Equipment virtualization orchestration method for heterogeneous resource pooling management

Abstract

The invention provides a heterogeneous resource pooling management-oriented equipment virtualization orchestration method, and relates to the technical field of equipment virtualization orchestration, and the method comprises the steps: carrying out the equipment virtualization of to-be-orchestrated equipment, unifying the to-be-orchestrated equipment into a same resource pool, obtaining a production resource formula of each production plan, and obtaining a production resource formula of each production plan; establishing a resource control atlas according to a historical production resource formula, obtaining a production resource formula of a future production plan as a demand resource, and arranging resources in a resource pool through a resource allocation algorithm according to the real-time digital label and the demand resource of each virtual device type, the method comprises the steps of arranging virtual devices, outputting a logic connection relation and arranged virtual device resources through a resource control atlas, marking timestamps on the arranged virtual device resources, obtaining the usage amount of resource types of each device type through the same time interval, and calculating a historical resource allocation equilibrium coefficient as an optimization target. And optimizing the single arrangement ratio, the weight adjustment coefficient and the fitting coefficient.

Description

Device virtualization orchestration method for heterogeneous resource pooling management Technical Field

[0001] This invention relates to the field of device virtualization orchestration technology, specifically a device virtualization orchestration method for heterogeneous resource pooling management. Background Technology

[0002] Traditional industrial resource management systems typically involve different types of control equipment. Even equipment of the same type can vary greatly in usage due to different manufacturers. For example, PLC control equipment, switches, and storage servers are often heterogeneous devices with significant differences in their communication protocols, resource interfaces, and management methods, making resource integration and unified scheduling difficult.

[0003] Furthermore, the rigid binding of physical device resources through fixed configurations lacks unified virtualization abstraction capabilities, resulting in idle or locally overloaded computing and bandwidth resources. The lack of modeling and optimization of inter-device dependencies leads to neglect of underlying transmission latency or bottlenecks during production task scheduling, easily causing execution timeouts. Traditional static allocation schemes cannot dynamically optimize based on the actual usage ratio of resource types, resulting in resource imbalances, frequent I / O blocking, or wasted computing power. Independent management interfaces for multi-vendor devices and closed ecosystems force a surge in manual maintenance costs, and hardware expansion requires re-adaptation, inhibiting system flexibility. These problems collectively constitute the technical predicament of "unconnectable, unmanageable, and poorly usable," urgently requiring a breakthrough through virtualization orchestration and intelligent scheduling.

[0004] Furthermore, the control logic relationships between equipment in industrial production are complex. Traditional methods often rely on manual experience to configure equipment networking and logical relationships, which is not only inefficient but also makes it difficult to ensure the consistency between resource allocation and actual control logic. At the same time, due to the lack of a continuous optimization mechanism for historical resource allocation effects, the system cannot adaptively adjust scheduling strategies, resulting in the inability to guarantee the balance and stability of the resource pool under long-term operation.

[0005] Existing technologies typically employ static or fixed-weight resource allocation strategies, which struggle to adapt to dynamically changing production plans and real-time resource status changes. This often leads to uneven resource allocation, low utilization rates, and a disconnect between control logic and physical resources. Therefore, there is an urgent need for a method that can achieve unified abstraction and dynamic weighting of heterogeneous equipment resources, and intelligently orchestrate and generate connection relationships based on real-time status and production needs.

[0006] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to provide a device virtualization orchestration method for heterogeneous resource pooling management, so as to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a device virtualization orchestration method for heterogeneous resource pooling management, comprising the following steps: Step 1: For devices to be orchestrated, virtualization is performed by identifying the device type and basic information of the devices, unifying the virtual devices into the same resource pool, and assigning a standardized, digitized digital tag to each virtual device, continuously reporting the basic information of the devices to the resource pool, and completing the registration of the resource pool; Step 2: Obtaining the device type, device quantity, resource type, and resource type quantity required for each production plan as the production resource formula, and based on the devices in the historical production resource formula... Step 3: Based on the real-time digital tags and required resources of each virtual device type, the resources in the resource pool are arranged using a resource allocation algorithm, and the logical connection relationship of the arranged virtual devices is output through the resource control graph. Step 4: The arranged virtual device resources are marked with timestamps, and the usage of each device type's resources is obtained at the same time interval. The single arrangement ratio, weight adjustment coefficient, and fitting coefficient are optimized by calculating the historical resource allocation balance coefficient as the optimization objective.

