Dynamic particle adaptive drawing method and system based on autonomous controllable GPU

Abstract

The invention provides a dynamic particle adaptive drawing method and system based on an autonomous controllable GPU. Optical fiber sensing arrays are deployed in a monitoring area in a distributed mode, and multi-frequency-band vibration waveform data of the dynamic particle flow under strong electromagnetic interference are collected. And in combination with a particle fluid mechanics constraint rule, generating an inter-node association topology of the dynamic particle knowledge graph. And through a federated learning framework, a plurality of GPU computing nodes cooperatively update the knowledge graph, and particle rendering resolutions are dynamically distributed according to the real-time variation of the topology. And finally, based on the vibration waveform data fluctuation interval fed back in real time, generating a GPU drawing control instruction, and triggering a GPU computing node to adaptively adjust the rendering granularity of the particle trajectory according to the priority of the energy transfer path. According to the technical scheme provided by the invention, efficient and accurate drawing of the dynamic particle flow in a strong electromagnetic interference environment can be realized, and the visualization effect of the particle flow in the mechanical cavity is improved.

Description

Technical Field

[0001] This application relates to the field of optical fiber sensing technology, and particularly to a dynamic particle adaptive rendering method and system based on an independently controllable GPU. Background Art

[0002] In multi-user collaborative virtual simulation particle systems in the fields of scientific research and education, dynamic particle adaptive rendering technology is a key requirement. Such systems are used to simulate complex physical, chemical, or biological processes, such as fluid dynamics or molecular motion. Users need to observe the dynamic changes of particles in real time and adjust the rendering accuracy according to requirements to achieve efficient visual presentation under limited hardware resources. In a multi-user collaborative scenario, different users may focus on different regions or details, and the system needs to dynamically allocate computing resources to achieve particle adaptive rendering while ensuring the real-time and stability of rendering. In addition, with the popularization of independently controllable GPUs, how to make full use of their computing power has become the core of technology implementation.

[0003] Currently, existing solutions mainly adopt a hierarchical particle rendering method based on GPU acceleration. This method constructs a hierarchical structure of the particle system, divides particles into different levels of detail, and dynamically adjusts the rendering accuracy according to the user's perspective and particle density. The system uses the parallel computing power of the GPU to perform spatial partitioning and hierarchical clustering of particles, and at the same time, by real-time analyzing the user's perspective and particle motion state, dynamically allocates computing resources to preferentially render high-precision particles in the region of interest of the user, while using low-precision rendering for distant or low-importance regions. In addition, this method also introduces a GPU load balancing algorithm to ensure efficient resource allocation in a multi-user collaborative scenario.

[0004] However, the hierarchical particle rendering method still has some defects. First, it depends on fixed hierarchical partitioning rules and is difficult to adapt to complex and variable particle distributions and user requirements, which may lead to over-simplification or resource waste. Second, the implementation of the GPU load balancing algorithm is complex, and performance degradation or rendering delay is likely to occur due to resource competition during multi-user concurrent access. In addition, the existing solutions lack sufficient optimization support for independently controllable GPUs and fail to make full use of their unique hardware characteristics, which limits the further improvement of rendering efficiency. These problems may affect the system performance stability and user collaboration experience. Summary of the Invention

[0005] This application provides a dynamic particle adaptive rendering method and system based on an independently controllable GPU to solve the problem of limited dynamic particle adaptive ability in the prior art.

[0006] In a first aspect, this application provides a dynamic particle adaptive rendering method based on an independently controllable GPU, including:

[0007] Distributively deploy an optical fiber sensing array within the enclosed monitoring area of a fluid machinery, and collect multi-band vibration waveform data of a dynamic particle flow in a strong electromagnetic interference environment through the optical fiber sensing array;

[0008] Based on the multi-band vibration waveform data and a preset particle fluid mechanics constraint rule, generate an inter-node association topology of a dynamic particle knowledge graph;

[0009] Through a federated learning framework, perform distributed collaborative updates on the dynamic particle knowledge graph among multiple GPU computing nodes, and based on the real-time change amount of the inter-node association topology obtained during the distributed collaborative update process, determine the dynamic allocation strategy of the particle rendering resolution for each GPU computing node;

[0010] According to the dynamic allocation strategy of the particle rendering resolution and the vibration waveform data fluctuation range feedback by the optical fiber sensing array in real time, generate a GPU drawing control instruction for the dynamic particle flow inside the mechanical cavity, and the drawing control instruction triggers the GPU computing node to perform an adaptive adjustment of the rendering granularity of the particle trajectory based on the priority of the energy transfer path in the dynamic particle knowledge graph.

[0011] Optionally, divide the inter-node association topology of the dynamic particle knowledge graph into multiple update units, and the update unit includes particle collision feature change information and the probability of energy transfer path breakage;

[0012] Establish a data transmission channel between GPU computing nodes in the federated learning framework through a hybrid communication protocol, and adopt a local constraint collaborative mechanism to synchronize only the inter-node topology fragments that have a mechanical association with the target GPU computing node area, so as to realize the distributed collaborative update of the dynamic particle knowledge graph;

[0013] Establish a priority queue in the federated learning framework, and generate a local update weight according to the product of the probability of energy transfer path breakage and the particle collision frequency in the update unit;

[0014] When the local update weight exceeds a preset threshold, trigger a rendering resolution adjustment event of the corresponding GPU computing node based on the sorting of the priority queue;

[0015] Based on the real-time change amount of the inter-node association topology obtained during the distributed collaborative update process, and in combination with the rendering resolution adjustment event and the real-time bandwidth occupancy rate of the data transmission channel, generate a dynamic allocation strategy of the particle rendering resolution.

[0016] Optionally, in the federated learning framework, establish a data transmission channel between GPU computing nodes through a hybrid communication protocol for transmitting update data of the dynamic particle knowledge graph;

[0017] Divide the boundary of the synchronization region of the spatial locality constraint according to the radius of curvature of the energy transfer path within the rendering area of the target GPU computing node, where the boundary of the synchronization region contains an inter-node correlation topology fragment that shares a continuous energy flux vector with the rendering area of the target GPU computing node;

[0018] Through the data transmission channel, only synchronize the inter-node correlation topology fragment within the boundary of the synchronization region to achieve distributed collaborative update of the dynamic particle knowledge graph.

[0019] Optionally, identify the mutation interval of the radius of curvature of the energy transfer path based on the phase difference jump feature of adjacent sensing nodes in the multi-band vibration waveform data collected by the fiber optic sensing array;

[0020] Generate a vortex core energy density distribution map according to the mutation interval of the radius of curvature of the energy transfer path, where the vortex core energy density distribution map contains energy focusing hotspots formed by the reflection of the particle flow on the inner wall of the mechanical cavity;

[0021] During the process of dividing the boundary of the synchronization region of the spatial locality constraint, use the radial diffusion front of the energy focusing hotspot in the vortex core energy density distribution map as the dynamic adjustment baseline of the synchronization region boundary;

[0022] When it is detected that the change rate of the radius of curvature of the energy transfer path exceeds the preset critical value of fluid instability, trigger a real-time re-division event of the synchronization region boundary. The real-time re-division event preferentially retains the inter-node correlation topology fragment that shares a continuous energy flux vector with the rendering area of the target GPU computing node, and performs backward compensation on the energy conservation error of the unsynchronized region based on the position offset of the energy focusing hotspot.

[0023] Optionally, based on the spatial distribution characteristics of the particle collision frequency in the multi-band vibration waveform data collected by the fiber optic sensing array, extract the probability of energy transfer path breakage that has hydrodynamic coupling with the rendering area of the target GPU computing node in the dynamic particle knowledge graph;

[0024] Perform a time-domain convolution operation on the probability of energy transfer path breakage and the particle collision frequency in the corresponding region to generate a local topology update weight.

[0025] Optionally, identify the instantaneous spatial coordinates of the energy focusing hotspot based on the high-frequency phase difference jump feature in the multi-band vibration waveform data collected by the fiber optic sensing array;

[0026] Extract the radial diffusion front trajectory of the energy focusing hotspot in the vortex core energy density distribution map;

[0027] Generate a front expansion rate compensation factor according to the position offset of the energy focusing hot spot within a preset time window, where the compensation factor is dynamically calculated by the ratio of the fluctuation amplitude of the high-frequency phase difference jump event to the thermal expansion coefficient of the inner wall of the cavity;

[0028] During the dynamic adjustment of the boundary of the synchronization region, project the tangent direction of the radial diffusion front trajectory onto the atlas incremental transmission channel of the federated learning framework;

[0029] When it is detected that the deviation angle of the energy flux vector direction exceeds the preset turbulence anisotropy threshold, trigger a local retraction operation of the boundary of the synchronization region, and redefine the dynamic adjustment baseline of the boundary of the synchronization region based on the distribution of the curvature mutation points of the radial diffusion front trajectory.

[0030] Optionally, screen and extract the high-frequency phase difference jump event and the low-frequency energy attenuation gradient fluctuation interval from the fluctuation intervals of the multi-band vibration waveform data real-time feedback by the fiber optic sensing array;

[0031] Perform spatio-temporal matching on the screened high-frequency phase difference jump event and the priority of the energy transfer path in the dynamic particle knowledge graph to generate a spatio-temporal matching result;

[0032] Based on the dynamic allocation strategy of the particle rendering resolution, and combined with the spatio-temporal matching result, generate a dynamic resolution attenuation gradient table;

[0033] Generate a GPU rendering control instruction for the dynamic particle flow inside the mechanical cavity according to the dynamic resolution attenuation gradient table and the spatial distribution density of the high-frequency phase difference jump event obtained in real time.

[0034] In a second aspect, the present application provides a dynamic particle adaptive rendering system based on an autonomous and controllable GPU, including:

[0035] An acquisition module, which distributes and deploys a fiber optic sensing array in the closed monitoring area of the fluid machinery, and acquires multi-band vibration waveform data of the dynamic particle flow in a strong electromagnetic interference environment through the fiber optic sensing array;

[0036] A generation module, which generates the inter-node association topology of the dynamic particle knowledge graph based on the multi-band vibration waveform data and the preset particle hydrodynamics constraint rules;

[0037] A calculation module, which performs distributed collaborative update on the dynamic particle knowledge graph among multiple GPU calculation nodes through the federated learning framework, and determines the dynamic allocation strategy of the particle rendering resolution of each GPU calculation node based on the real-time change amount of the inter-node association topology obtained during the distributed collaborative update process;

[0038] An adjustment module generates GPU rendering control instructions for the dynamic particle flow inside the mechanical cavity according to the dynamic allocation strategy of the particle rendering resolution and the fluctuation range of the vibration waveform data fed back in real time by the fiber optic sensing array. The rendering control instructions trigger the GPU computing node to perform adaptive adjustment of the rendering granularity of the particle trajectory based on the priority of the energy transfer path in the dynamic particle knowledge graph.

[0039] In a third aspect, an embodiment of the present application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a dynamic particle adaptive rendering method based on a domestically controllable GPU as described in the first aspect above.

[0040] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program, which when executed by a computer, implements a dynamic particle adaptive rendering method based on a domestically controllable GPU as described in the first aspect.

