Three-dimensional resource dynamic allocation method and system based on grain storage

Abstract

The invention provides a three-dimensional resource dynamic allocation method and system based on grain storage, and belongs to the technical field of digital twinning. The three-dimensional resource dynamic allocation method comprises the following steps: deploying a multi-level management architecture of a cloud rendering host; wherein the multi-level management architecture is used for carrying out hierarchical management on the cloud rendering host and dynamically scheduling three-dimensional resources; controlling the multi-level management architecture to obtain an access request of a user terminal for the three-dimensional resources, and searching and scheduling the three-dimensional resources corresponding to the access request to a local reservoir area; judging whether task processing resources of the cloud rendering hosts in the local reservoir area are sufficient or not; if the task processing resources are sufficient, distributing the three-dimensional resources to a cloud rendering host in a local reservoir area in a resource distributed deployment mode for cloud rendering; if the task processing resources are insufficient, scheduling and distributing three-dimensional resources upwards across the reservoir area or the area step by step; and forwarding the rendered video stream to the user terminal. According to the invention, the problems of inflexible resource allocation and insufficient cross-regional scheduling capability in the prior art can be solved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of digital twinning, and particularly relates to a three-dimensional resource dynamic allocation method and system based on grain storage. BACKGROUND

[0002] Digital twinning technology is accelerating the digital transformation of the grain storage industry. With its high-precision grain storage environment simulation capability, it can capture subtle changes in temperature, humidity, and insect infestation in the grain storage environment in real time, thereby accurately warning potential safety hazards. Through dynamic simulation and data analysis of the storage process, resource allocation schemes such as warehouse capacity utilization and material allocation can be continuously optimized. However, as the industry digitization process accelerates, customers' requirements for digital twinning models are becoming increasingly stringent. Not only do they pursue millimeter-level precision to accurately reproduce the storage facilities and grain storage state, but they also demand higher quality three-dimensional scene rendering effects, expecting a more realistic immersive experience. At the same time, real-time interactivity requirements are rising, requiring the system to quickly respond to instructions, support multi-person collaboration, and remote control. High-quality three-dimensional scene rendering requires high task processing resources of the host, and traditional local rendering methods cannot meet the needs of large-scale user concurrent access. The current mainstream cloud rendering distributed deployment mode places rendering tasks on multiple cloud servers and distributes digital twinning applications on multiple nodes. Through video streaming technologies such as pixel streams, rendering pictures are pushed to terminals, realizing a basic architecture of "lightweight terminal + heavy cloud rendering". However, how to dynamically and efficiently allocate and schedule three-dimensional resources required for cloud rendering according to user requests to ensure user experience and optimize resource utilization is a technical problem that needs to be solved. The existing distributed deployment architecture has the problems of insufficient flexibility in resource allocation and insufficient cross-regional scheduling capability. SUMMARY

[0003] The present application aims to provide a three-dimensional resource dynamic allocation scheme based on grain storage. It aims to solve the problem of insufficient flexibility in resource allocation and insufficient cross-regional scheduling capability of the existing distributed deployment architecture.

[0004] According to a first aspect of the present application, the present application provides a three-dimensional resource dynamic allocation method based on grain storage, comprising: According to the distribution of grain storage, a multi-level management architecture of cloud rendering hosts is deployed; the multi-level management architecture is used for hierarchical management of cloud rendering hosts and dynamic scheduling of three-dimensional resources; Controlling the multi-level management architecture to obtain access requests of user terminals to three-dimensional resources, finding and scheduling the corresponding three-dimensional resources of the access requests to local library areas; Judging whether the cloud rendering hosts in the local library area have sufficient task processing resources; If the task processing resources of the cloud rendering host are sufficient, the three-dimensional resources are distributed to the cloud rendering host in the local warehouse area for cloud rendering by a resource distributed deployment manner. If the task processing resources of the cloud rendering host are insufficient, the three-dimensional resources are scheduled and distributed to the cloud rendering host in the local warehouse area by a resource distributed deployment manner. The rendering video stream obtained by cloud rendering is sent to the user terminal.

[0005] Preferably, in the three-dimensional resource dynamic allocation method, the step of deploying a multi-level management architecture of the cloud rendering host according to the distribution of the grain storage includes: A cluster deployment architecture of the cloud rendering host corresponding to the distribution of the grain storage is designed, wherein the cluster deployment architecture includes a plurality of cloud rendering hosts and a resource storage host for storing three-dimensional resources. A corresponding multi-level management architecture is deployed for the cluster deployment architecture; wherein the multi-level management architecture includes a center server, a regional server and a bottom layer warehouse area server, and the warehouse area server governs all cloud rendering hosts and resource storage hosts in the warehouse area. Preferably, in the three-dimensional resource dynamic allocation method, the step of controlling the multi-level management architecture to obtain an access request of a user terminal to three-dimensional resources, and finding and scheduling the three-dimensional resources corresponding to the access request to the local warehouse area includes: The center server of the multi-level management architecture is used to obtain the access request and query the resource storage host where the three-dimensional resource file corresponding to the access request is located. When the resource storage host where the three-dimensional resource file is located is queried, the access request is forwarded to the resource storage host through the multi-level management architecture to retrieve the three-dimensional resource file corresponding to the three-dimensional resource. The three-dimensional resource file is retrieved to the warehouse area server of the local warehouse area, and the warehouse area server distributes the three-dimensional resource file to the cloud rendering host in the local warehouse area. When the resource storage host is not queried or the three-dimensional resource file is retrieved from the resource storage host, the three-dimensional resource file is queried from the warehouse area server step by step upwards across the warehouse area or across the region.

[0006] Preferably, in the three-dimensional resource dynamic allocation method, the step of distributing the three-dimensional resources to the cloud rendering host in the local warehouse area for cloud rendering by a resource distributed deployment manner includes: The rendering task of the three-dimensional resource is split into a plurality of GPU nodes using real-time streaming media transmission technology. In the local warehouse area, a plurality of GPU corresponding cloud rendering hosts are deployed in a rendering task cluster. The task processing resources of each cloud rendering host for processing the rendering task are dynamically allocated according to a load balancing mechanism.

