Edge-end cooperative computing power distribution method, device, equipment, medium and program product

Abstract

The embodiment of the invention discloses an edge-end cooperative computing power distribution method, device and equipment, a medium and a program product. Comprising the following steps: acquiring business logistics information of each terminal node in a current business period; inputting the logistics information of each industry into a pre-trained business classification prediction model, and determining a real-time terminal node and a real-time computing power prediction value; determining a real-time computing power demand according to the real-time terminal node and the real-time computing power predicted value, and dividing the total computing power resource into a real-time service container area and a non-real-time service container area according to the real-time computing power demand; and according to the non-real-time computing power resource of the non-real-time service container area and the node weight of each non-real-time terminal node, determining a non-real-time access node of each time slot corresponding to the next service period, and performing non-real-time computing power resource allocation on each non-real-time access node in the non-real-time service container area. The computing power resource of the edge node can be fully utilized, the distributed service can be flexibly deployed, and the service processing capability of the power internet of things is improved.

Description

Technical Field

[0001] The present invention relates to the field of Internet of Things technology, and in particular to an edge-end collaborative computing power allocation method, device, equipment, medium and program product. Background Art

[0002] The Internet of Things is an expanded application and network extension of communication networks and the Internet. It uses sensing technology and intelligent equipment to perceive and identify the physical world, and performs calculations, processing, and knowledge mining through network transmission and interconnection. It realizes information interaction and seamless connection between people and objects, and between objects, and achieves the purpose of controlling, accurately managing, and making scientific decisions about the physical world. It has been widely used in the power sector to form the power Internet of Things.

[0003] With the rapid development of the power Internet of Things, the explosive growth of data volume far exceeds the bearing capacity of network bandwidth, and the emergence of a large number of new intelligent applications has put forward higher requirements for system latency. Edge computing, as a new computing mode, can enable data to be processed promptly and effectively in the edge network near the source. In the edge computing mode, the terminal offloads tasks to nearby edge computing devices for execution, effectively alleviating the network congestion and high latency problems caused by transmitting all data to the data center in traditional cloud computing. However, edge computing still has bottlenecks, mainly because its computing, communication and storage resources are limited. When the number of terminals and the scale of data increase further, the edge nodes tend to be saturated, which will lead to a serious decline in the business processing capabilities of the edge-side fusion terminal.

[0004] The access scale of terminals, data and services in the current power Internet of Things is constantly growing, and massive power data needs to be analyzed and calculated on edge computing devices to achieve efficient and flexible business processing and decision-making. Due to the distributed nature of new power system services, the working mode of edge computing node devices has gradually changed from a single-machine independent operation mode to a distributed access and collaborative interaction mode between edge nodes and many terminal nodes. The current working mode of the edge side and the terminal side can no longer meet the growing urgent needs of distributed business deployment, which may lead to the inability to execute important businesses in the power network in a timely manner, reducing the business processing capacity of the power Internet of Things network. Summary of the invention

[0005] The present invention provides a method, device, equipment, medium and program product for allocating edge collaborative computing power. According to the different business types corresponding to the numerous terminal nodes connected to the edge nodes, computing power resources of different collaborative mechanisms are allocated to each terminal node in the edge node to ensure that more important businesses can be executed in a timely manner, and the computing power resources of the edge nodes can be fully utilized, so that distributed businesses can be flexibly deployed, thereby improving the processing capabilities of the power Internet of Things network for businesses.

[0006] In a first aspect, an embodiment of the present invention provides a method for allocating edge-end collaborative computing power, including:

[0007] Obtain the business logistics information of each terminal node in the current business cycle;

[0008] Input logistics information of various industries into the pre-trained business classification prediction model to determine the real-time terminal nodes and real-time computing power prediction values;

[0009] Determine the real-time computing power demand according to the real-time terminal node and the real-time computing power prediction value, and divide the total computing power resources into a real-time business container area and a non-real-time business container area according to the real-time computing power demand, so as to allocate the computing power resources corresponding to the real-time computing power demand to the real-time business container area;

[0010] According to the non-real-time computing power resources of the non-real-time service container area and the node weight of each non-real-time terminal node, the non-real-time access nodes for each time slot corresponding to the next service cycle are determined, and non-real-time computing power resources are allocated to each non-real-time access node in the non-real-time service container area.

[0011] In a second aspect, an embodiment of the present invention further provides an edge-end collaborative computing power allocation device, including:

[0012] The information acquisition module is used to obtain the business logistics information of each terminal node in the current business cycle;

[0013] The real-time computing power prediction module is used to input logistics information of various industries into the pre-trained business classification prediction model to determine the real-time terminal nodes and real-time computing power prediction values;

[0014] A business container division module is used to determine the real-time computing power demand according to the real-time terminal node and the real-time computing power prediction value, and divide the total computing power resources into a real-time business container area and a non-real-time business container area according to the real-time computing power demand, so as to allocate the computing power resources corresponding to the real-time computing power demand to the real-time business container area;

[0015] The non-real-time computing power allocation module is used to determine the non-real-time access nodes for each time slot corresponding to the next business cycle according to the non-real-time computing power resources of the non-real-time business container area and the node weights of each non-real-time terminal node, and allocate non-real-time computing power resources to each non-real-time access node in the non-real-time business container area.

[0016] In a third aspect, an embodiment of the present invention further provides an edge collaborative computing power allocation device, the edge collaborative computing power allocation device comprising:

[0017] at least one processor; and a memory communicatively coupled to the at least one processor;

[0018] Among them, the memory stores a computer program that can be executed by at least one processor, and the computer program is executed by at least one processor so that the at least one processor can implement the edge collaborative computing power allocation method of any embodiment of the present invention.

[0019] In a fourth aspect, an embodiment of the present invention further provides a storage medium comprising computer executable instructions, which, when executed by a computer processor, are used to execute the edge collaborative computing power allocation method of any embodiment of the present invention.