[0041] In the embodiment of the present application, a fiber optic sensing array is distributedly deployed in the enclosed monitoring area of the fluid machinery, and multi-band vibration waveform data of the dynamic particle flow in a strong electromagnetic interference environment is collected through the fiber optic sensing array; based on the multi-band vibration waveform data and the preset particle hydrodynamics constraint rules, an inter-node association topology of the dynamic particle knowledge graph is generated; the dynamic particle knowledge graph among multiple GPU computing nodes is distributedly and collaboratively updated through a federated learning framework, and the dynamic allocation strategy of the particle rendering resolution of each GPU computing node is determined according to the real-time change amount of the inter-node association topology; according to the dynamic allocation strategy of the particle rendering resolution and the fluctuation range of the vibration waveform data fed back in real time by the fiber optic sensing array, GPU rendering control instructions for the dynamic particle flow inside the mechanical cavity are generated, and the rendering control instructions trigger the GPU computing node to perform adaptive adjustment of the rendering granularity of the particle trajectory based on the priority of the energy transfer path in the dynamic particle knowledge graph.

[0042] The technical solution of the present application has the following beneficial effects:

[0043] The dynamic particle adaptive rendering method based on domestically controllable GPUs in this application collects multi-band vibration waveform data in a strong electromagnetic interference environment through distributed deployment of fiber optic sensing arrays, effectively achieving high-precision monitoring of dynamic particle flows; generates the inter-node association topology of a dynamic particle knowledge graph based on the vibration waveform data and particle fluid mechanics constraint rules, providing a structured physical model support for particle flow behavior; performs distributed collaborative updates on multiple GPU computing nodes through a federated learning framework, and determines the dynamic allocation strategy of particle rendering resolution according to the real-time change amount of the inter-node association topology, ensuring the intelligent optimization and load balancing of multi-GPU computing resources; finally, generates GPU rendering control instructions by combining the fluctuation range of the vibration waveform data, and adaptively adjusts the rendering granularity of particle trajectories based on the priority of the energy transfer path, significantly improving the real-time performance and accuracy of dynamic particle rendering, while reducing the redundant consumption of computing resources, and achieving efficient and stable rendering in complex environments.

[0044] Furthermore, through the dynamic calculation of spatio-temporal increment units and local topology update weights, precise capture of particle collisions and changes in energy transfer paths is achieved; the synchronization mechanism based on spatial locality constraints and the atlas increment transmission channel significantly reduce the communication overhead between GPU computing nodes and improve data transmission efficiency; through the intelligent triggering of the topology evolution priority queue and rendering resolution reallocation events, dynamic optimization and allocation of GPU computing resources are ensured, thus achieving high-precision and high-efficiency particle rendering in complex fluid environments, while reducing system latency and resource consumption, and meeting the high-performance rendering requirements of dynamic particle flows in sealed mechanical cavities.

[0045] These aspects or other aspects of this application will be more clearly understood in the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] To more clearly illustrate the technical solutions in the embodiments of this application or in the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of this application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0047] Figure 1 Shows the flowchart of a dynamic particle adaptive rendering method based on domestically controllable GPUs provided by this application;

[0048] Figure 2 Shows the structural schematic diagram of a dynamic particle adaptive rendering system based on domestically controllable GPUs provided by this application;

[0049] Figure 3The structural schematic diagram of a computing device provided by the present application is shown. Specific embodiments

[0050] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application.

[0051] In some processes described in the specification, claims and above-mentioned drawings of the present application, a plurality of operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first" and "second" in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are of different types.

[0052] This project aims to develop a multi-user collaborative virtual simulation particle system based on an autonomous and controllable GPU, collect multi-band vibration waveform data of dynamic particle flow through a distributed fiber optic sensing array, construct a dynamic particle knowledge graph in combination with particle fluid mechanics constraint rules, use a federated learning framework to realize distributed collaborative update and dynamic allocation of rendering resolution of multiple GPU computing nodes, and finally generate GPU rendering control instructions for dynamic particle flow inside the mechanical cavity to achieve adaptive adjustment of the rendering granularity of particle trajectories. This system can efficiently and accurately complete the real-time rendering of complex dynamic particle flow in a strong electromagnetic interference environment, meeting the high-performance requirements of multi-user collaborative virtual simulation in the scientific research and education fields.

[0053] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0054] Figure 1 A flowchart of a dynamic particle adaptive rendering method based on an autonomous and controllable GPU is provided for the embodiments of the present application. As Figure 1 shown, the method includes:

[0055] 101. Distributedly deploy a fiber optic sensing array in the sealed monitoring area of the fluid machinery, and collect multi-band vibration waveform data of the dynamic particle flow in a strong electromagnetic interference environment through the fiber optic sensing array;

[0056] In this step, the fiber optic sensing array is a distributed sensor network composed of multiple optical fibers. Its core principle is to utilize the optical properties of optical fibers (such as interference, scattering, or reflection) to detect physical changes in the environment. Within the enclosed monitoring area of the fluid machinery, the fiber optic sensing array is deployed to collect multi-band vibration waveform data of the dynamic particle flow under strong electromagnetic interference.

[0057] The dynamic particle flow refers to the particles or fluid flowing inside the fluid machinery, and its motion state will change due to factors such as flow velocity, pressure, and turbulence.

[0058] The multi-band vibration waveform data refers to the vibration signals in different frequency ranges collected by the fiber optic sensing array, including low-frequency, medium-frequency, and high-frequency vibrations. These data can comprehensively reflect the motion characteristics of the dynamic particle flow.

[0059] In the embodiment of the present application, first, within the enclosed monitoring area of the fluid machinery, the fiber optic sensing array is deployed in a distributed manner to ensure coverage of the entire monitoring area. The fiber optic sensing array monitors the vibration waveform of the dynamic particle flow inside the machinery in real time through the optical properties of the optical fiber (such as Brillouin scattering or Rayleigh scattering). Then, due to the characteristics of fiber optic sensing such as anti-electromagnetic interference, high temperature resistance, and corrosion resistance, it can work stably in a strong electromagnetic environment. The collected multi-band vibration waveform data includes vibration signals of low frequency (such as mechanical vibration or water flow impact), medium frequency (such as turbulent vibration), and high frequency (such as bubble rupture or micro-particle collision). Finally, these data are transmitted to the data processing center through the optical fiber and, after signal amplification, filtering, and digital processing, form a multi-band vibration waveform data set for subsequent analysis.

[0060] Inside the turbine of a large hydropower station, a distributed fiber optic sensing array is installed. Due to the strong electromagnetic interference and high temperature and high pressure environment inside the turbine, traditional sensors cannot work stably. The fiber optic sensing array has successfully collected multi-band vibration waveform data of the water flow dynamic particle flow inside the turbine, including low-frequency water flow impact vibration (0 - 100 Hz), medium-frequency turbulent vibration (100 - 1000 Hz), and high-frequency bubble rupture vibration (1000 - 10000 Hz). These data provide a solid foundation for subsequent analysis and help engineers better understand the hydrodynamic characteristics inside the turbine.

[0061] 102. Generate the inter-node association topology of the dynamic particle knowledge graph based on the multi-band vibration waveform data and the preset particle fluid mechanics constraint rules;

[0062] In this step, the dynamic particle knowledge graph is a graphical representation based on multi-band vibration waveform data and particle hydrodynamics constraint rules, used to describe the associated topology between nodes (such as particles, fluid regions, or energy transfer paths) in the dynamic particle flow.

[0063] The particle hydrodynamics constraint rules are rules set according to hydrodynamics principles (such as the continuity equation, the momentum conservation equation, and the energy conservation equation), used to guide the generation of the knowledge graph.

[0064] The nodes of the knowledge graph represent key elements (such as fluid regions or particle groups) in the dynamic particle flow, and the edges represent the interaction relationships between these elements.

[0065] In the embodiments of this application, first, the collected multi-band vibration waveform data is preprocessed, including filtering, denoising, and feature extraction. Then, based on the particle hydrodynamics constraint rules, the behavior patterns of the dynamic particle flow are analyzed. For example, the mass conservation characteristics of the fluid are analyzed through the continuity equation, the momentum transfer characteristics of the fluid are analyzed through the momentum conservation equation, and the energy distribution characteristics of the fluid are analyzed through the energy conservation equation. Next, graph theory algorithms are used to represent the nodes and their associated relationships in the dynamic particle flow as the nodes and edges of the knowledge graph. For example, the nodes can represent fluid regions or particle groups, and the edges can represent energy transfer paths or momentum exchange relationships. Finally, the associated topology between the nodes of the dynamic particle knowledge graph is generated, clearly showing the structure and interaction relationships of the dynamic particle flow.

[0066] Inside the turbine, based on the collected multi-band vibration waveform data, the system analyzes the dynamic behavior of the water flow through the particle hydrodynamics constraint rules. For example, the low-frequency vibration data is used to identify the overall flow direction of the water flow, the mid-frequency vibration data is used to identify the turbulent regions, and the high-frequency vibration data is used to identify the bubble distribution. Through the graph neural network, the associated topology of the dynamic particle flow inside the turbine is generated, clearly showing the interaction relationships between the water flow, the turbulence, and the bubbles. For example, a certain node in the knowledge graph represents the fluid region of the turbine blade, and the edge represents the energy transfer path between the water flow and the blade.

[0067] 103. Perform distributed collaborative update on the dynamic particle knowledge graph between multiple GPU computing nodes through the federated learning framework, and determine the dynamic allocation strategy of the particle rendering resolution for each GPU computing node based on the real-time change amount of the associated topology obtained during the distributed collaborative update process;

[0068] In this step, the federated learning framework is a distributed machine learning method that allows multiple GPU computing nodes to collaboratively update the model without sharing the original data.

[0069] The distributed collaborative update of the dynamic particle knowledge graph means that multiple GPU computing nodes perform local updates on the knowledge graph based on local data and conduct global integration through the federated learning framework.

[0070] The dynamic allocation strategy for particle rendering resolution is to allocate different rendering resolutions for each GPU computing node according to the real-time change amount of the inter-node association topology to optimize computing resources. For example, a high rendering resolution is allocated to areas with high dynamic changes, and a low rendering resolution is allocated to areas with low dynamic changes.

[0071] In the embodiments of this application, first, each GPU computing node performs local updates on the dynamic particle knowledge graph based on local data, including the adjustment of the node association topology and the optimization of the energy transfer path. For example, a certain GPU computing node is responsible for updating the association topology of the turbine blade area, and another GPU computing node is responsible for updating the association topology of the water flow area. Then, through the federated learning framework, the local update results of each node are integrated to generate a global dynamic particle knowledge graph. For example, the update results of each node are fused into a global knowledge graph using methods such as weighted average or gradient aggregation. Next, based on the real-time change amount of the inter-node association topology, the system allocates different particle rendering resolutions for each GPU computing node. For example, a high rendering resolution is allocated to the turbulent area with high dynamic changes, and a low rendering resolution is allocated to the stable area with low dynamic changes. Finally, the dynamic allocation strategy for the particle rendering resolution of each GPU computing node is determined, optimizing the use of computing resources.

[0072] In the turbine monitoring system, multiple GPU computing nodes are respectively responsible for updating the dynamic particle knowledge graph of different areas. Through the federated learning framework, the update results of each node are integrated into a global knowledge graph. For example, when the water flow speed inside the turbine suddenly increases, the system detects the change in the inter-node association topology and dynamically adjusts the rendering resolution of the GPU computing node: the rendering resolution of the turbulent area is increased (such as 1024x1024 pixels), while the rendering resolution of the stable area is decreased (such as 512x512 pixels), thereby optimizing the use of computing resources.