[0007] Preferably, in the three-dimensional resource dynamic allocation method, the step of scheduling and allocating the three-dimensional resources across the library areas or across the regions in a step-by-step upward manner comprises: sending, by a library area server of a local library area, a regional resource scheduling request to an upper-level regional server; when the regional resource scheduling request is received by the regional server, searching, by all library area servers in the jurisdiction of the regional server, for a cloud rendering host with available resources; if a cloud rendering host with available resources is found, allocating, by the regional server, the three-dimensional resources to the cloud rendering host with available resources according to a resource distributed deployment mode, and starting rendering service of the three-dimensional resources by the cloud rendering host; if a cloud rendering host with available resources is not found, initiating, by the regional server, a global resource scheduling request to a central server, searching, by the central server, for a cloud rendering host with available resources in the jurisdiction of all regional servers of the whole platform, allocating the three-dimensional resources to the cloud rendering host with available resources according to the resource distributed deployment mode, and starting rendering service of the three-dimensional resources.

[0008] Preferably, in the three-dimensional resource dynamic allocation method, the step of sending the rendered video stream obtained by cloud rendering to the user terminal comprises: establishing a WebRTC connection with the user terminal, and forwarding the rendered video stream to the user terminal using the WebRTC connection; real-time monitoring of network latency, packet loss rate and bandwidth delay on the user terminal side; comprehensively considering the network latency, packet loss rate and bandwidth delay, and adjusting the forwarding order and frequency of the rendered video stream through negative feedback.

[0009] Preferably, in the three-dimensional resource dynamic allocation method, after the step of sending the rendered video stream obtained by cloud rendering to the user terminal, the method further comprises: when the user terminal has downloaded the rendered video stream, fusing the rendered video stream into a digital twin picture generated by a digital twin model through a view alignment method; analyzing the rendered video stream to obtain multiple decoded images, inputting the multiple decoded images after format processing into the digital twin model, learning the texture features of the decoded images using the digital twin model in an adversarial learning manner, and enhancing the digital twin picture using the learned texture features.

[0010] According to a second aspect of the present application, the present application also provides a three-dimensional resource dynamic allocation system based on grain storage, comprising: a plurality of cloud rendering hosts arranged according to the distribution of the grain storage, and a multi-level management architecture corresponding to the plurality of cloud rendering hosts, the multi-level management architecture being used for hierarchical management of the cloud rendering hosts and dynamic scheduling of three-dimensional resources; wherein, A multi-level management architecture is configured to obtain an access request of a user terminal to a three-dimensional resource, find and schedule the three-dimensional resource corresponding to the access request to a local library area, judge whether the cloud rendering host in the local library area has sufficient task processing resources, if the cloud rendering host has sufficient task processing resources, allocate the three-dimensional resource to the cloud rendering host in the local library area for cloud rendering through a resource distributed deployment mode, and if the cloud rendering host has insufficient task processing resources, perform scheduling and allocation of the three-dimensional resource to a higher-level library area or a higher-level region. A cloud rendering host is configured to send a rendering video stream obtained through cloud rendering to a user terminal.

[0011] Preferably, the three-dimensional resource dynamic allocation system further comprises: A cloud rendering host cluster deployment architecture corresponding to the distribution of grain storage, wherein the cloud rendering host cluster deployment architecture comprises a plurality of cloud rendering hosts and a resource storage host configured to store three-dimensional resources. The multi-level management architecture is deployed corresponding to the cloud rendering host cluster deployment architecture, and the multi-level management architecture comprises a center server, a regional server and a lower-level library area server, the library area server governing all cloud rendering hosts and resource storage hosts in the library area.

[0012] Preferably, the center server of the multi-level management architecture is configured to obtain an access request and query a resource storage host where a three-dimensional resource file corresponding to the access request is located. The center server is further configured to forward the access request to the resource storage host through the multi-level management architecture when the resource storage host where the three-dimensional resource file is located is queried. The library area server is configured to call the three-dimensional resource file corresponding to the three-dimensional resource and allocate the three-dimensional resource file to the cloud rendering host in the local library area. The library area server is further configured to query the three-dimensional resource file from the library area server to a higher-level library area or a higher-level region when the three-dimensional resource file is not called from the resource storage host.

[0013] Preferably, in the three-dimensional resource dynamic allocation system, the library area server of the local library area sends a regional resource scheduling request to a higher-level regional server. The regional server is configured to find a cloud rendering host with available resources in all library area servers in the regional server jurisdiction when the regional resource scheduling request is received. The regional server is further configured to allocate the three-dimensional resource to the cloud rendering host with available resources according to a resource distributed deployment mode and start a rendering service of the three-dimensional resource by the cloud rendering host with available resources when the cloud rendering host with available resources is found. The regional server is further configured to initiate a global resource scheduling request to the center server through the regional server when the cloud rendering host with available resources is not found. The central server is also used to search for cloud rendering hosts with available resources within the jurisdiction of all regional servers on the entire platform, allocate three-dimensional resources to cloud rendering hosts with available resources in accordance with the resource distributed deployment method, and start rendering services for three-dimensional resources.

[0014] According to the third aspect of the present application, the present application also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the three-dimensional resource dynamic allocation method based on grain storage provided in any of the above embodiments.

[0015] The technical solution of this application has at least the following technical effects: The three-dimensional resource dynamic allocation solution based on grain storage provided by the present application can deploy a multi-level management architecture corresponding to the cloud rendering host according to the distribution of grain storage. The multi-level management architecture can perform hierarchical management of the cloud rendering host and can dynamically schedule three-dimensional resources. The above-mentioned multi-level management architecture can solve the problems of inflexible resource allocation and low scheduling efficiency in the existing technology. Specifically, by deploying a multi-level management architecture, when an access request for a three-dimensional resource is obtained from a user terminal, the three-dimensional resource corresponding to the access request is searched and uniformly scheduled to the local storage area. The local storage area is the storage area corresponding to the resource storage host where the three-dimensional resource is located. The storage area contains multiple cloud rendering hosts and the above-mentioned resource storage hosts, and the storage area server that manages the storage area performs unified scheduling. Then, it is determined whether the cloud rendering host in the local storage area has sufficient task processing resources. Since the three-dimensional resources required by the external digital twin scene may be called by a large number of user terminals, the cloud rendering host in the local storage area may have insufficient task processing resources. In view of the above situation, if the task processing resources of the cloud rendering host are sufficient, the three-dimensional resources are allocated to the cloud rendering host in the local library area through a resource distributed deployment method. The cloud rendering host in the local library area downloads the three-dimensional resources and performs cloud rendering. If the task processing resources of the cloud rendering host in the local library area are insufficient, the three-dimensional resources are scheduled and allocated step by step across the library area or even across regions. A corresponding regional server is set up in one area to manage the library area servers of multiple library areas. The above method can schedule available resources in a larger range. After downloading the above three-dimensional resources, the assigned cloud rendering host starts rendering, and then sends the rendered video stream obtained by cloud rendering to the user terminal. In summary, the above method can dynamically and efficiently allocate and schedule the three-dimensional resources required for cloud rendering according to user requests, and the startup process is efficient, thereby ensuring user experience and optimizing resource utilization. The above method can solve the problems of insufficient flexibility in resource allocation and insufficient cross-regional scheduling capabilities in the existing technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings: Figure 1 A structural schematic diagram of a three-level management architecture provided for an embodiment of the application; Figure 2 A flowchart of a three-dimensional resource dynamic allocation method based on grain storage provided for an embodiment of the application; Figure 3 A flowchart of a rendering and digital twin method for rendering a video stream provided for an embodiment of the application; Figure 4 A structural schematic diagram of a three-dimensional resource dynamic allocation system based on grain storage provided for an embodiment of the application. DETAILED DESCRIPTION