[0009] Furthermore, the device types include controllers, switches, and storage devices; the basic information of the controller includes working memory, communication interface and communication speed, and instruction execution speed; the basic information of the switch includes the number of ports and the switching capacity of the ports; the basic information of the storage device includes capacity, storage category and category capacity and read / write rate; the digital tag includes device type, device identifier, device basic information, and basic information status data.

[0010] Furthermore, the specific method for establishing the resource control graph is as follows: based on the control logic between equipment types in the historical production resource formula, including the data transmission direction and the instruction sending direction, the interaction and dependence between different equipment types in terms of resource demand are identified. A resource control graph is constructed through a graph neural network, where nodes represent different equipment types and edges represent the control logic between them. The equipment types in the production resource formula are used as the input to the resource control graph, and the control logic between equipment types is used as the output of the resource control graph, thereby training the resource control graph.

[0011] Furthermore, the resource allocation algorithm includes a dynamic weight allocation function, a resource allocation function, and node logic output. The dynamic weight allocation function is used to allocate dynamic weight ratios when orchestrating resources of virtual devices. The resource allocation function orchestrates the required resources to virtual devices through the dynamic weight ratios and obtains the control logic relationship between virtual devices after orchestration through node logic output.

[0012] Furthermore, the specific calculation steps of the dynamic weight allocation function are as follows: based on historical production resource formulas, the usage of different resource types for each type of equipment is statistically analyzed; the resource matching ratio between each resource type in the same equipment is calculated; based on the current usage ratio and resource matching ratio of each resource in each virtual equipment, the future reserved resource amount in the equipment is adjusted by calculating the dynamic weight ratio; the specific mathematical formula for calculating the dynamic weight ratio is: in, For resource type The usage ratio is Dynamic weighting ratio at time, For resource type The resource allocation ratio, This is the weighting adjustment factor. These are the fitting coefficients. For resource type The usage ratio.

[0013] Further, the specific steps of the resource allocation function are as follows: based on the current resource type usage ratio and resource demand of each virtual device, calculate the similarity value between the actual remaining ratio after allocating the resource demand to the device and the dynamic weight ratio after allocation; by setting the single orchestration ratio for each resource type, and based on the total amount and resource demand of different resource types in each device, determine the fitness between the resource demand and the total amount by calculating the ratio difference between the allocation ratio of the resource demand and the total resource amount and the single orchestration ratio; based on the similarity value and the ratio difference, calculate the orchestration coefficient for each device by allocating weights, and orchestrate the resource demand based on the orchestration coefficient.

[0014] Furthermore, the node logic output is used to obtain the control logic relationship between virtual devices by inputting the digital tags of the virtual devices with allocated resource requirements into the resource control graph.

[0015] Furthermore, the calculation method for the historical resource allocation equilibrium coefficient is as follows: in, The historical resource allocation equilibrium coefficient. for Moment Virtual Device resource types Usage for Moment Virtual Device resource types The average usage, This represents the number of times data was collected.

[0016] Furthermore, the method for optimizing the single arrangement ratio, weight adjustment coefficient, and fitting coefficient is based on the sparrow search algorithm, and the specific process includes discoverer and follower.