[0073] 104. Generate a GPU drawing control instruction for the dynamic particle flow inside the mechanical cavity according to the dynamic allocation strategy of the particle rendering resolution and the vibration waveform data fluctuation range feedback in real time by the fiber optic sensing array. The drawing control instruction triggers the GPU computing node to perform an adaptive adjustment of the rendering granularity of the particle trajectory based on the priority of the energy transfer path in the dynamic particle knowledge graph.

[0074] In this step, the GPU rendering control instructions are generated based on the dynamic allocation strategy of particle rendering resolution and the fluctuation range of the vibration waveform data feedback in real time by the fiber optic sensing array, and are used to control the GPU computing nodes to render the dynamic particle flow.

[0075] The priority of the energy transfer path refers to the importance of the influence of different paths in the knowledge graph on the overall dynamic particle flow. The adaptive adjustment of the rendering granularity means dynamically adjusting the rendering accuracy of the particle trajectory according to the priority. For example, high-precision rendering is performed on the energy transfer paths with high priority, and low-precision rendering is performed on the paths with low priority.

[0076] In the embodiments of the present application, first, according to the fluctuation range of the vibration waveform data feedback in real time by the fiber optic sensing array, the state of the current dynamic particle flow is determined. Then, based on the dynamic allocation strategy of particle rendering resolution, specific rendering control instructions are generated for each GPU computing node. These instructions include the number of particles to be rendered, the resolution, and the rendering order. For example, high-precision rendering is performed on the particle trajectories in the turbulent region to capture details; while low-precision rendering is performed on the particle trajectories in the stable region to save computing resources. Finally, the GPU computing nodes adaptively adjust the rendering granularity of the particle trajectories according to the priority of the energy transfer paths in the dynamic particle knowledge graph. For example, high-precision rendering is performed on the energy transfer paths with high priority, and low-precision rendering is performed on the paths with low priority.

[0077] In the turbine monitoring system, when it is detected that the water flow velocity suddenly increases, the system generates GPU rendering control instructions according to the real-time vibration waveform data fluctuation range. For example, high-precision rendering is performed on the particle trajectories in the turbulent region (such as rendering 10,000 particles per frame) to capture details; while low-precision rendering is performed on the particle trajectories in the stable region (such as rendering 1,000 particles per frame) to save computing resources. Finally, the GPU computing nodes complete the efficient rendering of the dynamic particle flow inside the turbine according to the priority of the energy transfer paths, providing clear visualization results for engineers.

[0078] In summary, steps 101 to 104 achieved the distributed deployment of an optical fiber sensing array within the enclosed monitoring area of the fluid machinery. The system successfully collected multi-band vibration waveform data of the dynamic particle flow in a strong electromagnetic interference environment. Based on this data and the particle fluid mechanics constraint rules, the inter-node association topology of the dynamic particle knowledge graph was generated, clearly showing the structure and interaction relationships of the dynamic particle flow. Through the federated learning framework, multiple GPU computing nodes collaborated to update the knowledge graph, and the particle rendering resolution was dynamically allocated according to the real-time change amount of the inter-node association topology, optimizing the use of computing resources. Finally, based on the rendering resolution strategy and the fluctuation range of the real-time vibration waveform data, GPU drawing control instructions were generated, realizing the efficient rendering of the dynamic particle flow inside the mechanical cavity. This solution significantly improved the accuracy and efficiency of dynamic particle flow monitoring and rendering in a strong electromagnetic interference environment, providing strong support for the intelligent monitoring and fault diagnosis of fluid machinery.

[0079] To solve the problems of uneven resource allocation, large communication overhead, and difficulty in balancing rendering accuracy and efficiency among multiple GPU computing nodes in dynamic particle flow simulation, a distributed collaborative update system based on the federated learning framework was developed. This system achieved efficient collaboration of multiple GPUs and a significant improvement in rendering performance by intelligently allocating computing resources, optimizing the communication mechanism, and dynamically adjusting the rendering resolution, providing high-performance support for complex dynamic particle flow simulation. In some embodiments, in step 103, the distributed collaborative update of the dynamic particle knowledge graph among multiple GPU computing nodes is performed through the federated learning framework, and based on the real-time change amount of the inter-node association topology obtained during the distributed collaborative update process, the dynamic allocation strategy of the particle rendering resolution for each GPU computing node is determined, including:

[0080] 201. Divide the inter-node association topology of the dynamic particle knowledge graph into spatio-temporal increment units, where the spatio-temporal increment units contain particle collision mode mutation features and the probability of energy transfer path breakage;

[0081] In step 201, the dynamic particle knowledge graph is a graph model used to describe the interaction and energy transfer relationships among particles. Nodes represent particles, and edges represent the association relationships among particles. The spatio-temporal increment unit is the smallest unit obtained by dividing the graph in the time and space dimensions, used to describe the particle behavior and energy transfer characteristics in a local area. The particle collision mode mutation feature is a feature that describes a significant change in the particle collision behavior, such as a mutation in the collision frequency or angle. The probability of energy transfer path breakage is the possibility of the interruption of the energy transfer path among particles, calculated based on the distance and interaction strength among particles.

[0082] In the embodiments of the present application, first, the dynamic particle knowledge graph is segmented in terms of space and time, and the graph is divided into multiple spatio-temporal incremental units. Each spatio-temporal incremental unit describes the particle behavior in the local area by analyzing the change characteristics of the particle collision pattern (such as the mutation of collision frequency and angle) and the break probability of the energy transfer path (calculated based on the distance and interaction strength between particles). Then, through this segmentation method, it is possible to focus on the dynamic changes in the local area, providing a basis for subsequent incremental processing and rendering optimization. In addition, the segmentation of the spatio-temporal incremental unit also considers the continuity of the time dimension, ensuring that the evolution process of particle behavior can be captured in the time series, thereby providing more accurate data support for subsequent real-time updates and rendering.

[0083] 202. Establish a graph incremental transmission channel between GPU computing nodes in the federated learning framework through a hybrid communication protocol, and adopt a collaborative mechanism for particle subsets with spatial locality constraints to synchronize only the associated topological fragments between nodes that have a hydrodynamic coupling relationship with the rendering area of the target GPU computing node, so as to achieve distributed collaborative update of the dynamic particle knowledge graph;

[0084] In step 202, the hybrid communication protocol is a protocol that combines multiple communication technologies (such as TCP / IP and RDMA) and is used for efficient data transmission. The graph incremental transmission channel is a communication channel used to transmit spatio-temporal incremental units between GPU computing nodes. The spatial locality constraint is a constraint condition based on physical space proximity and is used to screen the particle subsets that need to be synchronized. The hydrodynamic coupling relationship describes the mutual influence relationship between particles under hydrodynamic action.

[0085] In the embodiments of the present application, first, a graph incremental transmission channel between GPU computing nodes is established in the federated learning framework through a hybrid communication protocol. Then, a particle subset synchronization mechanism with spatial locality constraints is adopted to screen out the associated topological fragments between nodes that have a hydrodynamic coupling relationship with the rendering area of the target GPU computing node, and only synchronize these fragments. This mechanism reduces unnecessary data transmission, improves communication efficiency, and at the same time ensures the accuracy of the rendering area. In addition, the selection of the hybrid communication protocol also considers the dynamic changes in the network environment and can dynamically adjust the transmission strategy according to the real-time network bandwidth and latency conditions, further optimizing the communication performance.

[0086] 203. Establish a topological evolution priority queue for the dynamic particle knowledge graph in the federated learning framework, and generate a local topological update weight according to the product of the break probability of the energy transfer path and the particle collision frequency in the spatio-temporal incremental unit;

[0087] In step 203, the topology evolution priority queue is a queue structure used to manage the priority of local topology updates of the dynamic particle knowledge graph. The local topology update weight is an index generated by the product of the energy transfer path break probability and the particle collision frequency, and is used to measure the significance of local topology changes. The energy transfer path break probability refers to the probability that the particle energy transfer path breaks in the spatio-temporal increment unit, and is used to measure the stability of the topology structure. The particle collision frequency refers to the frequency of particle collisions in the spatio-temporal increment unit, and is used to measure the intensity of particle activity. The local topology update weight refers to the weight generated according to the product of the energy transfer path break probability and the particle collision frequency, and is used to guide the update of the local topology.

[0088] In the embodiments of the present application, first, a topology evolution priority queue of the dynamic particle knowledge graph is established in the federated learning framework. The establishment of the topology evolution priority queue usually adopts a priority sorting algorithm. For example, by using the local topology update weight as the basis for priority sorting, a priority queue is generated. Then, the local topology update weight is generated according to the product of the energy transfer path break probability and the particle collision frequency in the spatio-temporal increment unit. The calculation of the local topology update weight usually adopts a weighted product algorithm. For example, by multiplying the energy transfer path break probability and the particle collision frequency, a weight value is generated. Through this step, the system can scientifically quantify the priority of local topology updates and provide a basis for subsequent rendering optimization.

[0089] 204. When the local topology update weight exceeds a preset fluid turbulence threshold, activate the rendering resolution reallocation event of the corresponding GPU computing node based on the sorting of the topology evolution priority queue;

[0090] In step 204, the fluid turbulence threshold is a preset threshold used to determine whether to trigger a rendering resolution reallocation event. The fluid turbulence threshold refers to a preset threshold used to determine the intensity of fluid turbulence and trigger a rendering resolution reallocation event. The topology evolution priority queue refers to a priority queue generated according to the local topology update weight and is used to guide the order of topology updates. The GPU computing node refers to a GPU computing unit used to execute rendering tasks and is used to achieve rendering optimization. The rendering resolution reallocation event refers to an event of reallocating the rendering resolution according to the local topology update weight and is used to optimize the rendering quality.

[0091] In the embodiments of the present application, first, it is determined whether the local topology update weight exceeds a preset fluid turbulence threshold. The setting of the fluid turbulence threshold is usually based on a fluid dynamics model or experimental data. For example, by analyzing the impact of fluid turbulence on the rendering quality, a reasonable threshold is set. When the local topology update weight exceeds the fluid turbulence threshold, a rendering resolution reallocation event for the corresponding GPU computing node is activated based on the sorting of the topology evolution priority queue. The rendering resolution reallocation event usually adopts a dynamic resolution adjustment algorithm. For example, by increasing the rendering resolution of high-priority regions and decreasing the rendering resolution of low-priority regions, the rendering quality and system performance are optimized. Through this step, the system can optimize the rendering resolution in real time to ensure the maximization of the rendering quality and system performance.

[0092] 205. Generate a dynamic allocation strategy for the particle rendering resolution for the enclosed mechanical cavity based on the rendering resolution reallocation event and the real-time bandwidth occupancy rate of the atlas incremental transmission channel.

[0093] In step 205, the rendering resolution reallocation event is an event that triggers the GPU computing node to adjust the rendering resolution. The real-time bandwidth occupancy rate of the atlas incremental transmission channel is an indicator that describes the bandwidth usage of the current communication channel. The enclosed mechanical cavity refers to a closed physical space used to simulate or analyze particle behavior.