[0020] In a fifth aspect, an embodiment of the present invention further provides a computer program product, including a computer program, which, when executed by a processor, is used to execute the edge collaborative computing power allocation method of any embodiment of the present invention.

[0021] An edge collaborative computing power allocation method, device, equipment, medium and program product provided by an embodiment of the present invention obtains the business logistics information of each terminal node in the current business cycle; inputs the business logistics information into a pre-trained business classification prediction model to determine the real-time terminal nodes and the real-time computing power prediction value; determines the real-time computing power demand according to the real-time terminal nodes and the real-time computing power prediction value, and divides the total computing power resources into a real-time business container area and a non-real-time business container area according to the real-time computing power demand, so as to allocate the computing power resources corresponding to the real-time computing power demand to the real-time business container area; determines the non-real-time access nodes for each time slot corresponding to the next business cycle according to the non-real-time computing power resources of the non-real-time business container area and the node weights of each non-real-time terminal node, and allocates non-real-time computing power resources to each non-real-time access node in the non-real-time business container area. By adopting the above technical solution, different terminal nodes in the access edge nodes are divided into real-time terminal nodes and non-real-time terminal nodes according to the types of services required to be implemented in the terminal nodes, and then real-time computing power prediction is performed for the real-time terminal nodes, and the total computing power resources are divided into a real-time business container area that reserves and allocates full computing power resources for real-time terminal nodes and a non-real-time business container area that provides partial computing power resources for non-real-time terminal nodes, that is, a reservation-based coordination mechanism is adopted for real-time services. For non-real-time services, the non-real-time access nodes that can provide computing power in different time slots in the next business cycle are determined based on the non-real-time computing power resources that can be provided to non-real-time services in the non-real-time business container area and the node weights of different non-real-time terminal nodes, that is, a competitive coordination mechanism is adopted for non-real-time services, and computing power resources are provided for the most important non-real-time services in the same time period, ensuring that more important services can be executed in a timely manner, and the computing power resources of the edge nodes can be fully utilized, so that distributed services can be flexibly deployed, and the processing capabilities of the power Internet of Things network for services are improved.

[0022] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 A flowchart of a method for allocating edge-end collaborative computing power provided in Embodiment 1 of the present invention;

[0025] Figure 2 A flowchart of a method for allocating edge-end collaborative computing power provided in Embodiment 2 of the present invention;

[0026] Figure 3 This is a structural example diagram of a GRU unit provided in Embodiment 2 of the present invention;

[0027] Figure 4 A schematic diagram of the structure of an edge-end collaborative computing power allocation device provided in Embodiment 3 of the present invention;

[0028] Figure 5 A structural schematic diagram of an edge collaborative computing power allocation device provided in Example 4 of the present invention. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0031] Embodiment 1

[0032] Figure 1 A flow chart of an edge collaborative computing power allocation method provided for the first embodiment of the present invention. The embodiment of the present invention can be applied to the situation where computing power resources are divided for edge computing nodes of the electric power Internet of Things. The method can be executed by an edge collaborative computing power allocation device, which can be implemented by software and / or hardware, and the edge collaborative computing power allocation device can be configured in an edge collaborative computing power allocation device. Optionally, the edge collaborative computing power allocation device can be an electronic device, which can be a notebook, a desktop computer, a smart tablet, an edge computing device in the electric power Internet of Things, etc., and the embodiment of the present invention does not limit this.

[0033] like Figure 1 As shown, an edge-end collaborative computing power allocation method provided by an embodiment of the present invention specifically includes the following steps:

[0034] S101. Obtain the business logistics information of each terminal node in the current business cycle.

[0035] In this embodiment, the business cycle can be specifically understood as the time period required for a business to start and end. The current business cycle can be specifically understood as the business cycle at the current moment. The terminal node can be specifically understood as the power terminal device connected to the same edge computing device as the edge computing node in the edge network corresponding to the power Internet of Things. It can be understood that each terminal node can perform different businesses. Business logistics information can be specifically understood as business feature information related to the business executed by the terminal node.

[0036] Specifically, in order to make full use of the computing power resources of the edge computing device during the edge computing process, the edge computing device can be accessed based on the current business cycle, and the computing power resources in the edge computing device in the next business cycle can be pre-allocated using the business executed by the computing power resources of the edge computing device. To achieve the above purpose, when it is necessary to pre-allocate the computing power resources of the edge computing device in the next business cycle, the edge computing device can obtain the business flow information of the business flow transmitted to each terminal node accessed in the current business cycle through the southbound communication interface set therein.

[0037] S102: Input the logistics information of each industry into the pre-trained business classification prediction model to determine the real-time terminal node and real-time computing power prediction value.

[0038] In this embodiment, the business classification prediction model can be specifically understood as a neural network model for classifying the business types corresponding to the business logistics information according to the business characteristics in the input business logistics information, and predicting the computing power demand of the business belonging to the real-time business type. The real-time terminal node can be specifically understood as a terminal node whose business type is a real-time business type. The real-time computing power prediction value can be specifically understood as the computing power resource prediction required for the real-time business type in the next business cycle obtained by performing computing power prediction on the business of the real-time business type.

[0039] Specifically, the business flow information corresponding to each terminal node is input into the pre-trained business classification prediction model, and the business flow information is firstly feature extracted through the business classification prediction model, and the business flow information containing each feature is determined to be business flow information of the real-time business type or business flow information of the non-real-time business type based on the extracted features. Then, the terminal node corresponding to the business flow information belonging to the real-time business type is determined as the real-time terminal node, and after determining that the business flow information is business flow information of the real-time business type, the corresponding features are used to predict the computing power resources, and the real-time computing power prediction value that the real-time terminal node may need in the next business cycle is obtained.

[0040] S103. Determine the real-time computing power demand according to the real-time terminal node and the real-time computing power prediction value, and divide the total computing power resources into a real-time business container area and a non-real-time business container area according to the real-time computing power demand, so as to allocate the computing power resources corresponding to the real-time computing power demand to the real-time business container area.