[0017] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention virtualizes the devices to be orchestrated and unifies them into a single resource pool. It obtains the production resource formula for each production plan, establishes a resource control graph based on historical production resource formulas, and obtains the production resource formula for future production plans as required resources. Based on the real-time digital tags and required resources for each virtual device type, it orchestrates the resources in the resource pool using a resource allocation algorithm. The orchestrated virtual devices are then logically connected using the resource control graph. The orchestrated virtual device resources are timestamped. The usage of each device type's resource type is obtained at equal time intervals. The optimization objective is to calculate the historical resource allocation balance coefficient, and to optimize the single orchestration ratio, weight adjustment coefficient, and fitting coefficient. This invention, based on device virtualization and digital tagging technology, unifies heterogeneous devices into a resource pool and achieves efficient and balanced resource orchestration based on a dynamic weight allocation function and resource allocation algorithm. Furthermore, by combining the resource control graph, it automatically generates logical connections between devices, significantly improving resource utilization and scheduling efficiency. This method can also optimize key parameters based on historical data, further improving the balance and adaptability of resource allocation, and is particularly suitable for application scenarios involving the collaboration of multiple types of industrial equipment and the parallel operation of complex production plans. Attached Figure Description

[0018] Figure 1 is a schematic diagram of the overall method flow of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0020] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0021] Example: Please refer to Figure 1. The present invention provides a technical solution: a device virtualization orchestration method for heterogeneous resource pooling management. The specific steps include: Step 1: For the devices to be orchestrated, the device is virtualized by identifying the device type and the basic information of the device, the virtual devices are unified into the same resource pool, and each virtual device is assigned a standardized and digitally described digital tag. The basic information of the device is continuously reported to the resource pool to complete the registration of the resource pool.

[0022] In this embodiment, the device types include controllers, switches, and storage devices; the basic information of the controller includes working memory, communication interface and communication speed, and instruction execution speed; the basic information of the switch includes the number of ports and the switching capacity of the ports; the basic information of the storage device includes capacity, storage category and category capacity, and read / write rate; the digital tag includes device type, device identifier, device basic information, and basic information status data.

[0023] The process involves identifying the type of device to be virtualized, such as a controller, switch, or storage device, and collecting its basic information. For example, controller information includes working memory, communication interfaces, communication speed, and instruction execution speed; switch information includes the number of ports and port switching capacity; and storage device information includes capacity, storage type, type capacity, and read / write speed. Based on this basic information, a corresponding virtual device model is created. This abstracts the physical device, removing its specific hardware characteristics and presenting its functions and resources in a software-defined manner. During virtualization, a standardized digital tag is used to assign a unique identifier to each virtual device, described as a digital tag including device type, device identifier, basic information, and status data. The virtual devices are then managed within the same resource pool. Each virtual device periodically or in real-time reports its basic information and status data, ensuring the information in the resource pool remains up-to-date. Throughout this process, the virtualization layer provides a unified interface, allowing the upper-layer management system to focus on resource scheduling, orchestration, and optimization without needing to concern itself with the specific implementation details of the physical devices, ultimately completing the registration of the resource pool and the virtualization of the devices.

[0024] Industrial control equipment (such as PLCs, switches, and storage servers) comes from different manufacturers, with vastly different communication protocols, management interfaces, and performance parameters, creating serious "information silos." This method identifies the equipment type and extracts its key basic information (such as the controller's instruction execution speed and the switch's port capacity), transforming physical devices into virtual devices with standardized descriptions. This enables completely different types of devices to be understood and managed at the same level, laying a solid foundation for building a unified resource pool.

[0025] Secondly, the digital tag assigned to each virtual device not only contains static identity information (device type, identifier), but more importantly, it continuously updates its dynamic "basic information status data" (such as current memory usage, real-time read / write speed). This data is continuously reported to the resource pool, which is equivalent to equipping the resource management system with "real-time sensors," enabling it to accurately grasp the real-time health status, performance load, and availability of all resources in the pool, providing indispensable data for subsequent intelligent and precise resource orchestration.

[0026] A unified resource pool and standardized digital tags enable the management system to manage complex heterogeneous devices globally, just as it would manage a single homogeneous resource. Subsequent orchestration algorithms can match, allocate, and optimize resources based on this standardized tag information, rather than relying on the dedicated interfaces of specific devices, thereby truly achieving flexible, automated, and efficient management of heterogeneous resource pools. Without this step of abstraction and standardization, all subsequent intelligent orchestration would be impossible.