[0094] In the embodiments of the present application, first, according to the rendering resolution reallocation event and the real-time bandwidth occupancy rate of the atlas incremental transmission channel, the particle rendering resolution of each region in the enclosed mechanical cavity is dynamically adjusted. When the communication bandwidth is sufficient, the rendering resolution of high-dynamic change regions is increased; when the bandwidth is limited, the rendering accuracy of key regions is preferentially guaranteed. Finally, through this dynamic allocation strategy, while ensuring the rendering effect, the resource utilization rate can be optimized. In addition, the dynamic allocation strategy also takes into account the physical property differences of different regions, and allocates a higher rendering resolution to high-dynamic change regions, thereby further improving the accuracy of the simulation results.

[0095] The following is a specific example:

[0096] In a hydrodynamic simulation scenario within an aero-engine combustion chamber, the dynamic particle knowledge graph is first segmented into multiple spatio-temporal incremental units, and the particle collision patterns and energy transfer paths within each unit are analyzed. For example, in the high-temperature and high-pressure regions of the combustion chamber, the particle collision frequency increases significantly, and the probability of energy transfer path breakage is also relatively high. These regions are divided into spatio-temporal incremental units with high dynamic changes. Then, a graph incremental transmission channel is established between GPU computing nodes through a hybrid communication protocol, and a particle subset synchronization mechanism with spatial locality constraints is adopted to synchronize only the associated topological fragments between nodes that have a hydrodynamic coupling relationship with the rendering area of the target GPU computing node. For example, in the flame propagation region of the combustion chamber, only the particle subset related to flame propagation is synchronized, reducing unnecessary data transmission. Next, a topological evolution priority queue is established, and a local topological update weight is generated based on the product of the energy transfer path breakage probability and the particle collision frequency. For example, in the flame propagation region of the combustion chamber, since both the particle collision frequency and the energy transfer path breakage probability are relatively high, the local topological update weight exceeds the preset fluid turbulence threshold, triggering a rendering resolution reallocation event. Finally, based on the rendering resolution reallocation event and the real-time bandwidth occupancy rate of the graph incremental transmission channel, the particle rendering resolution of each region within the combustion chamber is dynamically adjusted. For example, the rendering resolution is increased in the flame propagation region to ensure accurate simulation of the flame shape and propagation process, while the rendering resolution is appropriately decreased in other regions to optimize resource utilization.

[0097] In summary, through steps 201 to 205, the high efficiency and high precision of particle behavior simulation within a sealed mechanical cavity are achieved. Meanwhile, resource utilization and communication efficiency are optimized, significantly improving data transmission efficiency and rendering area accuracy, providing reliable technical support for the simulation of complex physical scenarios. This method also considers the continuity of the time dimension and the physical property differences in different regions, further enhancing the accuracy and real-time performance of the simulation results, providing reliable technical support for the simulation of complex physical scenarios.

[0098] To address the problems of low data transmission efficiency between multiple GPU computing nodes, high transmission delay of emergency incremental units, and unreasonable bandwidth resource allocation, a graph incremental transmission system based on a hybrid communication protocol is developed. This system realizes high-priority transmission of emergency incremental units, intelligent allocation of bandwidth resources, and significant improvement in data transmission efficiency, providing reliable support for the efficient cooperation of multiple GPU computing nodes. In some embodiments, establishing a graph incremental transmission channel between GPU computing nodes in the federated learning framework through a hybrid communication protocol and adopting a particle subset synchronization mechanism with spatial locality constraints to synchronize only the associated topological fragments between nodes that have a hydrodynamic coupling relationship with the rendering area of the target GPU computing node in step 203 includes:

[0099] 301. In the federated learning framework, a data transmission channel is established between GPU computing nodes through a hybrid communication protocol for transmitting updated data of the dynamic particle knowledge graph.

[0100] In step 301, the federated learning framework refers to a framework that supports distributed computing and collaborative learning for updating and optimizing the dynamic particle knowledge graph. The hybrid communication protocol refers to a protocol that combines multiple communication technologies for efficient data transmission. The GPU computing node refers to a GPU computing unit used to execute rendering tasks for rendering optimization. The data transmission channel refers to a channel for transmitting data between GPU computing nodes to support distributed collaborative updates. The dynamic particle knowledge graph refers to a knowledge graph that describes the behavior and relationships of particles in spatio-temporal incremental units for guiding rendering optimization. The updated data refers to the updated information of the dynamic particle knowledge graph for synchronizing and optimizing the knowledge graph.

[0101] In the embodiment of the present application, first, in the federated learning framework, a data transmission channel is established between GPU computing nodes through a hybrid communication protocol. The hybrid communication protocol usually combines high-speed network communication and low-latency communication technologies. For example, by combining Ethernet and Infiniband technologies, efficient data transmission is achieved. Then, the updated data of the dynamic particle knowledge graph is transmitted using the data transmission channel. The data transmission process usually adopts data compression and block transmission technologies. For example, by compressing the updated data and transmitting it in blocks, the transmission efficiency and reliability are improved. Through this step, the system can efficiently establish a data transmission channel to support subsequent distributed collaborative updates.

[0102] 302. According to the curvature radius of the energy transfer path within the rendering area of the target GPU computing node, the synchronization area boundary of the spatial locality constraint is divided. The synchronization area boundary contains an inter-node association topology fragment that shares a continuous energy flux vector with the rendering area of the target GPU computing node.

[0103] In step 302, the target GPU computing node refers to the GPU computing unit currently executing the rendering task for determining the synchronization area boundary. The rendering area refers to the area responsible for rendering by the target GPU computing node for dividing the synchronization area boundary. The curvature radius of the energy transfer path refers to the curvature radius of the energy transfer path for measuring the degree of bending of the energy transfer path. The spatial locality constraint refers to a constraint condition divided based on spatial locality for determining the synchronization area boundary. The synchronization area boundary refers to the boundary divided according to the spatial locality constraint for defining the synchronization range. The continuous energy flux vector refers to the continuous energy flux vector in the energy transfer path for determining the inter-node association topology fragment. The inter-node association topology fragment refers to the topology fragment that shares a continuous energy flux vector with the rendering area of the target GPU computing node for synchronizing and updating the knowledge graph.

[0104] In the embodiments of the present application, first, the curvature radius of the energy transfer path within the rendering area of the target GPU computing node is calculated to divide the boundary of the synchronization area with spatial locality constraints. The division of the synchronization area boundary usually adopts the curvature radius threshold method. For example, by setting a curvature radius threshold, the area with a curvature radius smaller than the threshold is divided into the synchronization area boundary. Then, the inter-node association topology fragment that shares a continuous energy flux vector with the rendering area of the target GPU computing node within the synchronization area boundary is determined. The determination of the inter-node association topology fragment usually adopts the energy flux vector matching algorithm. For example, by matching the continuous energy flux vectors, the association topology fragment is determined. Through this step, the system can scientifically divide the synchronization area boundary, providing a basis for subsequent distributed collaborative updates.

[0105] 303. Through the data transmission channel, only synchronize the inter-node association topology fragments within the synchronization area boundary to achieve the distributed collaborative update of the dynamic particle knowledge graph.

[0106] In step 303, the data transmission channel refers to the channel used to transmit data between GPU computing nodes, which is used to support distributed collaborative updates. The synchronization area boundary refers to the boundary divided according to spatial locality constraints, which is used to define the synchronization range. The inter-node association topology fragment refers to the topology fragment that shares a continuous energy flux vector with the rendering area of the target GPU computing node, which is used to synchronize and update the knowledge graph. The distributed collaborative update refers to the collaborative update of the dynamic particle knowledge graph through distributed computing, which is used to optimize the knowledge graph.

[0107] In the embodiments of the present application, first, through the data transmission channel, only synchronize the inter-node association topology fragments within the area boundary. The synchronization process usually adopts the incremental synchronization technology. For example, by only transmitting the updated data, the amount of synchronized data is reduced. Then, the distributed collaborative update of the dynamic particle knowledge graph is realized. The distributed collaborative update usually adopts the consistency protocol or the distributed lock mechanism. For example, by ensuring the consistency of the data of each node, the collaborative update is realized. Through this step, the system can efficiently realize the distributed collaborative update of the dynamic particle knowledge graph, ensuring the accuracy and real-time performance of the knowledge graph.

[0108] The following is a specific example:

[0109] In a fluid monitoring system inside a turbine engine, the system first establishes a data transmission channel between GPU computing nodes in a federated learning framework through a hybrid communication protocol. For example, the TCP / IP protocol and RDMA technology are adopted to ensure the efficient transmission of updated data of the dynamic particle knowledge graph between GPU computing nodes. For example, the change in the energy transfer path in the high-frequency turbulent region is transmitted in real time to the target GPU node. Then, the system divides the boundary of the synchronous region with spatial locality constraints according to the curvature radius of the energy transfer path within the rendering region of the target GPU computing node. For example, when the curvature radius is less than 5 mm, the rendering region of the target GPU node and its adjacent regions are divided into the boundary of the synchronous region. This boundary contains the inter-node association topology segments that share continuous energy flux vectors with the rendering region of the target GPU node, such as the energy transfer paths between the high-frequency turbulent region and the low-frequency eddy region. Subsequently, the system synchronizes only the inter-node association topology segments within the boundary of the synchronous region through the data transmission channel. For example, only the data on the change in the energy transfer path in the high-frequency turbulent region is transmitted, reducing unnecessary data exchange and achieving the distributed collaborative update of the dynamic particle knowledge graph. Through this mechanism, the system can accurately monitor the fluid dynamics inside the turbine engine and provide reliable data support for mechanical performance optimization.

[0110] In summary, by dynamically adjusting the bandwidth allocation ratio of the map increment transmission channel through steps 301 to 304, the transmission requirements of high-priority data are preferentially met in case of emergency, while ensuring the basic transmission requirements of non-emergency data, further optimizing the resource utilization rate and communication efficiency. Finally, while ensuring the accuracy and real-time performance of the simulation results, this method significantly improves the overall performance and reliability of complex physical scene simulation.

[0111] To solve the problems of inaccurate division of the synchronous region boundary and cumulative energy conservation error caused by changes in the energy transfer path within the rendering regions of multiple GPU computing nodes, a dynamic adjustment system for the synchronous region boundary based on the curvature radius of the energy transfer path is developed. This system realizes the accurate division of the synchronous region boundary, the real-time tracking of the energy transfer path, and the effective control of the energy conservation error, providing a reliable guarantee for the efficient collaborative rendering of multiple GPU computing nodes. In some embodiments, in step 303, establishing a topological evolution priority queue of the dynamic particle knowledge graph in the federated learning framework, generating a local topology update weight according to the product of the energy transfer path break probability and the particle collision frequency in the spatio-temporal increment unit, and when the local topology update weight exceeds a preset fluid turbulence threshold, activating a rendering resolution reallocation event of the corresponding GPU computing node based on the sorting of the topological evolution priority queue includes:

[0112] 401. Identify the mutation interval of the curvature radius of the energy transfer path based on the phase difference jump characteristics between adjacent sensing nodes in the multi-band vibration waveform data collected by the fiber optic sensing array.