[0041] In this embodiment, the real-time computing power demand can be specifically understood as the amount of computing power resources required to execute real-time business in the next business cycle. The real-time business container area can be specifically understood as a container for executing real-time business type services, which contains computing power resources for executing real-time business type services. The non-real-time business container area can be specifically understood as a container for executing non-real-time business type services, which contains computing power resources for executing non-real-time business type services. It can be understood that computing power resources may include central processing unit (CPU) resources, container resources, memory resources, etc., and the embodiments of the present invention do not limit this.

[0042] Specifically, according to the determined real-time terminal nodes and the predicted values ​​of real-time computing power, the maximum amount of computing power resources required on average for executing real-time business type business at each moment in the next business cycle is determined, and the determined amount of computing power resources is used as the real-time computing power demand. Since the real-time business type business is of high importance in the power system and needs to ensure its timely execution, the computing power resources corresponding to the real-time computing power demand in the total computing power resources can be divided into the real-time business container area when the total computing power resources in the edge computing device can meet the real-time computing power demand, so that the real-time business of each real-time terminal node in the next business cycle can be executed using the computing power resources in the real-time business container area. At the same time, the computing power resources in the total computing power resources except the real-time computing power demand are divided into a container area as a non-real-time business container area, so that the business of the remaining terminal nodes except the real-time terminal nodes in the next business cycle can be executed in the non-real-time business container area.

[0043] In an embodiment of the present invention, different types of services in the edge network of the power Internet of Things are accessed and divided by dividing containers, thereby ensuring sufficient computing resources for more important real-time business types. By adjusting the containers, the computing resources of the edge nodes can be fully utilized and flexibly deployed, thereby improving the processing capacity of the power Internet of Things network for services.

[0044] S104. Determine the non-real-time access nodes for each time slot corresponding to the next service cycle according to the non-real-time computing power resources of the non-real-time service container area and the node weight of each non-real-time terminal node, and allocate non-real-time computing power resources to each non-real-time access node in the non-real-time service container area.

[0045] In this embodiment, non-real-time computing resources can be specifically understood as computing resources allocated by edge computing devices to non-real-time service container areas and available for non-real-time service types. Node weights can be specifically understood as weight values ​​used to indicate the importance of services executed at different times in non-real-time terminal nodes. Non-real-time access nodes can be specifically understood as terminal nodes that need to access services for a certain time slot.

[0046] Specifically, according to the non-real-time computing power resources that can be allocated to the non-real-time business container area in the next business cycle, the concurrency upper limit and processing upper limit for non-real-time business type services at the same time can be determined, and then the next business cycle can be divided into multiple time slots, and according to the node weight of each non-real-time terminal node, the weight of each non-real-time terminal node accessing the edge computing device in different time slots is determined, and according to each weight, the non-real-time terminal node that is most suitable for accessing and executing non-real-time business type services in each time slot in the next business cycle is selected as the non-real-time access node, and then within the maximum allowable range of the non-real-time computing power resources of the non-real-time business container area, in each time slot in the next business cycle, the non-real-time access node accessing the time slot is provided with the computing power resources it needs, and the non-real-time computing power resource allocation is completed. It can be understood that for a time slot in the next business cycle, if the non-real-time computing power resources are greater than the computing power resources required by the non-real-time access node in the time slot, it is only necessary to provide the non-real-time access node with the computing power resources it needs. If the non-real-time computing power resources are less than the computing power resources required by the non-real-time access node in the time slot, all the non-real-time computing power resources can be provided to the non-real-time access node. Although the non-real-time access node is still unable to execute its corresponding non-real-time business type business at the optimal speed at this time, since the importance of non-real-time business type business is relatively low compared to real-time business type business, it can still meet the business execution requirements in the power Internet of Things network. The embodiment of the present invention adopts this computing power resource division method. On the basis of ensuring that the real-time business type business can be executed in a timely and sufficient manner, it provides more sufficient computing power resources for non-real-time business type business to the greatest extent, so that the computing power resources of the edge computing device can be fully utilized.

[0047] The technical solution of this embodiment is to obtain the business logistics information of each terminal node in the current business cycle; input the business logistics information into the pre-trained business classification prediction model to determine the real-time terminal nodes and the real-time computing power prediction value; determine the real-time computing power demand according to the real-time terminal nodes and the real-time computing power prediction value, and divide the total computing power resources into a real-time business container area and a non-real-time business container area according to the real-time computing power demand, so as to allocate the computing power resources corresponding to the real-time computing power demand to the real-time business container area; determine the non-real-time access nodes for each time slot corresponding to the next business cycle according to the non-real-time computing power resources of the non-real-time business container area and the node weights of each non-real-time terminal node, and allocate non-real-time computing power resources to each non-real-time access node in the non-real-time business container area. By adopting the above technical solution, different terminal nodes in the access edge nodes are divided into real-time terminal nodes and non-real-time terminal nodes according to the types of services required to be implemented in the terminal nodes, and then real-time computing power prediction is performed for the real-time terminal nodes, and the total computing power resources are divided into a real-time business container area that reserves and allocates full computing power resources for real-time terminal nodes and a non-real-time business container area that provides partial computing power resources for non-real-time terminal nodes, that is, a reservation-based coordination mechanism is adopted for real-time services. For non-real-time services, the non-real-time access nodes that can provide computing power in different time slots in the next business cycle are determined based on the non-real-time computing power resources that can be provided to non-real-time services in the non-real-time business container area and the node weights of different non-real-time terminal nodes, that is, a competitive coordination mechanism is adopted for non-real-time services, and computing power resources are provided for the most important non-real-time services in the same time period, ensuring that more important services can be executed in a timely manner, and the computing power resources of the edge nodes can be fully utilized, so that distributed services can be flexibly deployed, and the processing capabilities of the power Internet of Things network for services are improved.