[0027] Step 2: Obtain the equipment type, quantity of equipment, resource type, and quantity of resource type required for each production plan as the production resource formula. Establish a resource control map based on the control logic between equipment types in the historical production resource formula, and obtain the production resource formula for the future production plan as the demand resources.

[0028] In this embodiment, the specific method for establishing the resource control graph is as follows: based on the control logic between equipment types in the historical production resource formula, including the data transmission direction and the instruction sending direction, the interaction and dependence between different equipment types in terms of resource demand are identified. A resource control graph is constructed through a graph neural network, where nodes in the graph represent different equipment types and edges represent the control logic between them. The equipment types in the production resource formula are used as the input to the resource control graph, and the control logic between equipment types is used as the output of the resource control graph, thereby training the resource control graph.

[0029] The system transforms specific production plans into precise resource requirement templates, or production resource formulas, enabling resource allocation to be based on precise quantitative demand rather than empirical guesswork. Simultaneously, by analyzing the co-occurrence relationships and collaboration patterns of equipment types in historical formulas, it deeply uncovers the implicit control logic and dependencies between equipment, constructing a visualized resource control graph. A graph neural network is specifically used to construct this graph because it can process graph-structured data extremely efficiently. By aggregating node neighbor information through message passing mechanisms, it can not only capture complex multidimensional relationships between equipment (such as the high-frequency collaboration patterns between PLCs and switches), but also represent equipment nodes as low-dimensional vectors through embedding learning. This allows for the prediction and reasoning of unknown equipment combination logic relationships. When faced with new production demands, the system can not only match based on historical patterns but also intelligently deduce the optimal equipment connection logic using the generalization ability of graph neural networks, greatly enhancing the adaptability, intelligence, and decision-making accuracy of the orchestration system.

[0030] Step 3: Based on the real-time digital tags and resource requirements of each type of virtual device, the resources in the resource pool are arranged using a resource allocation algorithm, and the logical connection relationship of the arranged virtual devices is output through a resource control graph.

[0031] In this embodiment, the resource allocation algorithm includes a dynamic weight allocation function, a resource allocation function, and node logic output. The dynamic weight allocation function is used to allocate dynamic weight ratios when orchestrating resources of virtual devices. The resource allocation function orchestrates the required resources to virtual devices through the dynamic weight ratios and obtains the control logic relationship between virtual devices after orchestration through node logic output.

[0032] By combining resource allocation algorithms, particularly dynamic weight allocation functions with resource allocation functions, resource allocation strategies can be dynamically adjusted to flexibly address the changing resource demands of different virtual devices during actual production, enhancing the adaptability and real-time performance of the orchestration process. The dynamic weight allocation function adjusts resource allocation priorities based on the actual operating status and resource requirements of virtual devices, ensuring sufficient resource support for critical equipment. Meanwhile, the resource allocation function, based on dynamic weight ratios, efficiently schedules and optimizes resources in the resource pool, avoiding waste and shortages and improving resource utilization efficiency. Node logic output, by outputting the control logic relationships between orchestrated virtual devices, accurately expresses the collaborative working relationships between devices, providing necessary reference and support for subsequent production task scheduling—crucial in heterogeneous resource pool management. In this way, device virtualization orchestration not only ensures reasonable resource allocation but also effectively supports the coordination and scheduling of complex production tasks, improving the flexibility, scalability, and intelligence of device virtualization, further optimizing the efficiency and effectiveness of resource pool management.

[0033] In this embodiment, the specific calculation steps of the dynamic weight allocation function are as follows: based on historical production resource formulas, the usage of different resource types for each equipment type is statistically analyzed; the resource coordination ratio between each resource type in the same equipment is calculated; based on the current usage ratio and resource coordination ratio of each resource in each virtual equipment, the future reserved resource amount in the equipment is adjusted by calculating the dynamic weight ratio; the calculation formula for the resource coordination ratio between each resource type is: in, For resource allocation ratio, For the first The middle of the production resource formula Usage of each type of resource The quantity of production resource formulas.