[0113] In step 401, the fiber optic sensing array refers to an array composed of multiple fiber optic sensors for collecting multi-band vibration waveform data. The multi-band vibration waveform data refers to the vibration waveform data collected by the fiber optic sensing array, which contains vibration information in multiple frequency bands. Adjacent sensing nodes refer to adjacent sensor nodes in the fiber optic sensing array for detecting the phase difference of the vibration waveform data. The phase difference jump characteristic refers to the phase difference jump characteristic of the vibration waveform data between adjacent sensing nodes, which is used to identify the mutation interval of the curvature radius of the energy transfer path. The curvature radius of the energy transfer path refers to the curvature radius of the energy transfer path, which is used to measure the degree of bending of the energy transfer path. The mutation interval refers to the interval where the curvature radius of the energy transfer path undergoes a mutation, which is used to guide the generation of the vortex core energy density distribution map.

[0114] In the embodiment of the present application, first, based on the multi-band vibration waveform data collected by the fiber optic sensing array, the phase difference jump characteristics between adjacent sensing nodes are extracted. The extraction of the phase difference jump characteristics usually adopts signal processing techniques, such as analyzing the phase change of the vibration waveform data through Fourier transform or wavelet transform. Then, according to the phase difference jump characteristics, the mutation interval of the curvature radius of the energy transfer path is identified. The identification of the mutation interval usually adopts threshold detection or mutation point detection algorithms, such as setting a phase difference jump threshold to identify the mutation interval. Through this step, the system can scientifically identify the mutation interval of the curvature radius of the energy transfer path, providing a basis for the subsequent generation of the vortex core energy density distribution map.

[0115] 402. Generate a vortex core energy density distribution map according to the mutation interval of the curvature radius of the energy transfer path. The vortex core energy density distribution map includes energy focusing hotspots formed by the reflection of the particle flow on the inner wall of the mechanical cavity.

[0116] In step 402, the mutation interval of the curvature radius of the energy transfer path refers to the interval where the curvature radius of the energy transfer path undergoes a mutation, which is used to guide the generation of the vortex core energy density distribution map. The vortex core energy density distribution map refers to a graph describing the energy density distribution of the vortex core, which is used to guide the dynamic adjustment of the boundary of the synchronization region. The particle flow refers to the particle flow in the mechanical cavity, which is used to form energy focusing hotspots. The inner wall of the mechanical cavity refers to the inner wall of the mechanical cavity, which is used to reflect the particle flow. The energy focusing hotspot refers to the energy focusing area formed by the reflection of the particle flow on the inner wall of the mechanical cavity, which is used to guide the dynamic adjustment of the boundary of the synchronization region.

[0117] In the embodiments of the present application, first, according to the mutation interval of the curvature radius of the energy transfer path, a vortex core energy density distribution map is generated. The generation of the vortex core energy density distribution map generally adopts energy density calculation and visualization techniques. For example, by calculating the energy density in the vortex core region and generating a distribution map. Then, in the vortex core energy density distribution map, energy focusing hotspots formed by the reflection of the particle flow on the inner wall of the mechanical cavity are marked. The marking of the energy focusing hotspots generally adopts peak detection or clustering analysis algorithms. For example, by detecting the peak of the energy density, the hotspot region is marked. Through this step, the system can scientifically generate the vortex core energy density distribution map, providing a basis for the subsequent dynamic adjustment of the synchronous region boundary.

[0118] 403. During the process of dividing the synchronous region boundary under the spatial locality constraint, the radial diffusion front of the energy focusing hotspots in the vortex core energy density distribution map is used as the dynamic adjustment baseline for the synchronous region boundary;

[0119] In step 403, the vortex core energy density distribution map refers to a graph describing the vortex core energy density distribution, which is used to guide the dynamic adjustment of the synchronous region boundary. The energy focusing hotspot refers to the energy focusing region formed by the reflection of the particle flow on the inner wall of the mechanical cavity, which is used to guide the dynamic adjustment of the synchronous region boundary. The radial diffusion front refers to the radial diffusion front of the energy focusing hotspot, which is used as the dynamic adjustment baseline for the synchronous region boundary. The synchronous region boundary refers to the boundary divided according to the spatial locality constraint, which is used to define the synchronous range. The dynamic adjustment baseline refers to the baseline used to dynamically adjust the synchronous region boundary, which is used to guide the adjustment of the boundary.

[0120] In the embodiments of the present application, first, during the process of dividing the synchronous region boundary under the spatial locality constraint, the radial diffusion front of the energy focusing hotspots in the vortex core energy density distribution map is used as the dynamic adjustment baseline for the synchronous region boundary. The determination of the dynamic adjustment baseline generally adopts a front detection or boundary tracking algorithm. For example, by detecting the diffusion front of the energy focusing hotspot, the baseline is determined. Then, according to the dynamic adjustment baseline, the synchronous region boundary is dynamically adjusted. The adjustment of the synchronous region boundary generally adopts a boundary expansion or contraction algorithm. For example, by expanding or contracting the boundary to adapt to the change of the energy focusing hotspot. Through this step, the system can scientifically adjust the synchronous region boundary to ensure the rationality and real-time nature of the synchronous range.

[0121] 404. When it is detected that the change rate of the curvature radius of the energy transfer path exceeds the preset critical value of fluid instability, a real-time re-partitioning event of the synchronous region boundary is triggered. The real-time re-partitioning event preferentially retains the node-to-node association topology segments that share a continuous energy flux vector with the rendering region of the target GPU computing node.

[0122] In step 404, the rate of change of the radius of curvature of the energy transfer path refers to the rate at which the radius of curvature of the energy transfer path changes, and is used to determine the fluid instability state. The fluid instability critical value refers to a preset threshold used to determine the fluid instability state and trigger a real-time re-partitioning event of the synchronization region boundary. The synchronization region boundary refers to the boundary divided according to spatial locality constraints and is used to define the synchronization range. The real-time re-partitioning event refers to an event of re-partitioning the synchronization region boundary in real time according to the rate of change of the radius of curvature of the energy transfer path, and is used to optimize the synchronization range. The target GPU computing node refers to the GPU computing unit currently executing the rendering task and is used to determine the synchronization region boundary. The continuous energy flux vector refers to the continuous energy flux vector in the energy transfer path and is used to determine the inter-node correlation topology segment. The inter-node correlation topology segment refers to the topology segment that shares the continuous energy flux vector with the rendering region of the target GPU computing node and is used to synchronize and update the knowledge graph.

[0123] In the embodiment of the present application, first, the rate of change of the radius of curvature of the energy transfer path is detected to determine whether it exceeds the preset fluid instability critical value. The setting of the fluid instability critical value is usually based on a fluid dynamics model or experimental data. For example, by analyzing the impact of fluid instability on the synchronization range, a reasonable threshold is set. When it is detected that the rate of change of the radius of curvature of the energy transfer path exceeds the fluid instability critical value, a real-time re-partitioning event of the synchronization region boundary is triggered. The real-time re-partitioning event usually adopts a dynamic re-partitioning algorithm. For example, by re-partitioning the synchronization region boundary, it adapts to the change of the energy transfer path. Then, in the real-time re-partitioning event, the inter-node correlation topology segments that share the continuous energy flux vector with the rendering region of the target GPU computing node are preferentially retained. The retention of the inter-node correlation topology segments usually adopts a topology matching or energy flux vector matching algorithm. For example, by matching the continuous energy flux vector, the associated topology segments are retained. Through this step, the system can optimize the synchronization region boundary in real time to ensure the rationality and real-time nature of the synchronization range.

[0124] The following is a specific embodiment:

[0125] In the internal fluid monitoring system of a turbine engine, the system first identifies the mutation interval of the curvature radius of the energy transfer path based on the phase difference jump characteristics of adjacent sensing nodes in the multi-band vibration waveform data collected by the fiber optic sensing array. For example, when the phase difference between adjacent sensing nodes suddenly increases by more than 30 degrees, the system identifies the mutation interval of the curvature radius in the high-frequency turbulence region. Then, the system generates a vortex core energy density distribution map according to the mutation interval of the curvature radius of the energy transfer path. For example, the figure shows the energy focusing hot spots formed by the reflection of the particle flow on the inner wall of the mechanical cavity, and these hot spots are usually located in the edge region of the turbine blade. Subsequently, during the process of dividing the synchronous region boundary under spatial locality constraints, the system uses the radial diffusion front of the energy focusing hot spots in the vortex core energy density distribution map as the dynamic adjustment baseline of the synchronous region boundary. For example, expanding 5 millimeters outward with the hot spot as the center is used as the synchronous region boundary. Finally, when the change rate of the curvature radius of the energy transfer path is detected to exceed the preset critical value of fluid instability, for example, the change rate of the curvature radius exceeds 2 millimeters per second, the system triggers a real-time redivision event of the synchronous region boundary. For example, priority is given to retaining the inter-node association topology fragments that share a continuous energy flux vector with the rendering region of the target GPU computing node, such as the energy transfer path in the high-frequency turbulence region. Through this mechanism, the system can accurately monitor the internal fluid dynamics of the turbine engine and provide reliable data support for mechanical performance optimization.

[0126] In summary, the high efficiency and high precision of particle behavior simulation in a sealed mechanical cavity are achieved through steps 401 to 404. At the same time, the resource utilization rate and communication efficiency are optimized, providing reliable technical support for the simulation of complex physical scenarios. By combining the multi-band vibration waveform data of the fiber optic sensing array and the dynamic particle knowledge graph, the fracture probability of the energy transfer path can be accurately quantified, providing a reliable basis for local topology update. In addition, by introducing multi-band spectrum analysis, adaptive convolution kernels, multi-scale spatial sorting algorithms, and dynamic phase correction mechanisms, the robustness and adaptability of the method are further improved, significantly enhancing the overall performance and reliability of complex physical scenario simulation.

[0127] To solve the problems of untimely local topology update in the dynamic particle knowledge graph and unreasonable rendering resolution allocation of GPU computing nodes, a rendering resolution reallocation system based on a topology evolution priority queue is developed. This system realizes the precise triggering of local topology update, the dynamic optimization of rendering resolution, and the high synchronization of system timing, providing reliable guarantee for the efficient collaborative rendering of multi-GPU computing nodes. In some embodiments, establishing a priority queue in the federated learning framework in step 203 and generating a local update weight according to the product of the fracture probability of the energy transfer path and the particle collision frequency in the update unit includes:

[0128] 501. Based on the spatial distribution characteristics of the particle collision frequency in the multi-band vibration waveform data collected by the fiber optic sensing array, extract the probability of the energy transfer path breakage that has hydrodynamic coupling with the rendering area of the target GPU computing node in the dynamic particle knowledge graph;

[0129] In step 501, the fiber optic sensing array refers to an array composed of multiple fiber optic sensors, which is used to collect multi-band vibration waveform data. The multi-band vibration waveform data refers to the vibration waveform data collected by the fiber optic sensing array, which contains vibration information of multiple frequency bands. The particle collision frequency refers to the frequency of particle collisions in the spatio-temporal increment unit, which is used to measure the intensity of particle activity. The spatial distribution characteristics refer to the distribution characteristics of the particle collision frequency in space, which are used to extract the probability of the energy transfer path breakage. The dynamic particle knowledge graph refers to the knowledge graph that describes the behavior and relationships of particles in the spatio-temporal increment unit, which is used to guide rendering optimization. The target GPU computing node refers to the GPU computing unit that is currently executing the rendering task, which is used to determine the rendering area. The rendering area refers to the area responsible for rendering by the target GPU computing node, which is used to extract the probability of the energy transfer path breakage. The hydrodynamic coupling refers to the hydrodynamic interaction between the rendering area of the target GPU computing node and the surrounding area, which is used to determine the probability of the energy transfer path breakage. The probability of the energy transfer path breakage refers to the probability of the energy transfer path breaking in the spatio-temporal increment unit, which is used to measure the stability of the topological structure.