[0048] Embodiment 2

[0049] Figure 2A flow chart of an edge collaborative computing power allocation method provided for embodiment 2 of the present invention. The embodiment of the present invention is further optimized on the basis of the above-mentioned optional technical solutions, and directly completes the classification and prediction processing of business flow data provided by each terminal node through a business classification prediction model including a business classification sub-model and a computing power prediction sub-model, thereby reducing the process complexity of data processing. At the same time, the real-time computing power demand is pre-determined based on the business arrival volume of each real-time terminal node, the average business data volume of each real-time terminal node, and the real-time computing power prediction value, and the computing power resources of the real-time computing power demand are divided into the real-time business container area on the edge computing device to realize computing power reservation for real-time business. The computing power resources in the edge computing device other than the real-time computing power demand are used as the non-real-time computing power resources of the non-real-time business container area, and the next business cycle is divided into multiple time slots. Each non-real-time terminal node adopts a competitive access mechanism. According to their respective node weights and the uplink transmission upper limit and concurrent processing upper limit of the non-real-time terminal node to the edge computing device, the non-real-time access node that should be accessed in each time slot is determined, and the non-real-time computing power resources that can be allocated to the non-real-time access node are determined based on the historical computing power resources and predicted computing power resources of each non-real-time access node, or by random allocation. This ensures that more important businesses can be executed in a timely manner, the computing power resources of the edge nodes can be fully utilized, and distributed businesses can be flexibly deployed, thereby improving the business processing capabilities of the power Internet of Things network.

[0050] like Figure 2 As shown, an edge-end collaborative computing power allocation method provided by an embodiment of the present invention specifically includes the following steps:

[0051] S201. Obtain the business logistics information of each terminal node in the current business cycle.

[0052] S202. For each piece of business logistics information, the business logistics information is input into the business classification sub-model in the pre-trained business classification prediction model to determine the business type of the terminal node corresponding to the business logistics information.

[0053] In this embodiment, the business classification sub-model can be specifically understood as a neural network sub-model used in the business classification prediction model to extract features and classify the business logistics information input therein, and determine whether the business type corresponding to the business logistics information is a real-time business type or a non-real-time business type based on the classification results. Optionally, the business classification sub-model can be a probabilistic classification model based on Bayes' theorem.

[0054] For example, assume that a service flow information has n features, namely {u 1 ,u 2 ,...,u n}, and each logistics feature can be classified into real-time business x1 and non-real-time services x 2 Then, by using the business classification sub-model to classify it, the probability that the business logistics information belongs to real-time business can be determined. and the probability of belonging to non-real-time business The one with a greater probability between the two is taken as the business type corresponding to the logistics information.

[0055] Specifically, for each business logistics information, the business logistics information is input into the business classification sub-model in the pre-trained business classification prediction model, and the business classification sub-model is used to extract features of the business flow information to obtain multiple business features, and each business feature is classified to determine whether it is a business feature of real-time business or a business feature of non-real-time business. After the types of all business features are determined, the number of real-time business features and the number of total business features can be determined as the probability that the business logistics information belongs to real-time business, and the number of non-real-time business features and the number of total business features can be determined as the probability that the business logistics information belongs to non-real-time business. The two probabilities can be compared, and the business type with a larger probability can be determined as the business type to which the business flow information belongs.

[0056] S203. When the business type of the terminal node is a real-time business type, the terminal node corresponding to the business logistics information is determined as a real-time terminal node, and the business logistics information is input into the computing power prediction sub-model in the pre-trained business classification prediction model to determine the real-time computing power prediction value of the real-time terminal node.

[0057] In this embodiment, the computing power prediction sub-model can be specifically understood as a neural network sub-model used in the business classification prediction model to iteratively predict the business features input therein to determine the computing power demand value corresponding to the business. Optionally, the computing power prediction sub-model can be a GRU model based on time series prediction, which is not limited in this embodiment of the present invention.

[0058] For example, taking the computing power prediction sub-model as a GRU model, Figure 3 This is a structural example diagram of a GRU unit provided in the second embodiment of the present invention, and its state equation can be expressed as:

[0059]

[0060] Among them, y t-1 is the hidden layer output of the previous GRU unit, x t is the business feature of this round of input, σ is the sigmoid function, z t is the update gate of the GRU unit, y t ' is the updated intermediate state, y t is the output of this round of GRU units, U z is the weight of the update gate, Wz is the weight of the intermediate state, b z is the offset.

[0061] Specifically, after the business classification sub-model classifies the business flow information input therein, if it is determined that the business type of the terminal node corresponding to the business flow information is a non-real-time business type, it can be considered that there is no need to predict the computing power resources required for the terminal node in the next business cycle, and at this time, there is no need to input the business features extracted from the business flow information into the computing power prediction sub-model in the pre-trained business classification prediction model. When the business type of the terminal node is a real-time business type, the terminal node corresponding to the business flow information can be first determined as a real-time terminal node, and the business features extracted from the business flow information can be input into the computing power prediction sub-model in the pre-trained business classification prediction model to perform real-time computing power prediction, and the output of the computing power prediction sub-model is determined as the real-time computing power prediction value of the real-time terminal node.

[0062] S204: Determine the business computing power requirement based on the business arrival volume of each real-time terminal node and the average business data volume of each real-time terminal node.

[0063] In this embodiment, the service arrival volume can be specifically understood as the number of real-time services that can reach the edge computing device from the real-time terminal node within a service cycle. The average service data volume can be specifically understood as the average amount of data transmitted from the real-time services in each real-time terminal node to the edge computing device. The service computing power demand can be specifically understood as the amount of computing power resources required by the edge computing device to process the real-time services of each real-time terminal node within a service cycle.

[0064] Specifically, based on the business arrival volume of each real-time terminal node and the average business data volume of each real-time terminal node, the amount of real-time business data that the edge computing device will receive from each real-time terminal node at each moment in a business cycle is determined, and it is determined as the business computing power demand.

[0065] For example, the business computing power requirement can be expressed by the following formula:

[0066]

[0067] Among them, T is the next business cycle, Y t is the traffic arrival volume at time t, D t is the average business data volume at time t, and η is the proportional factor between the business data volume and the computing power demand.