[0034] By statistically analyzing historical data, this study uncovers the inherent and common resource ratios (such as memory, instruction execution speed, and communication ports) within the same type of equipment (e.g., PLC) during production activities. It calculates the ratio of the total usage of each resource type in all historical production formulas. This ratio represents a proven, stable, and efficient resource combination pattern, providing an "ideal reference template" for current resource allocation. It ensures that resource allocation is no longer isolated but takes into account the synergy between resources.

[0035] Suppose there were 10 historical production plans corresponding to 10 production resource recipes, all using PLC equipment. In these 10 plans, the total memory usage of all PLCs was 300GB, the total instruction execution speed usage was 400 MIPS, and the total number of communication ports used by all PLCs was 30. The calculated resource allocation ratio is 300:400:30, which simplifies to 3:4:0.3. This ratio of 3:4:0.3 is an important historical lesson, telling the system that a "typical" production task consumes approximately these three resources from a single PLC in this proportional relationship. When allocating resources for new tasks in the future, efforts should be made to ensure that the remaining resource ratio after allocation is as close to this pattern as possible, thus guaranteeing the balance and stability of the resource pool.

[0036] The formula for calculating the usage ratio of each resource is as follows: in, For resource type The usage ratio, For resource type Usage Resource types in virtual devices The total amount.

[0037] Resource types The usage ratio refers to the consumption of a certain type of resource in a specific virtual device. For resource type Usage Resource types in virtual devices The total amount, It intuitively reflects the surplus of the resource. The higher the value (closer to 1), the more scarce the resource is; The lower the value (closer to 0), the more abundant the resource is. This value provides a real-time basis for resource allocation.

[0038] The specific mathematical formula for calculating the dynamic weighting ratio is as follows: in, For resource type Using proportion Dynamic weighting ratio at time, For resource type The resource allocation ratio, This is the weighting adjustment factor. The fitting coefficients are denoted as .

[0039] Its dynamic weighting ratio combines historical experience and real-time status to calculate which resources should have higher "priority" or "weight" when allocating them to the current device.

[0040] Representing the "remaining proportion" of resources, it achieves the goal of load balancing. The higher the remaining proportion (i.e., the less currently used), the greater its weight, and the more the system is inclined to assign new tasks to it. At the same time, this trend is also modulated by the historical resource allocation ratio. For example, if memory demand has historically been high, then its weight will be more sensitive to changes.

[0041] The weight adjustment factor is a constant adjustment term, which can be understood as "basic weight reduction". It is used to fine-tune the baseline of the calculation so that the weight value is more in line with the needs of the actual scenario.

[0042] This is used to simulate the nonlinear effects of resource scarcity, when the resource utilization rate is very high (close to 1). It is a very small value; when the resource utilization rate is very low. When the value is close to 1, its function is to appropriately increase the weight when resources are abundant to encourage allocation. When resources are scarce, the value of this item becomes very small, thus significantly reducing its weight along with the main part, effectively preventing further allocation of overused resources, and acting as a "brake".

[0043] In this embodiment, the specific steps of the resource allocation function are as follows: based on the current resource type usage ratio and resource demand of each virtual device, a similarity value is calculated between the actual remaining ratio after allocating the required resources to the device and the dynamic weight ratio after allocation. The method for calculating the similarity value is as follows: in, For proportional similarity values, Resource type at the current moment The usage ratio, For the allocated resource type exist The dynamic weighting ratio at any given time. The number of resource types; theoretically, a good resource allocation should make the state of the allocated equipment as close as possible to the "ideal state" calculated by the dynamic weighting function.

[0044] The proportional similarity value can comprehensively measure the overall deviation between the "expected state after resource allocation" and the "ideal target state" by analyzing all resource types. Find the Euclidean norm of the difference (i.e., the square root of the sum of squares). The final proportional similarity value is obtained. It is a comprehensive difference value. The smaller the proportional similarity value, the more the overall resource status after the new task is assigned to the device matches the "ideal state" as perceived by the system, and the more "healthy" and "high-quality" the assignment is.