[0130] In the embodiment of the present application, first, based on the multi-band vibration waveform data collected by the fiber optic sensing array, analyze the spatial distribution characteristics of the particle collision frequency. The spatial distribution characteristics of the particle collision frequency usually adopt statistical analysis methods, such as by calculating the number of particle collisions per unit volume to quantify the particle collision frequency. Then, extract the probability of the energy transfer path breakage that has hydrodynamic coupling with the rendering area of the target GPU computing node in the dynamic particle knowledge graph. The extraction of the probability of the energy transfer path breakage usually adopts machine learning or regression analysis algorithms, such as by analyzing the relationship between the particle collision frequency and the breakage of the energy transfer path to calculate the breakage probability. Through this step, the system can scientifically extract the probability of the energy transfer path breakage, providing a basis for generating the local topology update weight in the subsequent step.

[0131] 502. Perform a time-domain convolution operation on the probability of the energy transfer path breakage and the particle collision frequency in the corresponding area to generate a local topology update weight.

[0132] In step 502, the energy transfer path breakage probability refers to the probability that the energy transfer path breaks in the spatio-temporal increment unit, which is used to measure the stability of the topological structure. The particle collision frequency refers to the frequency at which particles collide in the spatio-temporal increment unit, which is used to measure the intensity of particle activity. The time-domain convolution operation refers to the convolution operation of a signal in the time domain, which is used to generate the local topology update weight. The local topology update weight refers to the weight generated based on the time-domain convolution operation of the energy transfer path breakage probability and the particle collision frequency, which is used to guide the update of the local topology.

[0133] In the embodiments of the present application, first, the energy transfer path breakage probability and the particle collision frequency in the corresponding region are subjected to a time-domain convolution operation. The time-domain convolution operation usually adopts discrete convolution or fast Fourier transform technology. For example, by convolving the energy transfer path breakage probability and the particle collision frequency, the local topology update weight is generated. The generation of the local topology update weight usually adopts a weighted convolution algorithm. For example, by combining the convolution result with the weight coefficient, the update weight is generated. Through this step, the system can scientifically generate the local topology update weight, providing a basis for subsequent topology updates.

[0134] The following is a specific example:

[0135] In the internal fluid monitoring system of a turbine engine, the system first extracts the energy transfer path breakage probability that has a hydrodynamic coupling with the rendering area of the target GPU computing node in the dynamic particle knowledge graph based on the spatial distribution characteristics of the particle collision frequency in the multi-band vibration waveform data collected by the fiber optic sensing array. For example, the system analyzes the particle collision frequency in the high-frequency turbulent region and finds that its spatial distribution characteristics are highly correlated with the breakage probability of the energy transfer path. When the collision frequency exceeds 1000 times per second, the breakage probability increases significantly. Then, the system performs a time-domain convolution operation on the energy transfer path breakage probability and the particle collision frequency in the corresponding region to generate the local topology update weight. For example, through the convolution operation, the system calculates that the local topology update weight in the high-frequency turbulent region is 0.8, and the weight in the low-frequency eddy region is 0.3. Through this mechanism, the system can accurately identify the potential breakage risk of the energy transfer path and provide reliable data support for the rendering strategy of the GPU computing node, ensuring the real-time monitoring and optimization of the internal fluid dynamics of the turbine engine.

[0136] In summary, the high efficiency and high precision of particle behavior simulation in a sealed mechanical cavity are achieved through steps 501 to 503. Meanwhile, the resource utilization rate and communication efficiency are optimized, providing reliable technical support for the simulation of complex physical scenarios. By combining the multi-band vibration waveform data of the fiber optic sensing array and the dynamic particle knowledge graph, the fracture probability of the energy transfer path can be accurately quantified, providing a reliable basis for local topology update. In addition, by introducing multi-band spectrum analysis, adaptive convolution kernels, multi-scale spatial sorting algorithms, and dynamic phase correction mechanisms, the robustness and adaptability of the method are further improved, significantly enhancing the overall performance and reliability of complex physical scenario simulation.

[0137] To solve the problems of boundary offset and misalignment of the energy flux vector direction caused by the dynamic change of the energy focusing hot spot during the synchronous region boundary division process, a dynamic adjustment system for the synchronous region boundary based on a radial diffusion front is developed. This system realizes the precise dynamic adjustment of the synchronous region boundary, the real-time tracking of the energy transfer path, and the effective control of boundary offset, providing stable support for the efficient collaborative rendering of multi-GPU computing nodes. In some embodiments, in the synchronous region boundary division process under the spatial locality constraint described in step 404, using the radial diffusion front of the energy focusing hot spot in the vortex core energy density distribution map as the dynamic adjustment baseline for the synchronous region boundary includes:

[0138] 601. Based on the high-frequency phase difference jump feature in the multi-band vibration waveform data collected by the fiber optic sensing array, identify the instantaneous spatial coordinates of the energy focusing hot spot;

[0139] In step 601, the fiber optic sensing array is a device for collecting multi-band vibration waveform data, which can monitor the vibration conditions in a sealed mechanical cavity in real time. The high-frequency phase difference jump feature is a feature that describes the sudden change of the phase difference in the high-frequency vibration waveform data, usually related to the formation or dissipation of the energy focusing hot spot. The energy focusing hot spot refers to the region where the energy is highly concentrated in the sealed mechanical cavity. The instantaneous spatial coordinates describe the spatial position of the energy focusing hot spot at a certain moment.

[0140] In the embodiments of the present application, first, based on the multi-band vibration waveform data collected by the fiber optic sensing array, analyze the high-frequency phase difference jump feature. This process can accurately capture the position change of the energy focusing hot spot, providing data support for subsequent energy diffusion analysis and synchronous region adjustment. Then, through the spatio-temporal distribution feature of the phase difference jump, identify the instantaneous spatial coordinates of the energy focusing hot spot. In addition, by introducing the spectrum analysis of the multi-band vibration waveform data, the accuracy of energy focusing hot spot identification is further improved, ensuring that subtle energy concentration changes can be captured in complex fluid mechanics scenarios.

[0141] 602. Extract the radial diffusion front trajectory of the energy - focused hot spot from the vortex - core energy density distribution map;

[0142] In step 602, the vortex - core energy density distribution map is a map describing the energy density distribution in the vortex - core region of a closed mechanical cavity. The radial diffusion front trajectory is the front path describing the outward diffusion of the energy - focused hot spot.

[0143] In the embodiment of the present application, first, the radial diffusion front trajectory of the energy - focused hot spot is extracted from the vortex - core energy density distribution map. This process determines the diffusion direction and path of the energy - focused hot spot by analyzing the gradient change of the energy density, providing a basis for subsequent front - expansion rate compensation and synchronization - region adjustment. In addition, by introducing a multi - scale energy - density analysis method, the trajectory extraction strategy can be dynamically adjusted according to the diffusion characteristics of the energy - focused hot spot, further improving the accuracy of the diffusion front trajectory.

[0144] 603. Generate a front - expansion rate compensation factor according to the position offset of the energy - focused hot spot within a preset time window. The compensation factor is dynamically calculated by the ratio of the fluctuation amplitude of the high - frequency - band phase - difference jump event to the thermal expansion coefficient of the cavity inner wall;

[0145] In step 603, the preset time window is the time range used to analyze the position change of the energy - focused hot spot. The front - expansion rate compensation factor is a dynamic factor used to correct the expansion rate of the energy - diffusion front. The fluctuation amplitude of the high - frequency - band phase - difference jump event is an index describing the intensity of the phase - difference jump. The thermal expansion coefficient of the cavity inner wall is a physical parameter describing the degree of expansion of the cavity - inner - wall material under temperature change.

[0146] In the embodiment of the present application, first, a front - expansion rate compensation factor is generated according to the position offset of the energy - focused hot spot within a preset time window. Then, the compensation factor is dynamically calculated by the ratio of the fluctuation amplitude of the high - frequency - band phase - difference jump event to the thermal expansion coefficient of the cavity inner wall. This process can dynamically correct the expansion rate of the energy - diffusion front according to the dynamic change of the energy - focused hot spot and the physical characteristics of the cavity inner wall, improving the accuracy of the diffusion analysis. In addition, by introducing an adaptive compensation algorithm, the calculation parameters of the compensation factor can be dynamically adjusted according to the diffusion characteristics of the energy - focused hot spot, further improving the compensation accuracy.

[0147] 604. During the dynamic adjustment of the synchronization - region boundary, project the tangent direction of the radial diffusion front trajectory onto the map - increment transmission channel of the federated - learning framework;

[0148] In step 604, the synchronization region boundary is the range of the region divided based on the spatial locality constraint, which is used to determine the associated topological fragments between nodes that need to be synchronized. The tangent direction of the radial diffusion front trajectory is a vector describing the instantaneous propagation direction of the energy diffusion front. The atlas incremental transmission channel is a communication channel used to transmit spatio-temporal incremental units between GPU computing nodes.

[0149] In the embodiment of the present application, first, during the dynamic adjustment of the synchronization region boundary, the tangent direction of the radial diffusion front trajectory is projected onto the atlas incremental transmission channel of the federated learning framework. This process can dynamically adjust the position and range of the synchronization region boundary according to the propagation direction of the energy diffusion front, ensuring that the synchronization region is consistent with the energy diffusion front, and improving the efficiency and accuracy of data synchronization. In addition, by introducing a multi-scale projection algorithm, the projection strategy can be dynamically adjusted according to different scales of the energy diffusion front, further optimizing the adjustment efficiency of the synchronization region boundary.

[0150] 605. When it is detected that the direction offset angle of the energy flux vector exceeds the preset turbulence anisotropy threshold, trigger the local retraction operation of the synchronization region boundary, and re-determine the dynamic adjustment baseline of the synchronization region boundary based on the distribution of curvature mutation points of the radial diffusion front trajectory.

[0151] In step 605, the direction offset angle of the energy flux vector is an index describing the offset angle between the energy transfer direction and the preset direction. The turbulence anisotropy threshold is a preset threshold used to determine whether the energy transfer direction has a significant offset. The local retraction operation of the synchronization region boundary refers to the operation of contracting the synchronization region boundary inward when it is detected that the energy transfer direction has an offset. The distribution of curvature mutation points is the spatial distribution of points where the curvature of the radial diffusion front trajectory undergoes a mutation.

[0152] In the embodiment of the present application, first, when it is detected that the direction offset angle of the energy flux vector exceeds the preset turbulence anisotropy threshold, trigger the local retraction operation of the synchronization region boundary. Then, based on the distribution of curvature mutation points of the radial diffusion front trajectory, re-determine the baseline of the synchronization region boundary. This process can timely adjust the synchronization region boundary when the energy transfer direction has a significant offset, ensuring that the synchronization region is consistent with the dynamic changes of the energy diffusion front, and further improving the accuracy and real-time performance of data synchronization. In addition, by introducing a dynamic baseline adjustment algorithm, the position and shape of the baseline can be dynamically adjusted according to the changes in the distribution of curvature mutation points, further optimizing the adjustment effect of the synchronization region boundary.