[0017] In order to more clearly illustrate the overall concept of the application, the following will be described in detail with reference to the accompanying drawings.

[0018] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the application. However, the application can be practiced without the specific details. In other instances, well-known methods, procedures, components, and circuits have not been described in detail since such can be readily understood by persons of ordinary skill in the art. The present application is not limited in scope by the exemplified embodiments, which are intended as illustrations of one or more aspects of the application. Any implementation shown is preferred only as an example of one or more aspects and implementation of the present application and do not imply any limitation on the scope of the present application. The overall description is intended to encompass any and all embodiments and variations within the scope of the present application along with full use of its general principles. It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of this application, as claimed. Other embodiments, having some advantages, as well as alternatives and modifications, will be apparent to those of ordinary skill in the art. Further, to assist clarity of exposition, not all features of an actual implementation can be described in absolute detail, but only those features necessary to appreciate the aspects of the present application.

[0019] In the present application, unless specifically stated and limited otherwise, the first feature is "on" or "under" the second feature can be that the first and second features are in direct contact, or the first and second features are indirectly in contact through an intermediate medium. In the description of the specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples.

[0020] The prior art has the following defects: High-quality 3D scene rendering requires significant processing resources from the host computer. Traditional local rendering methods struggle to meet the demands of large-scale concurrent user access. The current mainstream distributed cloud rendering deployment model places rendering tasks on multiple cloud servers, distributes digital twin applications across multiple nodes, and pushes rendered images to the terminal using video streaming technologies such as pixel streaming, achieving a "lightweight terminal + heavy cloud rendering" infrastructure. However, how to dynamically and efficiently allocate and schedule the 3D resources required for cloud rendering based on user requests, ensuring a positive user experience and optimizing resource utilization, remains a pressing technical challenge. Existing distributed deployment architectures suffer from inflexible resource allocation and insufficient cross-regional scheduling capabilities.

[0021] See Figure 1 , Figure 1 A schematic diagram of a three-level management architecture provided for an embodiment of the present application. The system architecture deployment solution of the present application adopts a three-level management system, with a central server deployed at the top to coordinate global scheduling and management; multiple regional servers are deployed in the middle layer, specifically dividing responsibilities by geographical region or business type; and multiple warehouse servers are set up under each regional server at the bottom layer to achieve localized resource management. At the end, each warehouse server directly manages multiple cloud rendering hosts, taking on the rendering tasks of specific three-dimensional resources, forming a four-level architecture of "center-region-warehouse-host" to ensure efficient resource allocation and centralized control.

[0022] The following is based on Figure 1 The three-level management architecture provided in the illustrated embodiment illustrates the dynamic three-dimensional resource allocation solution for grain storage provided in the following embodiment. This dynamic three-dimensional resource allocation solution, based on a panoramic digital grain warehouse for the grain industry, provides a dynamic resource allocation method for distributed cloud rendering deployment to address the inflexible resource allocation and low scheduling efficiency issues in existing technologies. This solution implements hierarchical management and dynamic scheduling of cloud rendering hosts 4 by deploying a multi-level management architecture, which includes at least a central server 1, regional servers 2, and warehouse servers 3. When a user requests access to the cloud rendering service for a specific warehouse area, the system first attempts to allocate resources in the local warehouse area. If local resources are insufficient, requests are sent step by step to the regional servers 2 and even to the central server 1 to schedule available resources within a larger area. The assigned cloud rendering host 4 downloads the required three-dimensional resources, initiates rendering, and forwards the rendered video stream to the user terminal 5 via WebRTC.

[0023] like Figure 2 As shown, the three-dimensional resource dynamic allocation method based on grain storage provided in the embodiment of the present application includes: S110: deploying a multi-level management architecture of the cloud rendering host according to the distribution of the grain storage; the multi-level management architecture is used for hierarchical management of the cloud rendering host and dynamic scheduling of the three-dimensional resources.

[0024] Specifically combined with Figure 1 As shown in the three-level management architecture, in the three-dimensional resource dynamic allocation method provided by the embodiments of the present application, the step of deploying the multi-level management architecture of the cloud rendering host according to the distribution of the grain storage includes: Designing a cluster deployment architecture of the cloud rendering host corresponding to the distribution of the grain storage, wherein the cluster deployment architecture includes a plurality of cloud rendering hosts and a resource storage host used for storing three-dimensional resources.

[0025] For the cluster deployment architecture, a corresponding multi-level management architecture is deployed; wherein the multi-level management architecture includes a center server, a regional server and a bottom layer warehouse area server, and the warehouse area server governs all cloud rendering hosts and resource storage hosts in the warehouse area. It should be noted that the warehouse area server can also act as a resource storage host, so that when the three-dimensional resource file needs to be called, it only needs to be called from the warehouse area server, and then the warehouse area server distributes the three-dimensional resource file to the cloud rendering host in the warehouse area.