[0045] By setting a single orchestration ratio for each resource type, and based on the total amount and required resources of different resource types in each device, the fitness between the required resources and the total resources is determined by calculating the ratio difference between the allocation ratio of the required resources to the total resources for each resource type and the single orchestration ratio. The formula for calculating the ratio difference is as follows: in, The difference in ratios, For resource type The resources required Resource types in virtual devices The total amount of resources, For the number of resource types, This represents the ratio of single-time arrangement.

[0046] The ratio difference effectively quantifies the "impact" and "matching degree" of a single production task on the resource capacity of candidate equipment. That is, it indirectly reflects the impact of differences in equipment capacity by quantifying the ratio of a single resource demand to the total equipment capacity, thus ensuring the adaptability between task requirements and equipment scale. (Resource type) The demand for resources represents the actual resource requirements of production tasks. Resource types in virtual devices The total amount of resources reflects the inherent resource scale of the equipment, and the ratio between the two is... This directly reflects the proportion of the task requirements to the total resources of the equipment. The single-time allocation ratio serves as a preset benchmark threshold (e.g., 0.2) and defines the ideal resource allocation range for a single task. By calculating and summing the absolute differences between the actual ratio and the ideal threshold, the ratio difference comprehensively reflects the degree of deviation between the task requirements and the equipment resource capacity. The smaller the ratio difference, the higher the matching degree between the task requirements and the equipment resource scale. This avoids both the resource waste and potential frequent scheduling overhead of "overloading" and the resource overload risk of "overloading". This ensures that the system selects the most efficient equipment to carry the task and operates stably for long-term resource allocation.

[0047] Different devices have varying resource capacities. Larger resources are more advantageous for handling larger resource demands. When a large resource is used to accommodate smaller resource demands, it is inevitable that a large number of devices will be needed to fill the resource capacity. As a result, the resource will need to constantly switch between different resource demands during operation, which will reduce its efficiency. The ratio difference indirectly reflects the impact of device capacity differences by quantifying the ratio of a single resource demand to the total capacity of the device. The calculation method of "required resource / total resource" naturally results in a smaller calculation result for devices with a larger total resource capacity when dealing with the same resource demand, i.e., a lower ratio difference value. This gives large-scale devices a greater advantage in orchestration scoring.

[0048] Based on the similarity ratio and the ratio difference, the orchestration coefficient of each device is calculated by assigning weights, and the orchestration resources are arranged according to the orchestration coefficient. The specific calculation formula is as follows: in, For arrangement coefficients, These are the weights of the proportional similarity value and the ratio difference, respectively. , .

[0049] By weighted fusion of "proportional similarity values" "and ratio difference" "Two key indicators constitute a comprehensive scheduling coefficient." Its core function lies in achieving balance and unity in multi-objective optimization decision-making; proportional similarity value The focus is on assessing the long-term health and ideality of equipment status after resource allocation (i.e., whether the allocated weights are close to the ideal weights), while the ratio difference... This focuses on assessing the immediate impact of a single task allocation on equipment capacity. In actual orchestration, these two objectives often conflict (for example, a device in near-ideal condition but with limited resources). Although the value is small, The value may be very large, so it is necessary to assign weights ( This reflects the system's strategy preferences under different business scenarios (e.g., whether to prioritize ensuring the long-term stability of the resource pool or to prioritize avoiding large tasks blocking high-performance devices). Based on the calculated comprehensive orchestration coefficient (the smaller the value, the lower the overall cost), all candidate devices are ranked to select the optimal orchestration target under the given strategy, rather than making a one-sided decision based solely on a single indicator. Static pre-defined parameters based on expert experience: Domain experts, drawing upon their extensive operational experience and deep understanding of business needs, directly provide... and Assign a fixed value.

[0050] In this embodiment, the node logic output is used to obtain the control logic relationship between virtual devices by inputting the digital tags of the virtual devices with allocated resource requirements into the resource control graph.