[0153] The following is a specific embodiment:

[0154] In a hydrodynamic simulation scenario within an aeroengine combustion chamber, first, based on the multi-band vibration waveform data collected by the fiber optic sensing array, the high-frequency phase difference jump characteristics are analyzed. For example, in the high-temperature and high-pressure region of the combustion chamber, due to the formation of energy-focusing hotspots, significant jumps occur in the high-frequency phase difference. Then, during the dynamic adjustment process of the synchronous region boundary, the tangent direction of the radial diffusion front trajectory is projected onto the atlas incremental transmission channel of the federated learning framework. For example, in the flame propagation region, the synchronous region boundary is dynamically adjusted according to the propagation direction of the energy diffusion front. Finally, when the deviation angle of the energy flux vector direction exceeds the preset turbulence anisotropy threshold, a local retraction operation of the synchronous region boundary is triggered, and the baseline of the synchronous region boundary is redrawn based on the distribution of curvature mutation points of the radial diffusion front trajectory. For example, in the flame propagation region, due to a significant deviation in the energy transfer direction, the synchronous region boundary shrinks inward, and the baseline is redrawn according to the distribution of curvature mutation points.

[0155] In summary, through steps 601 to 605, the accurate identification and dynamic tracking of energy-focusing hotspots and diffusion fronts in the sealed mechanical cavity are achieved. At the same time, the adjustment efficiency and accuracy of the synchronous region boundary are optimized, providing reliable technical support for the simulation of complex hydrodynamic scenarios. By projecting the tangent direction of the radial diffusion front trajectory onto the atlas incremental transmission channel, the synchronous region boundary can be dynamically adjusted to ensure its consistency with the energy diffusion front. By detecting the deviation angle of the energy flux vector direction and triggering the local retraction operation of the synchronous region boundary, the synchronous region boundary can be adjusted in a timely manner, further improving the accuracy and real-time performance of data synchronization. Finally, while ensuring the simulation effect, this method significantly improves the overall performance and reliability of the simulation of complex hydrodynamic scenarios.

[0156] To solve the problems of inaccurate resolution allocation and cumulative drawing errors of particle trajectories during the rendering process of dynamic particle flows, a particle trajectory adaptive adjustment system based on GPU drawing control instructions is developed. This system realizes the precise control of the rendering granularity of particle trajectories, the dynamic optimization of resolution allocation, and the effective compensation of drawing errors, providing reliable support for the high-performance rendering of complex dynamic particle flows. In some embodiments, in step 105, according to the dynamic allocation strategy of the particle rendering resolution and the fluctuation range of the vibration waveform data feedback by the fiber optic sensing array in real time, GPU drawing control instructions for the dynamic particle flow inside the mechanical cavity are generated, and the drawing control instructions trigger the GPU computing node to perform adaptive adjustment of the rendering granularity of the particle trajectory based on the priority of the energy transfer path in the dynamic particle knowledge graph, including:

[0157] 701. Screen and extract high-frequency phase difference jump events and low-frequency energy attenuation gradient fluctuation ranges from the multi-band vibration waveform data feedback by the fiber optic sensing array in real time;

[0158] In step 701, the fiber optic sensing array is a device for collecting multi-band vibration waveform data, capable of monitoring the vibration conditions in a sealed mechanical cavity in real time. The high-frequency phase difference jump event is a characteristic describing the sudden change in the phase difference in the high-frequency vibration waveform data, usually related to the formation or dissipation of an energy focusing hot spot. The low-frequency energy attenuation gradient fluctuation interval is an interval describing the fluctuation of the energy attenuation gradient in the low-frequency vibration waveform data, usually related to the breakage or weakening of the energy transfer path.

[0159] In the embodiment of the present application, first, the high-frequency phase difference jump event and the low-frequency energy attenuation gradient fluctuation interval are screened and extracted from the multi-band vibration waveform data fed back in real time by the fiber optic sensing array. This process can accurately capture the dynamic changes of the energy focusing hot spot and the energy transfer path by analyzing the spectral characteristics of the vibration waveform data, providing data support for subsequent spatio-temporal matching and resolution adjustment. In addition, by introducing an adaptive filtering algorithm, the screening and extraction strategy can be adjusted according to the dynamic characteristics of the vibration waveform data, further improving the accuracy and efficiency of data extraction.

[0160] 702. Perform spatio-temporal matching between the screened high-frequency phase difference jump event and the priority of the energy transfer path in the dynamic particle knowledge graph to generate a spatio-temporal matching result;

[0161] In step 702, the dynamic particle knowledge graph is a graph model for describing the interaction between particles and the energy transfer relationship. The priority of the energy transfer path is an index generated based on the energy transfer path breakage probability and the particle collision frequency, used to measure the importance of the energy transfer path. Spatio-temporal matching is the process of matching the high-frequency phase difference jump event with the energy transfer path in the time and space dimensions.

[0162] In the embodiment of the present application, first, perform spatio-temporal matching between the screened high-frequency phase difference jump event and the priority of the energy transfer path in the dynamic particle knowledge graph. This process determines its correspondence with the energy transfer path by analyzing the spatio-temporal distribution characteristics of the high-frequency phase difference jump event, generating a spatio-temporal matching result. This result provides a reliable basis for subsequent resolution adjustment and particle trajectory compensation. In addition, by introducing a multi-scale matching algorithm, the matching strategy can be dynamically adjusted according to different scales of the high-frequency phase difference jump event, further improving the matching accuracy.

[0163] 703. Based on the dynamic allocation strategy of the particle rendering resolution, and in combination with the spatio-temporal matching result, generate a dynamic resolution attenuation gradient table.

[0164] In step 703, the particle rendering resolution dynamic allocation strategy refers to the strategy of dynamically allocating the resolution according to the particle motion state and rendering requirements, which is used to optimize the rendering quality. The spatio-temporal matching result refers to the matching result of the particle motion state and rendering requirements in terms of time and space, which is used to generate the dynamic resolution attenuation gradient table. The dynamic resolution attenuation gradient table refers to the table that describes the attenuation gradient of the resolution in the spatio-temporal dimension, which is used to guide the generation of GPU rendering control instructions.

[0165] In the embodiments of the present application, first, based on the particle rendering resolution dynamic allocation strategy and combined with the spatio-temporal matching result, a dynamic resolution attenuation gradient table is generated. The generation of the dynamic resolution attenuation gradient table generally adopts gradient calculation and interpolation techniques. For example, by calculating the attenuation gradient of the resolution in the spatio-temporal dimension and generating a gradient table. The generation process of the gradient table generally adopts an adaptive algorithm. For example, the attenuation gradient is dynamically adjusted according to the particle motion state and rendering requirements. Through this step, the system can scientifically generate the dynamic resolution attenuation gradient table, providing a basis for the subsequent generation of GPU rendering control instructions.

[0166] 704. Generate GPU rendering control instructions for the dynamic particle flow inside the mechanical cavity according to the dynamic resolution attenuation gradient table and the spatial distribution density of the high-frequency phase difference jump events obtained in real time.

[0167] In step 704, the dynamic resolution attenuation gradient table refers to the table that describes the attenuation gradient of the resolution in the spatio-temporal dimension, which is used to guide the generation of GPU rendering control instructions. The high-frequency phase difference jump event refers to the phase difference jump event in the high-frequency vibration waveform data, which is used to measure the intensity of particle activity. The spatial distribution density refers to the distribution density of the high-frequency phase difference jump events in space, which is used to guide the generation of GPU rendering control instructions. The GPU rendering control instructions refer to the instructions used to control the GPU to render the dynamic particle flow, which is used to optimize the rendering quality.

[0168] In the embodiments of the present application, first, according to the dynamic resolution attenuation gradient table and the spatial distribution density of the high-frequency phase difference jump events obtained in real time, GPU rendering control instructions for the dynamic particle flow inside the mechanical cavity are generated. The generation of the GPU rendering control instructions generally adopts an instruction generation algorithm. For example, by combining the dynamic resolution attenuation gradient table with the spatial distribution density of the high-frequency phase difference jump events, rendering instructions are generated. The generation process of the rendering instructions generally adopts an optimization algorithm. For example, the instruction generation is optimized according to the rendering requirements and system performance. Through this step, the system can efficiently generate GPU rendering control instructions, ensuring the rendering quality of the dynamic particle flow and system performance.

[0169] The following is a specific example:

[0170] In the internal fluid monitoring system of a turbine engine, the system first screens high-frequency phase difference jump events and low-frequency energy attenuation gradient fluctuation intervals from the multi-band vibration waveform data real-time feedback by the fiber optic sensing array. For example, high-frequency phase difference jump events are concentrated at the edges of turbine blades, and low-frequency energy attenuation gradient fluctuation intervals appear in the center of the cavity. Then, the system performs spatio-temporal matching between the high-frequency phase difference jump events and the priorities of the energy transfer paths in the dynamic particle knowledge graph to generate a matching result. For example, high-frequency turbulent regions match high-priority energy transfer paths, and low-frequency eddy regions match low-priority energy transfer paths. Subsequently, the system generates a dynamic resolution attenuation gradient table based on the dynamic allocation strategy of particle rendering resolution and the spatio-temporal matching result. For example, the resolution attenuation gradient in high-frequency turbulent regions is reduced by 10% per second, and in low-frequency eddy regions is reduced by 5% per second. Finally, the system generates GPU rendering control instructions according to the dynamic resolution attenuation gradient table and the spatial distribution density of the high-frequency phase difference jump events obtained in real-time. For example, when the density of phase difference jump events in high-frequency turbulent regions exceeds 5 times per second, the system triggers the GPU computing node to increase the particle rendering resolution in this region to ensure monitoring accuracy and efficiency. Through this mechanism, the system can accurately capture the dynamics of the internal fluid of the turbine engine and provide reliable support for mechanical performance optimization.

[0171] In summary, through steps 701 to 704, high-precision drawing and dynamic adjustment of particle trajectories in a sealed mechanical cavity are achieved, while the rendering efficiency and simulation accuracy are optimized, providing reliable technical support for the simulation of complex fluid dynamics scenarios. By combining the multi-band vibration waveform data of the fiber optic sensing array and high-frequency phase difference jump events, the dynamic changes of energy focusing hotspots and energy transfer paths can be accurately captured. By performing spatio-temporal matching between high-frequency phase difference jump events and the priorities of energy transfer paths, reliable spatio-temporal matching results can be generated. By generating a dynamic resolution attenuation gradient table and a particle trajectory compensation factor, the drawing accuracy of particle trajectories can be dynamically adjusted. By integrating the dynamic resolution attenuation gradient table and the particle trajectory compensation factor into the GPU rendering control instructions, the GPU computing node can be triggered to execute particle trajectory drawing and adaptive adjustment according to the rendering granularity corresponding to the priority of the energy transfer path. Finally, while ensuring the simulation effect, this method significantly improves the overall performance and reliability of the simulation of complex fluid dynamics scenarios.