[0068] S205. Determine a maximum real-time computing power prediction error according to the real-time computing power prediction value and the prediction error corresponding to the real-time computing power prediction value.

[0069] For example, assume that the real-time computing power prediction value is yt , ε t is the prediction error of the real-time computing power prediction value at time t within period T, then the maximum real-time computing power prediction error can be expressed as

[0070] S206. Determine the sum of the business computing power demand and the maximum real-time computing power prediction error as the real-time computing power demand.

[0071] S207. The computing power resources in the total computing power resources corresponding to the real-time computing power demand are used as real-time computing power resources, and a real-time business container area with a resource amount of real-time computing power resources is constructed.

[0072] Specifically, in order to ensure that real-time services can be executed in a timely manner in the next service cycle, when the total computing power resources are sufficient, the computing power resources corresponding to the real-time computing power demand can be used as real-time computing power resources, and a real-time business container area can be divided in the edge computing device, and the determined real-time computing power resources can be used as the computing power resources of the real-time business container area.

[0073] It is understandable that when the total computing power resources are insufficient to support the real-time computing power demand, all the total computing power resources can be used as real-time computing power resources; or a maximum resource ratio that can be provided to real-time services can be divided from the total computing power resources of the edge computing device, and all computing power resources within this resource ratio can be used as real-time computing power resources. The method for determining real-time computing power resources can also be set according to other actual needs, and the embodiments of the present invention are not limited to this.

[0074] S208. The computing resources other than the real-time computing resources in the total computing resources are used as non-real-time computing resources, and a non-real-time business container area with a resource amount of non-real-time computing resources is constructed.

[0075] Specifically, after meeting the needs of real-time business, in order to make full use of the computing power resources in the edge computing device, the computing power resources other than the real-time computing power resources in the total computing power resources can be used as non-real-time computing power resources, and a non-real-time business container area with a resource amount of non-real-time computing power resources can be constructed to carry the operation of non-real-time business in the next business cycle through the non-real-time business container area.

[0076] S209: Determine the upper limit of uplink transmission and the upper limit of concurrent processing according to the average bandwidth demand and average computing power demand of each non-real-time terminal node, the non-real-time computing power resources and the total bandwidth of the non-real-time service container area.

[0077] In this embodiment, the average bandwidth requirement can be specifically understood as the average bandwidth required by each non-real-time terminal node when transmitting business logistics information to the edge computing device. The average computing power requirement can be specifically understood as the average amount of computing power resources required for processing non-real-time services in each non-real-time terminal node. The uplink transmission upper limit can be specifically understood as the upper limit of the number of concurrent access services for uplink transmission of terminal node data. The concurrent processing upper limit can be specifically understood as the upper limit of the number of services concurrently processed by the computing power of the edge computing device.

[0078] For example, suppose B total is the total bandwidth, B avg is the average bandwidth requirement, C total is the non-real-time computing resource, ηD avg is the average computing power requirement, then the upper limit of uplink transmission is M 1 It can be expressed as:

[0079] Concurrent processing limit M 2 It can be expressed as:

[0080] It can be understood that after determining the uplink transmission upper limit and the concurrent processing upper limit, the method further includes: broadcasting the uplink transmission upper limit and the concurrent processing upper limit to each non-real-time terminal node.

[0081] Specifically, since each non-real-time terminal node itself also needs to determine the access time slot required for itself in the next business cycle, after the edge computing device clearly defines the uplink transmission upper limit and the concurrent processing upper limit, the above two types of information need to be broadcast to each non-real-time terminal node, so that each non-real-time terminal node can calculate and determine the access time slot required for itself based on the obtained uplink transmission upper limit and concurrent processing upper limit.

[0082] S210. For each non-real-time terminal node, determine the probability of contention access of the non-real-time terminal node in each time slot in the next service cycle according to the uplink transmission upper limit, the concurrent processing upper limit, the total number of time slots corresponding to the next service cycle and the node weight of the non-real-time terminal node.

[0083] Continuing with the above example, assuming that the next service cycle T can be divided into N time slots, then given the known uplink transmission upper limit M 1 and the concurrent processing limit M 2 When an intermediate parameter can be determined Then, for each non-real-time terminal node, its probability of competing for access in the i-th time slot can be calculated: i for:

[0084]

[0085] Among them, i∈[1,N], ω iis the node weight of the non-real-time terminal node in the i-th time slot.

[0086] It can be understood that after the calculation of S210, the edge computing node can clearly determine the contention access probability of each non-real-time terminal node in each time slot in the next service cycle, that is, assuming that there are m non-real-time terminal nodes, the edge computing node can calculate m*N contention access probabilities.

[0087] S211: Determine the non-real-time terminal node as the non-real-time access node of the time slot corresponding to the maximum contention access probability.

[0088] Specifically, since the edge computing node can determine the contention access probability of each non-real-time terminal node in different time slots, it can be considered that the time slot corresponding to the maximum contention access probability is the time slot that the non-real-time terminal node most needs to access in the next business cycle. At this time, the non-real-time terminal node can be determined as a non-real-time access node for the time slot corresponding to its maximum contention access probability.

[0089] It is understandable that, since different non-real-time terminal nodes have different node weights, the time slots corresponding to the maximum contention access probability of each non-real-time terminal node should be different, that is, the maximum contention access probability can be used to determine a unique non-real-time access node for different time slots in the next service cycle.

[0090] It is understandable that each non-real-time terminal node itself also needs to calculate the contention access probability of all time slots in the next business cycle, and access the edge computing device as a non-real-time access node when reaching the time slot corresponding to the maximum contention access probability.

[0091] S212: Allocate non-real-time computing resources to each non-real-time access node in the non-real-time service container area.

[0092] Optionally, the manner of allocating non-real-time computing resources to each non-real-time access node in the non-real-time service container area includes at least one of the following:

[0093] 1) According to the historical computing power resources of each non-real-time access node, determine the non-real-time computing power demand corresponding to each non-real-time access node, and allocate non-real-time computing power resources to each non-real-time access node according to each non-real-time computing power demand;

[0094] 2) predicting the computing power demand of each non-real-time access node, determining the non-real-time computing power demand corresponding to each non-real-time access node, and allocating non-real-time computing power resources to each non-real-time access node according to each non-real-time computing power demand;

[0095] 3) Randomly allocate non-real-time computing resources to each non-real-time access node.