[0051] For example, a production plan requires two PLCs for resource orchestration, two switches, and three storage types. Based on the constructed knowledge graph, a network topology diagram is created to connect the relationships between PLCs, the relationships between PLCs and switches, and the relationships between storage types. This outputs the connection logic between PLCs, switches, and storage devices.

[0052] Step 4: Arrange the virtual device resources and mark them with timestamps. At the same time interval, obtain the usage of resource type for each device type. Optimize the single arrangement ratio, weight adjustment coefficient and fitting coefficient by calculating the historical resource allocation balance coefficient as the optimization objective.

[0053] In this embodiment, the historical resource allocation balance coefficient is calculated as follows: in, The historical resource allocation equilibrium coefficient. for Moment Virtual Device resource types Usage for Moment Virtual Device resource types The average usage, This represents the number of times data was collected.

[0054] The system records timestamps for each orchestration and periodically collects resource usage data. Using the calculation formula of historical resource allocation balance coefficient, it comprehensively averages the deviations of various resource usages of all virtual devices from their average values ​​at all collection times. This accurately measures the global balance of resource allocation over a period of time. The smaller the allocation balance coefficient, the more balanced the resource allocation and the smaller the load difference between devices. This effectively avoids the bottleneck problem of resource utilization where some devices are overused while others are idle.

[0055] Furthermore, with minimizing the allocation balance coefficient as the optimization objective, the system employs an optimization algorithm (sparrow search algorithm) to automatically optimize key control parameters (single orchestration ratio, weight adjustment coefficient, and fitting coefficient). This means that the algorithm can self-adjust and correct resource allocation strategies based on historical data from actual operation, making the dynamic weight calculation and resource allocation function more adaptable to specific production environments and load characteristics. This process enables the system to leap from single static orchestration to continuous dynamic self-learning, ensuring that the heterogeneous resource pool maintains high resource utilization and stable performance output under large-scale, long-cycle operation, while reducing reliance on manual parameter tuning and significantly improving the level of intelligent management.

[0056] In this embodiment, the method for optimizing the single arrangement ratio, weight adjustment coefficient, and fitting coefficient is based on the sparrow search algorithm. The specific process includes discoverers and followers. Discoverers are a subset of individuals in the population with higher fitness (i.e., the optimization target value of the corresponding solution is better, or in this case, the historical resource allocation equilibrium coefficient is smaller). They are responsible for exploring new and potentially better food sources (i.e., parameter combination regions).

[0057] The discoverer fine-tunes the algorithm near the better parameter combination it finds to approximate the optimal solution in that region, thereby improving the search accuracy and convergence speed of the algorithm; the discoverer's position update formula is: in, For the updated number The middle generation Only sparrows in the first The position of the sparrow To optimize the types of weights used in the optimization model within the data, the dimension... To optimize the weight dimensions in the model running on the data, For random numbers that follow a normal distribution, A uniformly random number in [0,1]. This is the warning threshold, with a value range of [0.5, 1]; the follower position update formula; in, The current iteration number The sparrow with the best fitness in the population was in the 1st month. The position of the dimension To generate a random number between -1 and 1.

[0058] All the above formulas use dimensionless numerical values ​​for calculation, and the numerical values ​​substituted into the formulas are all in the International System of Units (SI). The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0059] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0060] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0061] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A device virtualization orchestration method for heterogeneous resource pooling management, characterized in that, The specific steps include: Step 1: For the equipment to be orchestrated, virtualize the equipment by identifying the equipment type and basic information, unify the virtual equipment into the same resource pool, assign a standardized and digitally described digital tag to each virtual equipment, and continuously report the basic information of the equipment to the resource pool to complete the registration of the resource pool; Step 2: Obtain the equipment type, quantity of equipment, resource type, and quantity of resource type required for each production plan as the production resource formula, establish a resource control graph based on the control logic between equipment types in the historical production resource formula, and obtain the production resource formula of the future production plan as the demand resources; Step 3: Based on the real-time digital tags and demand resources of each virtual equipment type, orchestrate the resources in the resource pool through a resource allocation algorithm, and output the logical connection relationship of the orchestrated virtual equipment through the resource control graph; Step 4: Mark the orchestrated virtual equipment resources with timestamps, obtain the usage of resource type for each equipment type at the same time interval, and optimize the single orchestration ratio, weight adjustment coefficient, and fitting coefficient by calculating the historical resource allocation balance coefficient as the optimization objective.