[0026] The technical scheme provided by the embodiments of the present application, by designing a cluster deployment architecture of the cloud rendering host corresponding to the distribution of the grain storage, the cloud rendering host can quickly monitor and render the three-dimensional resources of the corresponding grain storage. The cluster deployment architecture of the cloud rendering host includes a plurality of cloud rendering hosts and a resource storage host, the cloud rendering host is used for cloud rendering of three-dimensional resources, and the resource storage host is used for storing three-dimensional resources. Combined with Figure 1 As shown in the three-level management architecture, corresponding to the cluster deployment architecture of the cloud rendering host, the multi-level management architecture at least includes a three-level architecture of a center server, a regional server and a bottom layer warehouse area server; the center server is responsible for overall global scheduling and management; the middle layer is equipped with a plurality of regional servers, which can divide responsibilities according to geographical areas or business types; the regional server is divided into a plurality of warehouse area servers, so as to implement localized resource management of the warehouse area servers in the region. At the end of each warehouse area server, a plurality of cloud rendering hosts are directly governed, which can undertake specific rendering task execution. Through the above-mentioned cluster deployment architecture corresponding to the cloud rendering host, the multi-level management architecture including the center server, the regional server and the bottom layer warehouse area server is designed, which can form a four-level overall system architecture of "center-region-warehouse area-host", thereby ensuring efficient allocation of resources and centralized management and control of business processing resources of the cloud rendering host.

[0027] Figure 2 The three-dimensional resource dynamic allocation method provided by the embodiments shown in the figure, after deploying the multi-level management architecture of the cloud rendering host, further includes: S120: The multi-level management architecture obtains an access request of a user terminal to a three-dimensional resource, finds and schedules the three-dimensional resource corresponding to the access request to a local library area.

[0028] By obtaining the access request of the user terminal to the three-dimensional resource, finding and scheduling the three-dimensional resource corresponding to the access request to the local library area, the three-dimensional resource can be uniformly found and scheduled through the multi-level management architecture, and the corresponding cloud rendering host is allocated to perform a rendering task. Through the above-mentioned manner, the hierarchical dynamic scheduling of the three-dimensional resource can be realized. Figure 1 The three-level management architecture shown in the figure realizes the hierarchical dynamic allocation and scheduling of cloud rendering freedom through the three-level architecture of the center server, the area server and the library area server, thereby improving the flexibility and response speed of resource allocation.

[0029] Specifically, as a preferred embodiment, in the three-dimensional resource dynamic allocation method, the step of controlling the multi-level management architecture to obtain an access request of a user terminal to a three-dimensional resource, find and schedule the three-dimensional resource corresponding to the access request to a local library area comprises: The center server of the multi-level management architecture is used to obtain the access request and query the resource storage host where the three-dimensional resource file corresponding to the access request is located.

[0030] The application can store a resource file list in the center server, which specifically indicates the storage location of the three-dimensional resource file. Through the above-mentioned manner, the resource storage host where the three-dimensional resource file is located can be found through the resource file list.

[0031] When the resource storage host where the three-dimensional resource file is located is queried, the access request is forwarded to the resource storage host through the multi-level management architecture to retrieve the three-dimensional resource file corresponding to the three-dimensional resource.

[0032] Here, the access request is forwarded to the resource storage host through the multi-level management architecture, and the dynamic scheduling and allocation of the three-dimensional resource file can be realized through the multi-level management architecture, thereby improving the flexibility and efficiency of resource allocation.

[0033] The library area server retrieves the three-dimensional resource file to the local library area, and the library area server allocates the three-dimensional resource file to the cloud rendering host in the local library area. The library area server manages a plurality of cloud rendering hosts at the bottom layer, retrieves the three-dimensional resource through the library area server, and distributes the three-dimensional resource to the cloud rendering host with sufficient task processing resources.

[0034] It should be noted that the library area server can also function as a resource storage host. Therefore, when the three-dimensional resource file needs to be retrieved, it only needs to be retrieved from the library area server and distributed to each cloud rendering host in the library area for rendering.

[0035] When the resource storage host is not queried or the three-dimensional resource file is not called from the resource storage host, the three-dimensional resource file is queried from the warehouse area server to the upper level across the warehouse area or across the area.

[0036] The technical scheme provided by the embodiment of the application comprises at least a center server, an area server and a warehouse area server. The access request of the user terminal is acquired by the center server, and then the three-dimensional resource file corresponding to the access request is queried according to the above multi-level management architecture. After the three-dimensional resource file is queried, the three-dimensional resource file is called by the warehouse area server and is distributed to one or more cloud rendering hosts according to the service processing resource of the multiple cloud rendering hosts in the warehouse area. If the resource storage host is not queried or the three-dimensional resource file is not called, the unqueried condition is reported to the upper level from the warehouse area server, and the three-dimensional resource file is queried by the area server across the warehouse area or by the center server across the area. Through the above method, the hierarchical dynamic scheduling and distribution of the three-dimensional resource file can be realized, and the flexibility and response speed of resource distribution are improved.

[0037] Figure 2 The three-dimensional resource dynamic distribution method provided by the embodiment further comprises the following steps after the three-dimensional resource corresponding to the access request is found and scheduled to the local warehouse area: S130: judging whether the task processing resource of the cloud rendering host in the local warehouse area is sufficient.

[0038] When the user accesses a specific warehouse area, the first stage, the resource self-checking of the client server is needed. Specifically, the user terminal initiates a rendering service request to the center server, and the rendering service request is distributed to the target warehouse area server through the center server-area server-warehouse area server. Then, the resource detection is performed, the cloud rendering host resource managed by the warehouse area server is checked, and if the cloud rendering host resource is sufficient, a cloud rendering host with available resource is selected to download the three-dimensional resource file. The cloud rendering host starts the cloud rendering service and performs the cloud rendering on the three-dimensional resource.

[0039] S140: if the task processing resource of the cloud rendering host is sufficient, the three-dimensional resource is distributed to the cloud rendering host in the local warehouse area for cloud rendering through the resource distributed deployment method.

[0040] After the library area server in the embodiment of the application acquires the three-dimensional resource file, self-checking of task processing resources of all cloud rendering hosts in the local library area needs to be performed by the library area server; if the task processing resources for rendering the three-dimensional resource file in the local library area are sufficient, the three-dimensional resource is distributed to the cloud rendering host in the local library area through the resource distributed deployment mode of Cinderella, and through the above-mentioned mode, the load balancing of each cloud rendering host in the local library area can be achieved, the GPU parallel computing capability of each cloud rendering server is fully released, and the response speed of heavy load tasks such as rendering of a three-dimensional model and real-time processing of data is significantly improved.

