim 6, characterized in that: The task execution module is also used to perform risk assessment on the structured data to obtain a feedback report if the structured data is determined to be dangerous, and report the feedback report as a vulnerability.

Images