2. The device virtualization orchestration method for heterogeneous resource pooling management according to claim 1, characterized in that: The device types include controllers, switches, and storage devices; the basic information of controllers includes working memory, communication interface and communication speed, and instruction execution speed; the basic information of switches includes the number of ports and the switching capacity of the ports; the basic information of storage devices includes capacity, storage category and category capacity, and read / write rate; the digital tag includes device type, device identifier, device basic information, and basic information status data.

3. The device virtualization orchestration method for heterogeneous resource pooling management according to claim 1, characterized in that: The specific method for establishing the resource control graph is as follows: Based on the control logic between equipment types in the historical production resource formula, including the data transmission direction and the instruction sending direction, the interaction and dependence between different equipment types in terms of resource demand are identified. A resource control graph is constructed through a graph neural network. The nodes in the graph represent different equipment types, and the edges represent the control logic between them. The equipment types in the production resource formula are used as the input of the resource control graph, and the control logic between equipment types is used as the output of the resource control graph. The resource control graph is then trained.

4. The device virtualization orchestration method for heterogeneous resource pooling management according to claim 1, characterized in that: The resource allocation algorithm includes a dynamic weight allocation function, a resource allocation function, and node logic output. The dynamic weight allocation function is used to allocate dynamic weight ratios when orchestrating resources of virtual devices. The resource allocation function orchestrates the required resources to virtual devices through the dynamic weight ratios and obtains the control logic relationship between virtual devices after orchestration through node logic output.

5. The device virtualization orchestration method for heterogeneous resource pooling management according to claim 4, characterized in that: The specific calculation steps of the dynamic weight allocation function are as follows: Based on historical production resource formulas, the usage of different resource types for each equipment type is statistically analyzed; the resource matching ratio between each resource type within the same equipment is calculated; based on the current usage ratio and resource matching ratio of each resource in each virtual equipment, the future reserved resource amount in the equipment is adjusted by calculating the dynamic weight ratio; the specific mathematical formula for calculating the dynamic weight ratio is: in, For resource type The usage ratio is Dynamic weighting ratio at time, For resource type The resource allocation ratio, This is the weighting adjustment factor. These are the fitting coefficients. For resource type The usage ratio.

6. The device virtualization orchestration method for heterogeneous resource pooling management according to claim 5, characterized in that: The specific steps of the resource allocation function are as follows: Based on the current resource type usage ratio and resource demand of each virtual device, calculate the similarity value between the actual remaining ratio after allocating the resource demand to the device and the dynamic weight ratio after allocation; by setting the single orchestration ratio for each resource type, and based on the total amount and resource demand of different resource types in each device, determine the fitness between the resource demand and the total amount by calculating the ratio difference between the allocation ratio of the resource demand and the total resource amount and the single orchestration ratio; based on the similarity value and the ratio difference, calculate the orchestration coefficient for each device by allocating weights, and orchestrate the resource demand based on the orchestration coefficient.

7. A device virtualization orchestration method for heterogeneous resource pooling management according to claim 5, characterized in that: The node logic output is used to obtain the control logic relationship between virtual devices by inputting the digital tags of the virtual devices with allocated resource requirements into the resource control graph.

8. The device virtualization orchestration method for heterogeneous resource pooling management according to claim 1, characterized in that: The method for calculating the historical resource allocation equilibrium coefficient is as follows: in, The historical resource allocation equilibrium coefficient. for Moment Virtual Device resource types Usage for Moment Virtual Device resource types The average usage, This represents the number of times data was collected.

9. A device virtualization orchestration method for heterogeneous resource pooling management according to claim 1, characterized in that: The method for optimizing the single arrangement ratio, weight adjustment coefficient, and fitting coefficient is based on the sparrow search algorithm, and the specific process includes discoverer and follower.

Images