[0041] Specifically, the step of distributing the three-dimensional resource to the cloud rendering host in the local library area for cloud rendering through the resource distributed deployment mode includes: The rendering task of the three-dimensional resource is split into multiple GPU nodes using a real-time streaming media transmission technology. The real-time streaming media transmission technology can select the UnityRenderStreaming technology of the Unity platform, and through the real-time streaming media transmission technology, the rendering task of the three-dimensional resource can be split into multiple sub-tasks and sent to multiple GPU nodes with sufficient task processing resources.

[0042] In the local library area, multiple GPUs respectively corresponding cloud rendering hosts are deployed in a rendering task clustering manner.

[0043] According to the load balancing mechanism, the task processing resources corresponding to the rendering tasks are dynamically distributed to each cloud rendering host.

[0044] The technical scheme provided by the embodiment of the application can efficiently cope with a high-concurrency access scene through the resource distributed deployment mode, based on the real-time streaming media transmission technology such as UnityRenderStreaming, by splitting the rendering task of the three-dimensional resource into multiple GPU nodes (usually one GPU node corresponding to each cloud rendering server) and performing cluster deployment on multiple cloud rendering servers. The mode dynamically allocates task processing resources for cloud rendering through the load balancing mechanism, can fully release the parallel computing capability of the GPU, and significantly improves the response speed of heavy load tasks such as rendering of a three-dimensional model and real-time processing of data. In the three-dimensional digital twin system of the grain depot, distributed deployment can realize simultaneous access of multiple hosts to the high-precision model of the library area, real-time viewing of temperature and humidity data and device state information, and low-latency transmission of the rendering result to the terminal by cooperating with the pixel stream technology, and both immersive experience and system stability are taken into account.

[0045] Figure 2 The three-dimensional resource dynamic distribution method provided by the embodiment includes the following steps after the three-dimensional resource is distributed to the cloud rendering host in the local library area for cloud rendering through the resource distributed deployment mode: S150: If the task processing resource of the cloud rendering host is insufficient, the three-dimensional resource is scheduled and allocated across the library area or across the area step by step. The technical scheme provided by the embodiment of the application designs a multi-level management architecture. When the task processing resource of the cloud rendering host is insufficient, the regional server performs cross-library area scheduling. If the regional server does not search a cloud rendering host with available resources in the region, the central server performs cross-area scheduling on the cloud rendering host in the global range. Through the above-mentioned manner, the problem of inflexible resource allocation, insufficient cross-area capability or low resource download and startup process in the prior art can be solved.

[0046] Specifically, as a preferred embodiment, in the three-dimensional resource dynamic allocation method, the step S150 of scheduling and allocating the three-dimensional resource across the library area or across the area step by step comprises: The library area server of the local library area sends a regional resource scheduling request to the upper-level regional server.

[0047] When the regional server receives the regional resource scheduling request, all library area servers in the jurisdiction of the regional server search a cloud rendering host with available resources.

[0048] If the cloud rendering host with available resources is searched, the regional server allocates the three-dimensional resource to the cloud rendering host with available resources according to the resource distributed deployment mode, and starts the rendering service of the cloud rendering host on the three-dimensional resource.

[0049] If the cloud rendering host with available resources is not searched, the regional server initiates a global resource scheduling request to the central server, the central server searches a cloud rendering host with available resources in the jurisdiction of all regional servers, allocates the three-dimensional resource to the cloud rendering host with available resources according to the resource distributed deployment mode, and starts the rendering service on the three-dimensional resource.

[0050] The technical scheme provided by the embodiment of the application is that after the regional server receives the resource access request of the library area server, the regional server searches a cloud rendering host with available resources in all library area servers under its jurisdiction, and matches the three-dimensional resource to the cloud rendering host with available resources. Specifically, the library area server of the target library area where the cloud rendering host with available resources is located is instructed to schedule the three-dimensional resource, and the cloud rendering host of the target library area is instructed to download the three-dimensional resource file and start the rendering service. The rendering video stream is directly forwarded to the user terminal through WebRTC (avoiding multi-level forwarding delay). If the cloud rendering host with available resources is not found, the regional server initiates a global resource request to the central server.

[0051] The central server performs cross-region scheduling, specifically including requesting upgrade: after the central server receives a resource request of a region server, it searches for available cloud rendering hosts within the jurisdiction of all region servers of the platform; then it performs global scheduling, if a cloud rendering server with available resources is found, it assigns the target library area host of the target region to perform the download and rendering task across regions; finally, the rendering result is directly connected to the user terminal through WebRTC technology to ensure low-latency transmission.

[0052] Figure 2 The three-dimensional resource dynamic allocation method provided by the illustrated embodiment further includes the following steps after scheduling and allocating three-dimensional resources across libraries or across regions at a level: S160: sending the rendering video stream obtained by cloud rendering to the user terminal. Here, the cloud rendering host of the multi-level management architecture described above can send the rendering video stream to the user terminal.

[0053] The application embodiment adopts the WebRTC protocol to realize the bidirectional data communication between the user terminal and the multi-level management architecture described above: the technical solution provided by the application embodiment can support real-time data transmission between the client and the cloud rendering platform, and provide a data channel with customizable business logic, providing basic capabilities for cross-terminal interaction. The mouse and keyboard interaction solution is further optimized under this framework: through state subdivision technology (such as splitting mouse operations into multiple dimensions such as pressing / maintaining / lifting), combined with event frequency dynamic control algorithm, the user can realize low-latency, high-fidelity real-time interaction with the three-dimensional scene deployed on the cloud through terminal devices such as mouse and keyboard. This solution ultimately establishes a bidirectional data channel between the front-end interface and the cloud service, which not only guarantees smooth interaction, but also supports multi-scene adaptation through business logic customization.

[0054] Specifically, as a preferred embodiment, in the three-dimensional resource dynamic allocation method described above, the step of sending the rendering video stream obtained by cloud rendering to the user terminal includes: Establishing a WebRTC connection with the user terminal and forwarding the rendering video stream to the user terminal using the WebRTC connection.

[0055] Real-time monitoring of network latency, packet loss rate and bandwidth delay on the user terminal side; Based on the network latency, packet loss rate and bandwidth delay, the forwarding order and frequency of the rendering video stream are adjusted in a negative feedback manner. Here, according to the corresponding weights of network latency, packet loss rate and bandwidth delay, the forwarding order and frequency of the rendering video stream are adjusted in a negative feedback manner according to the comprehensive score.