[0096] Specifically, any one of the above three methods can be used to determine the non-real-time computing power resources allocated to the corresponding non-real-time access node by the non-real-time service container area when it reaches the corresponding non-real-time access node access time slot. If the first method is used, the historical computing power resources of the non-real-time access node accessing the time slot can be averaged at each time slot, or the historical computing power resources with the highest frequency can be taken according to the normal distribution as the non-real-time computing power demand; if the second method is used, the historical computing power resources of the non-real-time access node accessing the time slot can be input into the pre-trained computing power resource prediction model for prediction at each time slot, and the obtained output can be determined as the non-real-time computing power demand of the corresponding non-real-time access node; if the third method is used, a value within the allowable range of non-real-time computing power resources can be directly randomly determined at each time slot as the non-real-time computing power demand of the corresponding non-real-time access node. And when the non-real-time computing power demand is completed, the non-real-time computing power resources are allocated to the corresponding non-real-time access node.

[0097] The technical solution of this embodiment directly completes the classification and prediction processing of business flow data provided by each terminal node through a business classification prediction model including a business classification sub-model and a computing power prediction sub-model, thereby reducing the complexity of the data processing process. At the same time, the real-time computing power demand is pre-determined based on the business arrival volume of each real-time terminal node, the average business data volume of each real-time terminal node, and the real-time computing power prediction value, and the computing power resources of the real-time computing power demand are divided into the real-time business container area on the edge computing device to realize the computing power reservation for the real-time business. The computing power resources in the edge computing device other than the real-time computing power demand are used as the non-real-time computing power resources of the non-real-time business container area, and the next business cycle is divided into multiple time slots. Each non-real-time terminal node adopts a competitive access mechanism. According to their respective node weights and the uplink transmission upper limit and concurrent processing upper limit of the non-real-time terminal node to the edge computing device, the non-real-time access node that should be accessed in each time slot is determined, and the non-real-time computing power resources that can be allocated to the non-real-time access node are determined based on the historical computing power resources and predicted computing power resources of each non-real-time access node, or by random allocation. This ensures that more important businesses can be executed in a timely manner, the computing power resources of the edge nodes can be fully utilized, and distributed businesses can be flexibly deployed, thereby improving the business processing capabilities of the power Internet of Things network.

[0098] Embodiment 3

[0099] Figure 4 A schematic diagram of the structure of an edge-end collaborative computing power allocation device provided in Embodiment 3 of the present invention is shown in FIG. Figure 4 As shown, the edge collaborative computing power allocation device includes an information acquisition module 31, a real-time computing power prediction module 32, a business container division module 33 and a non-real-time computing power allocation module 34.

[0100] Among them, the information acquisition module 31 is used to obtain the business logistics information of each terminal node in the current business cycle; the real-time computing power prediction module 32 is used to input the business logistics information of each terminal node into the pre-trained business classification prediction model to determine the real-time terminal node and the real-time computing power prediction value; the business container division module 33 is used to determine the real-time computing power demand according to the real-time terminal node and the real-time computing power prediction value, and divide the total computing power resources into a real-time business container area and a non-real-time business container area according to the real-time computing power demand, so as to allocate the computing power resources corresponding to the real-time computing power demand to the real-time business container area; the non-real-time computing power allocation module 34 is used to determine the non-real-time access nodes of each time slot corresponding to the next business cycle according to the non-real-time computing power resources of the non-real-time business container area and the node weights of each non-real-time terminal node, and allocate non-real-time computing power resources to each non-real-time access node in the non-real-time business container area.

[0101] The technical solution of the embodiment of the present invention divides different terminal nodes in the access edge node into real-time terminal nodes and non-real-time terminal nodes according to the type of business required to be implemented in the terminal node, and then predicts the real-time computing power for the real-time terminal node, and divides the total computing power resources into a real-time business container area that reserves and allocates the full amount of computing power resources for the real-time terminal node and a non-real-time business container area that provides part of the computing power resources for the non-real-time terminal node, that is, a reservation-based coordination mechanism is adopted for real-time services. For non-real-time services, the non-real-time access nodes that can provide computing power in different time slots in the next business cycle are determined based on the non-real-time computing power resources that can be provided to the non-real-time services in the non-real-time business container area and the node weights of different non-real-time terminal nodes, that is, a competitive coordination mechanism is adopted for non-real-time services, and computing power resources are provided for the most important non-real-time services in the same time period, ensuring that more important services can be executed in time, and the computing power resources of the edge nodes can be fully utilized, so that distributed services can be flexibly deployed, and the processing capability of the power Internet of Things network for services is improved.

[0102] Optionally, the business classification prediction model includes a business classification sub-model and a computing power prediction sub-model; the real-time computing power prediction module 32 is specifically used for:

[0103] For each business logistics information, the business logistics information is input into the business classification sub-model in the pre-trained business classification prediction model to determine the business type of the terminal node corresponding to the business logistics information;

[0104] When the business type of the terminal node is a real-time business type, the terminal node corresponding to the business logistics information is determined as the real-time terminal node, and the business logistics information is input into the computing power prediction sub-model in the pre-trained business classification prediction model to determine the real-time computing power prediction value of the real-time terminal node.

[0105] Optionally, the service container division module 33 is specifically used to:

[0106] Determine the business computing power requirement based on the business arrival volume of each real-time terminal node and the average business data volume of each real-time terminal node;

[0107] Determine a maximum real-time computing power prediction error according to the real-time computing power prediction value and the prediction error corresponding to the real-time computing power prediction value;

[0108] The sum of the business computing power demand and the maximum real-time computing power prediction error is determined as the real-time computing power demand;

[0109] The computing power resources corresponding to the real-time computing power demand in the total computing power resources are used as real-time computing power resources, and a real-time business container area with the resource amount of real-time computing power resources is constructed;

[0110] The computing resources other than the real-time computing resources in the total computing resources are regarded as non-real-time computing resources, and a non-real-time business container area with the resource amount of non-real-time computing resources is constructed.