[0172] Figure 2 The structural schematic diagram of a dynamic particle adaptive rendering system based on an autonomous and controllable GPU is provided for an embodiment of this application, as Figure 2 shown, this system includes:

[0173] The acquisition module 21 is distributedly deployed with an optical fiber sensing array within the enclosed monitoring area of the fluid machinery, and acquires multi-band vibration waveform data of the dynamic particle flow in a strong electromagnetic interference environment through the optical fiber sensing array;

[0174] The generation module 22 generates the inter-node association topology of the dynamic particle knowledge graph based on the multi-band vibration waveform data and the preset particle fluid mechanics constraint rules;

[0175] The calculation module 23 performs distributed collaborative update on the dynamic particle knowledge graph among multiple GPU computing nodes through a federated learning framework, and determines the dynamic allocation strategy of the particle rendering resolution for each GPU computing node based on the real-time change amount of the inter-node association topology obtained during the distributed collaborative update process;

[0176] The adjustment module 24 generates a GPU rendering control instruction for the dynamic particle flow inside the mechanical cavity according to the dynamic allocation strategy of the particle rendering resolution and the vibration waveform data fluctuation range feedback by the optical fiber sensing array in real time, and the rendering control instruction triggers the GPU computing node to perform adaptive adjustment of the rendering granularity of the particle trajectory based on the priority of the energy transfer path in the dynamic particle knowledge graph.

[0177] Figure 2 The described dynamic particle adaptive rendering system based on a domestically controllable GPU can execute Figure 1 The described dynamic particle adaptive rendering method based on a domestically controllable GPU in the illustrated embodiment, and its implementation principle and technical effects will not be elaborated further. For the dynamic particle adaptive rendering system based on a domestically controllable GPU in the above embodiment, the specific ways for each module and unit to perform operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0178] In a possible design, Figure 2 The dynamic particle adaptive rendering system based on a domestically controllable GPU in the illustrated embodiment can be implemented as a computing device, such as Figure 3 shown, and this computing device can include a storage component 31 and a processing component 32;

[0179] The storage component 31 stores one or more computer instructions, and among them, the one or more computer instructions are called and executed by the processing component 32.

[0180] The processing component 32 is used for the Figure 1 described dynamic particle adaptive rendering method based on a domestically controllable GPU in the above embodiment.

[0181] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-mentioned method. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above-mentioned method.

[0182] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0183] Of course, the computing device may also necessarily include other components, such as input / output interfaces, display components, communication components, etc.

[0184] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above-mentioned peripheral interface module may be an output device, an input device, etc.

[0185] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, etc.

[0186] Among them, the computing device may be a physical device or an elastic computing host provided by a cloud computing platform, etc. At this time, the computing device may refer to a cloud server, and the above-mentioned processing component, storage component, etc. may be basic server resources leased or purchased from a cloud computing platform.

[0187] The embodiment of the present application also provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the above-mentioned Figure 1 a dynamic particle adaptive rendering method based on an autonomous and controllable GPU shown in the embodiment.

[0188] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0189] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative work.

[0190] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0191] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A dynamic particle adaptive rendering method based on a self - controllable GPU, characterized in that, Including: Distributively deploy an optical fiber sensing array within the enclosed monitoring area of the fluid machinery, and collect multi-band vibration waveform data of the dynamic particle flow in a strong electromagnetic interference environment through the optical fiber sensing array; Generate the inter-node association topology of the dynamic particle knowledge graph based on the multi-band vibration waveform data and the preset particle fluid mechanics constraint rules; Distributively and collaboratively update the dynamic particle knowledge graph among multiple GPU computing nodes through the federated learning framework, and determine the dynamic allocation strategy of the particle rendering resolution for each GPU computing node based on the real-time change amount of the inter-node association topology obtained during the distributive collaborative update process; Generate a GPU rendering control instruction for the dynamic particle flow inside the mechanical cavity according to the dynamic allocation strategy of the particle rendering resolution and the vibration waveform data fluctuation range real-time fed back by the optical fiber sensing array, and the rendering control instruction triggers the GPU computing node to perform adaptive adjustment of the rendering granularity of the particle trajectory based on the priority of the energy transfer path in the dynamic particle knowledge graph.

2. The method according to claim 1, characterized in that, Distributively and collaboratively update the dynamic particle knowledge graph among multiple GPU computing nodes through the federated learning framework, and determine the dynamic allocation strategy of the particle rendering resolution for each GPU computing node based on the real-time change amount of the inter-node association topology obtained during the distributive collaborative update process, including: Divide the inter-node association topology of the dynamic particle knowledge graph into multiple update units, and the update unit contains particle collision feature change information and the probability of energy transfer path breakage; Establish a data transmission channel among GPU computing nodes in the federated learning framework through a hybrid communication protocol, and adopt a local constraint collaborative mechanism to only synchronize the inter-node topology segments that have mechanical associations with the target GPU computing node area, so as to realize the distributive collaborative update of the dynamic particle knowledge graph; Establish a priority queue in the federated learning framework, and generate a local update weight according to the product of the probability of energy transfer path breakage and the particle collision frequency in the update unit; When the local update weight exceeds the preset threshold, trigger the rendering resolution adjustment event of the corresponding GPU computing node based on the sorting of the priority queue; Generate a dynamic allocation strategy of the particle rendering resolution based on the real-time change amount of the inter-node association topology obtained during the distributive collaborative update process, and in combination with the rendering resolution adjustment event and the real-time bandwidth occupancy rate of the data transmission channel.

3. The method according to claim 2, wherein Establish a data transmission channel among GPU computing nodes in the federated learning framework through a hybrid communication protocol, and adopt a local constraint collaborative mechanism to only synchronize the inter-node topology segments that have mechanical associations with the target GPU computing node rendering area, so as to realize the distributive collaborative update of the dynamic particle knowledge graph, including: In the federated learning framework, establish a data transmission channel among GPU computing nodes through a hybrid communication protocol for transmitting the update data of the dynamic particle knowledge graph; Divide the boundary of the synchronization region with spatial locality constraints according to the radius of curvature of the energy transfer path within the rendering region of the target GPU computing node. The boundary of the synchronization region contains the inter-node correlation topology segment that shares a continuous energy flux vector with the rendering region of the target GPU computing node. Through the data transmission channel, only synchronize the inter-node correlation topology segment within the boundary of the synchronization region to achieve the distributed collaborative update of the dynamic particle knowledge graph.

4. The method according to claim 3, wherein Dividing the boundary of the synchronization region with spatial locality constraints according to the radius of curvature of the energy transfer path within the rendering region of the target GPU computing node, where the boundary of the synchronization region contains the inter-node correlation topology segment that shares a continuous energy flux vector with the rendering region of the target GPU computing node, includes: Based on the phase difference jump characteristics of adjacent sensing nodes in the multi-band vibration waveform data collected by the fiber optic sensing array, identify the mutation interval of the radius of curvature of the energy transfer path. Generate a vortex core energy density distribution map according to the mutation interval of the radius of curvature of the energy transfer path. The vortex core energy density distribution map contains the energy focusing hotspots formed by the reflection of the particle flow on the inner wall of the mechanical cavity. During the process of dividing the boundary of the synchronization region with spatial locality constraints, use the radial diffusion front of the energy focusing hotspots in the vortex core energy density distribution map as the dynamic adjustment baseline for the boundary of the synchronization region. When it is detected that the change rate of the radius of curvature of the energy transfer path exceeds the preset critical value of fluid instability, trigger the real-time re-division event of the boundary of the synchronization region. The real-time re-division event preferentially retains the inter-node correlation topology segment that shares a continuous energy flux vector with the rendering region of the target GPU computing node.

5. The method according to claim 2, wherein Establish a priority queue in the federated learning framework and generate local update weights according to the product of the energy transfer path break probability and the particle collision frequency in the update unit, including: Based on the spatial distribution characteristics of the particle collision frequency in the multi-band vibration waveform data collected by the fiber optic sensing array, extract the energy transfer path break probability that has hydrodynamic coupling with the rendering region of the target GPU computing node in the dynamic particle knowledge graph. Perform a time-domain convolution operation on the energy transfer path break probability and the particle collision frequency in the corresponding region to generate local topology update weights.

6. The method according to claim 4, wherein During the process of dividing the boundary of the synchronization region with spatial locality constraints, using the radial diffusion front of the energy focusing hotspots in the vortex core energy density distribution map as the dynamic adjustment baseline for the boundary of the synchronization region, includes: Based on the high-frequency phase difference jump characteristics in the multi-band vibration waveform data collected by the fiber optic sensing array, identify the instantaneous spatial coordinates of the energy focusing hotspots. Extract the radial diffusion front trajectory of the energy focusing hotspots in the vortex core energy density distribution map. Generate a front expansion rate compensation factor according to the position offset of the energy focusing hotspots within a preset time window. The compensation factor is dynamically calculated by the ratio of the fluctuation amplitude of the high-frequency phase difference jump event to the thermal expansion coefficient of the inner wall of the cavity. During the dynamic adjustment of the synchronization region boundary, project the tangent direction of the radial diffusion front trajectory onto the atlas incremental transmission channel of the federated learning framework; When it is detected that the deviation angle of the energy focusing hot spot vector direction exceeds the preset turbulence anisotropy threshold, trigger the local retraction operation of the synchronization region boundary, and redefine the dynamic adjustment baseline of the synchronization region boundary based on the distribution of curvature mutation points of the radial diffusion front trajectory.

7. The method according to claim 1, characterized in that Generate GPU rendering control instructions for the dynamic particle flow inside the mechanical cavity according to the dynamic allocation strategy of the particle rendering resolution and the fluctuation range of the vibration waveform data real-time feedback by the fiber optic sensing array, including: Screen and extract high-frequency phase difference jump events and low-frequency energy attenuation gradient fluctuation ranges from the fluctuation ranges of multi-band vibration waveform data real-time feedback by the fiber optic sensing array; Perform spatio-temporal matching between the screened high-frequency phase difference jump events and the priorities of the energy transfer paths in the dynamic particle knowledge graph to generate a spatio-temporal matching result; Generate a dynamic resolution attenuation gradient table based on the dynamic allocation strategy of the particle rendering resolution and in combination with the spatio-temporal matching result; Generate GPU rendering control instructions for the dynamic particle flow inside the mechanical cavity according to the dynamic resolution attenuation gradient table and the spatial distribution density of the high-frequency phase difference jump events obtained in real time.

8. A dynamic particle adaptive rendering system based on an autonomous and controllable GPU, characterized in that Including: An acquisition module that distributes and deploys a fiber optic sensing array in the closed monitoring area of the fluid machinery, and acquires multi-band vibration waveform data of the dynamic particle flow in a strong electromagnetic interference environment through the fiber optic sensing array; A generation module that generates the inter-node association topology of the dynamic particle knowledge graph based on the multi-band vibration waveform data and the preset particle hydrodynamics constraint rules; A calculation module that performs distributed collaborative update of the dynamic particle knowledge graph between multiple GPU calculation nodes through the federated learning framework, and determines the dynamic allocation strategy of the particle rendering resolution for each GPU calculation node based on the real-time change amount of the inter-node association topology obtained during the distributed collaborative update process; An adjustment module that generates GPU rendering control instructions for the dynamic particle flow inside the mechanical cavity according to the dynamic allocation strategy of the particle rendering resolution and the fluctuation range of the vibration waveform data real-time feedback by the fiber optic sensing array, and the rendering control instructions trigger the GPU calculation node to perform adaptive adjustment of the rendering granularity of the particle trajectory based on the priorities of the energy transfer paths in the dynamic particle knowledge graph.

9. A computing device, characterized in that, Including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a dynamic particle adaptive rendering method based on a domestically controllable GPU as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, Stores a computer program, and when the computer program is executed by the computer, it implements a dynamic particle adaptive rendering method based on a domestically controllable GPU as described in any one of claims 1 to 7.

Images