[0056] Specifically, the WebRTC protocol is used to forward the rendered video stream to the user terminal, and a low-delay WebRTC is used to implement video stream transmission, supporting plug-in-free access on the browser side, and the delay can be controlled within 50-200 ms. In addition, the network delay, packet loss rate, and bandwidth delay of the user terminal side are monitored in real time, and the forwarding order and frequency of the rendered video stream are adjusted by negative feedback, which can improve the data transmission efficiency and quality, and reduce the data transmission delay of the rendered video stream.

[0057] In summary, the three-dimensional resource dynamic allocation method based on grain storage provided by the embodiments of the present application can deploy a multi-level management architecture corresponding to the cloud rendering host according to the distribution of the grain storage. The multi-level management architecture can manage the cloud rendering host in layers and dynamically schedule three-dimensional resources. Through the above multi-level management architecture, the problem of inflexible resource allocation and low scheduling efficiency in the prior art can be solved. Specifically, by deploying a multi-level management architecture, when the access request of the user terminal to the three-dimensional resource is obtained, the corresponding three-dimensional resource of the access request is found and uniformly scheduled to the local warehouse area. The local warehouse area is the warehouse area corresponding to the resource storage host where the three-dimensional resource is located. The warehouse area contains multiple cloud rendering hosts and the above resource storage host, which are uniformly scheduled by the warehouse area server managing the warehouse area. Then, it is judged whether the cloud rendering host in the local warehouse area has sufficient task processing resources. Since the three-dimensional resources required by the external digital twin scene may be called by a large number of user terminals, the cloud rendering host in the local warehouse area may have insufficient task processing resources. In view of the above situation, if the task processing resources of the cloud rendering host are sufficient, the three-dimensional resources are allocated to the cloud rendering host in the local warehouse area by a resource distributed deployment manner. The cloud rendering host in the local warehouse area downloads the three-dimensional resources and performs cloud rendering. If the task processing resources of the cloud rendering host in the local warehouse area are insufficient, the three-dimensional resources are scheduled and allocated to the cloud rendering host in the local warehouse area by a resource distributed deployment manner. In which, a region server is arranged in a region, which is used to manage the warehouse area servers of multiple warehouse areas. Through the above manner, the available resources can be scheduled in a larger range. The allocated cloud rendering host downloads the three-dimensional resources and starts rendering, and then sends the rendered video stream obtained by cloud rendering to the user terminal. In summary, through the above manner, the three-dimensional resources required by cloud rendering can be dynamically and efficiently allocated and scheduled according to user requests, and the starting process is efficient, thereby ensuring user experience and optimizing resource utilization. Through the above manner, the problem of inflexible resource allocation and insufficient cross-region scheduling capability in the prior art can be solved.

[0058] In addition, in combination with Figure 3 the rendering and digital twin method of the rendered video stream as shown in the figure, as a preferred embodiment, the rendering method comprises the following steps: The user terminal applies for a session (equivalent to the access request of the three-dimensional resource), and then the cloud rendering platform (equivalent toFigure 1 The user terminal queues and creates a session with the cloud rendering platform, and returns a queue number to the cloud rendering platform; the cloud rendering platform receives the rendering video stream from the user terminal through the WebRTC connection; then the cloud rendering platform starts the digital twin model, the digital twin model starts the digital twin service, the digital twin model generates the digital twin picture in combination with the cloud rendering video stream, and synchronizes the digital twin picture to the user terminal; when the user terminal is idle, the connection with the digital twin model is temporarily disconnected, and when the digital twin picture needs to be synchronized again, the user terminal reconnects with the digital twin model and returns the digital twin picture. Finally, when the user terminal ends the session, the cloud rendering platform releases the session, and the digital twin service with the digital twin model is also ended.

[0059] In addition, in combination with Figure 3 The method shown above, the three-dimensional resource dynamic allocation method provided by the above embodiment further includes, after the step of sending the rendering video stream obtained by cloud rendering to the user terminal: The cloud rendering host or the center server of the multi-level management architecture synchronously sends the rendering video stream to the digital twin model; and the digital twin model synchronizes the digital twin picture with the user terminal.

[0060] When the user terminal downloads the rendering video stream, the rendering video stream is fused into the digital twin picture generated by the digital twin model through a perspective alignment manner.

[0061] The rendering video stream is parsed to obtain multiple decoded images, the multiple decoded images after format processing are input into the digital twin model, the digital twin model is used to learn the texture features of the decoded images in an adversarial learning manner, and the learned texture features are used to enhance the digital twin picture.

[0062] The embodiment of the present application can fuse the rendering video stream into the digital twin scene, specifically binds the shooting perspective of the rendering video stream and the specific perspective of the digital twin model in a perspective alignment manner, and then embeds the rendering video stream into the digital twin picture of the digital twin model. Here, the rendering video stream is parsed to obtain multiple decoded images, the decoded images are then format-processed, the decoded images are input into the digital twin model, the digital twin model is used to learn the texture features of the images in an adversarial learning manner, and the learned texture features are used to enhance the digital twin picture produced by the digital twin model, thereby improving the authenticity of the digital twin picture.

[0063] In addition, referring to Figure 4 , Figure 4 A three-dimensional resource dynamic allocation system based on grain storage is provided for the embodiment of the present application, in combination with Figure 1 The three-dimensional resource dynamic allocation system includes: The plurality of cloud rendering hosts are arranged according to the distribution of the grain storage, and the plurality of levels of management architecture correspond to the plurality of cloud rendering hosts, and are used for hierarchical management of the cloud rendering hosts and dynamic scheduling of the three-dimensional resources; the plurality of levels of management architecture include a center server, a regional server and a warehouse area server three-level architecture.

[0064] The plurality of levels of management architecture are used for obtaining an access request of a user terminal to the three-dimensional resources, finding and scheduling the three-dimensional resources corresponding to the access request to a local warehouse area; judging whether the cloud rendering hosts in the local warehouse area are sufficient in task processing resources; if the cloud rendering hosts are sufficient in the task processing resources, distributing the three-dimensional resources to the cloud rendering hosts in the local warehouse area for cloud rendering through a resource distributed deployment mode; and if the cloud rendering hosts are insufficient in the task processing resources, scheduling and distributing the three-dimensional resources to a higher level across the warehouse areas or across the regions. The cloud rendering host (i.e., the server rendering end shown in Figure 4 ) is used for cloud rendering of the three-dimensional resources, and sending the obtained rendering video stream to the user terminal (i.e., the client shown in Figure 4 ).