[0111] Optionally, the non-real-time computing power allocation module 34 is specifically used for:

[0112] Determine the upper limit of uplink transmission and concurrent processing based on the average bandwidth demand and average computing power demand of each non-real-time terminal node, the non-real-time computing power resources and total bandwidth of the non-real-time service container area;

[0113] For each non-real-time terminal node, determine the probability of contention access of the non-real-time terminal node in each time slot in the next service cycle according to the uplink transmission upper limit, the concurrent processing upper limit, the total number of time slots corresponding to the next service cycle and the node weight of the non-real-time terminal node;

[0114] The non-real-time terminal node is determined as the non-real-time access node of the time slot corresponding to the maximum contention access probability.

[0115] Optionally, after determining the uplink transmission upper limit and the concurrent processing upper limit, the following is further included:

[0116] The upstream transmission upper limit and the concurrent processing upper limit are broadcasted to each non-real-time terminal node.

[0117] Optionally, allocating non-real-time computing resources to each non-real-time access node in the non-real-time service container area includes at least one of the following:

[0118] According to the historical computing power resources of each non-real-time access node, determine the non-real-time computing power demand corresponding to each non-real-time access node, and allocate non-real-time computing power resources to each non-real-time access node according to each non-real-time computing power demand;

[0119] Predict the computing power demand of each non-real-time access node, determine the non-real-time computing power demand corresponding to each non-real-time access node, and allocate non-real-time computing power resources to each non-real-time access node according to each non-real-time computing power demand;

[0120] Randomly allocate non-real-time computing resources to each non-real-time access node.

[0121] The edge collaborative computing power allocation device provided in an embodiment of the present invention can execute the edge collaborative computing power allocation method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0122] Embodiment 4

[0123] Figure 5 A schematic diagram of the structure of an edge collaborative computing power allocation device provided for Embodiment 4 of the present invention. The edge collaborative computing power allocation device 40 may be intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The edge collaborative computing power allocation device 40 may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are intended to be examples only and are not intended to limit the implementation of the present invention described and / or required herein.

[0124] like Figure 5 As shown, the edge collaborative computing power distribution device 40 includes at least one processor 41, and a memory connected to the at least one processor 41, such as a read-only memory (ROM) 42, a random access memory (RAM) 43, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 41 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 42 or the computer program loaded from the storage unit 48 to the random access memory (RAM) 43. In RAM 43, various programs and data required for the operation of the edge collaborative computing power distribution device 40 can also be stored. The processor 41, ROM 42 and RAM 43 are connected to each other via a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0125] Multiple components in the edge collaborative computing power distribution device 40 are connected to the I / O interface 45, including: an input unit 46, such as a keyboard, a mouse, etc.; an output unit 47, such as various types of displays, speakers, etc.; a storage unit 48, such as a disk, an optical disk, etc.; and a communication unit 49, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 49 allows the edge collaborative computing power distribution device 40 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0126] The processor 41 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 41 executes the various methods and processes described above, such as the edge-end collaborative computing power allocation method.

[0127] In some embodiments, the edge collaborative computing power allocation method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 48. In some embodiments, part or all of the computer program can be loaded and / or installed on the computing power allocation device 40 via ROM 42 and / or communication unit 49. When the computer program is loaded into RAM 43 and executed by processor 41, one or more steps of the edge collaborative computing power allocation method described above can be performed. Alternatively, in other embodiments, processor 41 can be configured to execute the edge collaborative computing power allocation method in any other appropriate manner (for example, by means of firmware).

[0128] Optionally, an embodiment of the present invention further provides a computer program product, including a computer program, which, when executed by a processor, implements the edge collaborative computing power allocation method provided in any embodiment of the present invention.

[0129] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0130] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0131] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0132] In order to provide interaction with the user, the systems and techniques described herein can be implemented on a computing power distribution device, which has: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball), through which the user can provide input to the computing power distribution device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0133] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0134] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0135] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0136] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for allocating edge collaborative computing power, characterized in that: include: Obtain the business logistics information of each terminal node in the current business cycle; Input the logistics information of each business into the pre-trained business classification prediction model to determine the real-time terminal node and real-time computing power prediction value; Determine the real-time computing power demand according to the real-time terminal node and the real-time computing power prediction value, and divide the total computing power resources into a real-time business container area and a non-real-time business container area according to the real-time computing power demand, so as to allocate the computing power resources corresponding to the real-time computing power demand to the real-time business container area; According to the non-real-time computing power resources of the non-real-time service container area and the node weight of each non-real-time terminal node, the non-real-time access nodes of each time slot corresponding to the next service cycle are determined, and non-real-time computing power resources are allocated to each non-real-time access node in the non-real-time service container area.

2. The method according to claim 1, characterized in that The business classification prediction model includes a business classification sub-model and a computing power prediction sub-model; The inputting of the logistics information of each business into the pre-trained business classification prediction model to determine the real-time terminal node and the real-time computing power prediction value includes: For each of the business logistics information, the business logistics information is input into the business classification sub-model in the pre-trained business classification prediction model to determine the business type of the terminal node corresponding to the business logistics information; When the business type of the terminal node is a real-time business type, the terminal node corresponding to the business logistics information is determined as a real-time terminal node, and the business logistics information is input into the computing power prediction sub-model in the pre-trained business classification prediction model to determine the real-time computing power prediction value of the real-time terminal node.

3. The method according to claim 1, characterized in that: The determining the real-time computing power requirement according to the real-time terminal node and the real-time computing power prediction value includes: Determine the business computing power requirement according to the business arrival volume of each of the real-time terminal nodes and the average business data volume of each of the real-time terminal nodes; Determining a maximum real-time computing power prediction error according to the real-time computing power prediction value and a prediction error corresponding to the real-time computing power prediction value; The sum of the business computing power demand and the maximum real-time computing power prediction error is determined as the real-time computing power demand.