[0065] Specifically, as a preferred embodiment, the three-dimensional resource dynamic allocation system provided by the application includes, as shown in Figure 1 and 4 The architecture shows that the three-dimensional resource dynamic allocation system provided by the application further includes: A cluster deployment architecture of the cloud rendering hosts corresponding to the distribution of the grain storage, wherein the cluster deployment architecture includes a plurality of cloud rendering hosts and a resource storage host used for storing the three-dimensional resources; The plurality of levels of management architecture are deployed corresponding to the cluster deployment architecture, and the plurality of levels of management architecture include a center server, a regional server and a bottom-layer warehouse area server, and the warehouse area server governs all the cloud rendering hosts and the resource storage host in the warehouse area.

[0066] Specifically, as a preferred embodiment, in combination with Figure 1 and Figure 4 The above three-dimensional resource dynamic allocation system, the center server of the plurality of levels of management architecture is used for obtaining an access request, and querying a resource storage host where a three-dimensional resource file corresponding to the access request is located; The center server is further used for forwarding the access request to the resource storage host through the plurality of levels of management architecture when the resource storage host where the three-dimensional resource file is located is queried, The warehouse area server is used for calling the three-dimensional resource file corresponding to the three-dimensional resource, and distributing the three-dimensional resource file to the cloud rendering host in the local warehouse area by the warehouse area server; The warehouse area server is further used for querying the three-dimensional resource file from the warehouse area server across the warehouse areas or across the regions when the three-dimensional resource file is not called from the resource storage host.

[0067] Preferably, in the above three-dimensional resource dynamic allocation system, the local storage area server sends a regional resource scheduling request to the superior regional server; The regional server is used to search for cloud rendering hosts with available resources in all the database servers within the jurisdiction of the regional server when receiving a regional resource scheduling request; The regional server is further configured to allocate 3D resources to the cloud rendering host with available resources and start the rendering service of the cloud rendering host for the 3D resources in accordance with the resource distributed deployment method when a cloud rendering host with available resources is found. The regional server is also used to initiate a global resource scheduling request to the central server through the regional server when no cloud rendering host with available resources is found; The central server is also used to search for cloud rendering hosts with available resources within the jurisdiction of all regional servers on the entire platform, allocate three-dimensional resources to cloud rendering hosts with available resources in accordance with the resource distributed deployment method, and start rendering services for three-dimensional resources.

[0068] like Figure 4 The system shown here shows that the three-level management architecture for calling 3D resources and allocating cloud rendering hosts includes the following steps: First, the user access and resource allocation method is as follows: When a user accesses a specific library area, the request processing flow of the cloud rendering service is as follows: The first stage is the self-check of the server resources in the repository area: Request initiation: The user initiates a cloud rendering service request to the central server, which is distributed through the three levels of central server → regional server → warehouse server and finally reaches the target warehouse server.

[0069] Resource detection: The repository server checks the task processing resources of the locally managed cloud rendering host: If there are sufficient task processing resources: Select an available host and trigger it to download the 3D resource file (the 3D resource file can be stored in the resource storage host in the local library area or deployed on the library area server).

[0070] The host starts the cloud rendering service and forwards the rendered video stream to the user terminal in real time through WebRTC technology.

[0071] If insufficient: the repository server sends a resource request to the superior regional server.

[0072] In the second phase, regional servers are dispatched across different storage areas: Request reception: After receiving the resource request from the database server, the regional server searches for available cloud rendering hosts among all database servers under its jurisdiction.

[0073] If available resources are found, resource allocation is performed, specifically instructing the host of the target warehouse area to download the three-dimensional resource file and start the rendering service. The rendered video stream is directly forwarded to the user terminal through WebRTC (avoiding multi-level forwarding delay).

[0074] If no available cloud rendering host is found, the regional server initiates a global resource request to the central server.

[0075] In the third stage, the central server schedules across regions: Request escalation: After receiving the resource request from the regional server, the central server searches for available cloud rendering hosts within the jurisdiction of all regional servers on the platform.

[0076] Global scheduling: If available resources are found, the target host of the target warehouse area in the target region is assigned to perform the download and rendering tasks. The rendering result is directly connected to the user terminal through WebRTC technology to ensure low-latency transmission.

[0077] Finally, the above multi-level management architecture is realized through the WebRTC protocol for real-time streaming and interaction with the user terminal: the embodiments of the application use low-latency WebRTC to realize video streaming, support browser-side plugin-free access, and the delay can be controlled within 50-200 ms.

[0078] In summary, the above-mentioned embodiments of the application provide a three-dimensional resource dynamic allocation scheme based on grain storage, which has the following advantages: (1) Hierarchical dynamic scheduling: Through the three-level architecture of the central server, regional server, and warehouse area server, the dynamic allocation and scheduling of cloud rendering resources are realized, improving the flexibility and response speed of resource allocation.

[0079] (2) Improved resource utilization: When local resources are insufficient, idle resources can be scheduled across warehouse areas and regions, effectively improving the utilization rate of the overall cloud rendering host and avoiding resource waste.

[0080] (3) User experience guarantee: Even if a specific warehouse area is resource-constrained, user requests can be scheduled to other available resources through higher-level servers, ensuring the success rate and smoothness of user access.

[0081] (4) On-demand loading: After obtaining the allocation task and confirmation, the cloud rendering host first checks whether the resource is deployed locally, and if not, it downloads the corresponding three-dimensional resource, reducing unnecessary storage and bandwidth occupation.

[0082] The modules described in the embodiments of the application can be implemented in software or hardware. In some cases, the module name does not constitute a limitation on the unit itself.

[0083] The various embodiments in the specification are described in progressive manner, and the same or similar parts between the various embodiments can be mutually referred to, and each embodiment focuses on the difference from other embodiments.

[0084] The above merely provides an example of the present application, and is not intended to limit the present application. The present application can have various modifications and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A three-dimensional resource dynamic allocation method based on grain storage, characterized in that: include: Deploy a multi-level management architecture for cloud rendering hosts based on the distribution of grain storage; The multi-level management architecture is used to hierarchically manage the cloud rendering hosts and dynamically schedule 3D resources; Controlling the multi-level management architecture to obtain a user terminal's access request for a three-dimensional resource, and searching and dispatching the three-dimensional resource corresponding to the access request to a local repository; Determine whether the cloud rendering host in the local library has sufficient task processing resources; If the task processing resources of the cloud rendering host are sufficient, the 3D resources are allocated to the cloud rendering host in the local library for cloud rendering through resource distributed deployment; If the task processing resources of the cloud rendering host are insufficient, the 3D resources are scheduled and allocated step by step across warehouses or regions. The rendered video stream obtained by cloud rendering is sent to the user terminal.