4. The method according to claim 3, characterized in that The dividing of the total computing power resources into a real-time service container area and a non-real-time service container area according to the real-time computing power demand includes: The computing power resources corresponding to the real-time computing power demand in the total computing power resources are used as real-time computing power resources, and a real-time business container area with a resource amount equal to the real-time computing power resources is constructed; The computing power resources other than the real-time computing power resources in the total computing power resources are used as non-real-time computing power resources, and a non-real-time business container area with a resource amount equal to the non-real-time computing power resources is constructed.

5. The method according to claim 1, characterized in that The determining, according to the non-real-time computing power resources of the non-real-time service container area and the node weights of each non-real-time terminal node, the non-real-time access node for each time slot corresponding to the next service cycle includes: Determine the uplink transmission upper limit and the concurrent processing upper limit according to the average bandwidth demand and average computing power demand of each non-real-time terminal node, the non-real-time computing power resources and the total bandwidth of the non-real-time service container area; For each non-real-time terminal node, determine the probability of contention access of the non-real-time terminal node in each time slot in the next service cycle according to the uplink transmission upper limit, the concurrent processing upper limit, the total number of time slots corresponding to the next service cycle, and the node weight of the non-real-time terminal node; The non-real-time terminal node is determined as a non-real-time access node in the time slot corresponding to the maximum contention access probability.

6. The method according to claim 5, characterized in that After determining the uplink transmission upper limit and the concurrent processing upper limit, the method further includes: The uplink transmission upper limit and the concurrent processing upper limit are broadcasted and sent to each of the non-real-time terminal nodes.

7. The method according to claim 1, characterized in that The allocating non-real-time computing resources to each of the non-real-time access nodes in the non-real-time service container area includes at least one of the following: Determine the non-real-time computing power requirements corresponding to each of the non-real-time access nodes according to the historical computing power resources of each of the non-real-time access nodes, and allocate non-real-time computing power resources to each of the non-real-time access nodes according to each of the non-real-time computing power requirements; Predicting computing power requirements for each of the non-real-time access nodes, determining the non-real-time computing power requirements corresponding to each of the non-real-time access nodes, and allocating non-real-time computing power resources to each of the non-real-time access nodes according to the non-real-time computing power requirements; Randomly allocate non-real-time computing resources to each of the non-real-time access nodes.

8. An edge collaborative computing power allocation device, characterized in that: include: The information acquisition module is used to obtain the business logistics information of each terminal node in the current business cycle; A real-time computing power prediction module is used to input the logistics information of each business into a pre-trained business classification prediction model to determine the real-time terminal node and the real-time computing power prediction value; A business container division module, used to determine the real-time computing power demand according to the real-time terminal node and the real-time computing power prediction value, and divide the total computing power resources into a real-time business container area and a non-real-time business container area according to the real-time computing power demand, so as to allocate the computing power resources corresponding to the real-time computing power demand to the real-time business container area; The non-real-time computing power allocation module is used to determine the non-real-time access nodes for each time slot corresponding to the next business cycle according to the non-real-time computing power resources of the non-real-time business container area and the node weights of each non-real-time terminal node, and allocate non-real-time computing power resources to each non-real-time access node in the non-real-time business container area.

9. The device according to claim 8, characterized in that The business classification prediction model includes a business classification sub-model and a computing power prediction sub-model; the real-time computing power prediction module is specifically used to: For each of the business logistics information, the business logistics information is input into the business classification sub-model in the pre-trained business classification prediction model to determine the business type of the terminal node corresponding to the business logistics information; When the business type of the terminal node is a real-time business type, the terminal node corresponding to the business logistics information is determined as a real-time terminal node, and the business logistics information is input into the computing power prediction sub-model in the pre-trained business classification prediction model to determine the real-time computing power prediction value of the real-time terminal node.

10. The device according to claim 8, characterized in that The service container division module is specifically used for: Determine the business computing power requirement according to the business arrival volume of each of the real-time terminal nodes and the average business data volume of each of the real-time terminal nodes; Determining a maximum real-time computing power prediction error according to the real-time computing power prediction value and a prediction error corresponding to the real-time computing power prediction value; Determine the sum of the business computing power requirement and the maximum real-time computing power prediction error as the real-time computing power requirement; The computing power resources corresponding to the real-time computing power demand in the total computing power resources are used as real-time computing power resources, and a real-time business container area with a resource amount equal to the real-time computing power resources is constructed; The computing power resources other than the real-time computing power resources in the total computing power resources are used as non-real-time computing power resources, and a non-real-time business container area with a resource amount equal to the non-real-time computing power resources is constructed.

11. The device according to claim 8, characterized in that The non-real-time computing power allocation module is specifically used for: Determine the uplink transmission upper limit and the concurrent processing upper limit according to the average bandwidth demand and average computing power demand of each non-real-time terminal node, the non-real-time computing power resources and the total bandwidth of the non-real-time service container area; For each non-real-time terminal node, determine the probability of contention access of the non-real-time terminal node in each time slot in the next service cycle according to the uplink transmission upper limit, the concurrent processing upper limit, the total number of time slots corresponding to the next service cycle, and the node weight of the non-real-time terminal node; The non-real-time terminal node is determined as a non-real-time access node in the time slot corresponding to the maximum contention access probability.

12. An edge collaborative computing power distribution device, characterized in that: include: at least one processor; and a memory communicatively coupled to the at least one processor; In which, the memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the edge collaborative computing power allocation method described in any one of claims 1-7.

13. A storage medium containing computer executable instructions, characterized in that: The computer executable instructions, when executed by a computer processor, are used to execute the edge collaborative computing power allocation method as described in any one of claims 1-7.

14. A computer program product, comprising a computer program, which, when executed by a processor, implements the edge collaborative computing power allocation method as described in any one of claims 1-7.

Images