2. The method according to claim 1, wherein The steps of deploying a multi-level management architecture of cloud rendering hosts based on the distribution of grain storage include: Designing a clustered deployment architecture of cloud rendering hosts corresponding to the distribution of the grain storage, wherein the clustered deployment architecture includes multiple cloud rendering hosts and a resource storage host for storing three-dimensional resources; For the clustered deployment architecture, the corresponding multi-level management architecture is deployed; wherein, the multi-level management architecture includes a central server, a regional server and an underlying library area server, and the library area server governs all cloud rendering hosts and resource storage hosts in the library area.

3. The method according to claim 1 or 2, wherein: The step of controlling the multi-level management architecture to obtain a user terminal's access request for a three-dimensional resource, and searching and dispatching the three-dimensional resource corresponding to the access request to a local repository includes: Using the central server of the multi-level management architecture to obtain the access request, querying the resource storage host where the three-dimensional resource file corresponding to the access request is located; When the resource storage host where the three-dimensional resource file is located is found, the access request is forwarded to the resource storage host via the multi-level management architecture to retrieve the three-dimensional resource file corresponding to the three-dimensional resource; Retrieving the three-dimensional resource file to a storage area server in the local storage area, and having the storage area server distribute the three-dimensional resource file to a cloud rendering host in the local storage area; When the resource storage host is not found or the three-dimensional resource file is retrieved from the resource storage host, the three-dimensional resource file is searched from the library server step by step upward across the library or across regions.

4. The method according to claim 1, wherein The step of allocating the three-dimensional resources to the cloud rendering host in the local library for cloud rendering through distributed resource deployment includes: Using real-time streaming technology, the rendering task of the three-dimensional resource is split among multiple GPU nodes; In the local library area, cloud rendering hosts corresponding to the multiple GPUs are clustered and deployed according to the rendering tasks; Dynamically allocate task processing resources corresponding to the rendering tasks to each cloud rendering host according to the load balancing mechanism.

5. The method according to claim 1, wherein The step of scheduling and allocating three-dimensional resources across storage areas or regions step by step upward includes: Sending a regional resource scheduling request to a superior regional server through the local regional server; When the regional server receives the regional resource scheduling request, it searches for cloud rendering hosts with available resources in all the database servers within the jurisdiction of the regional server; If a cloud rendering host with available resources is found, the regional server allocates the three-dimensional resources to the cloud rendering host with available resources according to the resource distributed deployment method, and starts a rendering service for the three-dimensional resources on the cloud rendering host; If no cloud rendering host with available resources is found, a global resource scheduling request is initiated to the central server through the regional server, and the central server searches for cloud rendering hosts with available resources within the jurisdiction of all regional servers on the entire platform. The three-dimensional resources are allocated to the cloud rendering hosts with available resources according to the resource distributed deployment method, and the rendering service for the three-dimensional resources is started.

6. The method according to claim 1, wherein The step of sending the rendered video stream obtained by cloud rendering to the user terminal includes: Establishing a WebRTC connection with the user terminal, and forwarding the rendered video stream to the user terminal using the WebRTC connection; Real-time monitoring of network latency, packet loss rate, and bandwidth delay on the user terminal side; Taking into account the network delay, packet loss rate and bandwidth delay, negative feedback is used to adjust the forwarding order and forwarding frequency of the rendered video stream.

7. The method according to claim 1, wherein After the step of sending the rendered video stream obtained by cloud rendering to the user terminal, the method further includes: After the user terminal downloads the rendered video stream, the rendered video stream is integrated into the digital twin screen generated by the digital twin model through perspective alignment; Parse the rendered video stream to obtain multiple frames of decoded images, input the formatted multiple frames of decoded images into the digital twin model, use the digital twin model to learn the texture features of the decoded images in an adversarial learning manner, and use the learned texture features to enhance the digital twin picture.

8. A three-dimensional resource dynamic allocation system based on grain storage, characterized in that: include: Multiple cloud rendering hosts are set up according to the distribution of grain storage, and a multi-level management architecture corresponding to the multiple cloud rendering hosts, the multi-level management architecture is used to hierarchically manage the cloud rendering hosts and dynamically schedule three-dimensional resources; wherein, The multi-level management architecture is used to obtain access requests for 3D resources from user terminals, search for and dispatch the 3D resources corresponding to the access requests to the local storage area; determine whether the cloud rendering host in the local storage area has sufficient task processing resources; if the task processing resources of the cloud rendering host are sufficient, allocate the 3D resources to the cloud rendering host in the local storage area for cloud rendering through a resource distributed deployment method; if the task processing resources of the cloud rendering host are insufficient, dispatch and allocate the 3D resources across storage areas or regions step by step upward; The cloud rendering host is used to send the rendered video stream obtained by cloud rendering to the user terminal.

9. The system according to claim 8, wherein The system further comprises: A clustered deployment architecture of cloud rendering hosts corresponding to the distribution of the grain storage, wherein the clustered deployment architecture includes multiple cloud rendering hosts and a resource storage host for storing three-dimensional resources; The multi-level management architecture is deployed corresponding to the clustered deployment architecture. The multi-level management architecture includes a central server, a regional server and an underlying library server. The library server governs all cloud rendering hosts and resource storage hosts in the library.

10. The system according to claim 8 or 9, characterized in that The local storage area server sends a regional resource scheduling request to the superior regional server; The regional server is configured to search for cloud rendering hosts with available resources in all storage area servers within the jurisdiction of the regional server upon receiving the regional resource scheduling request; The regional server is further configured to allocate the three-dimensional resource to the cloud rendering host with available resources according to the resource distributed deployment method if a cloud rendering host with available resources is found, and start a rendering service for the three-dimensional resource by the cloud rendering host; The regional server is further configured to initiate a global resource scheduling request to the central server via the regional server when no cloud rendering host with available resources is found; The central server is further used to search for cloud rendering hosts with available resources within the jurisdiction of all regional servers on the entire platform, allocate the three-dimensional resources to the cloud rendering hosts with available resources according to the resource distributed deployment method, and start rendering services for the three-dimensional resources.

Images