Edge computing load balancing method based on artificial intelligence

Abstract

The invention discloses an artificial intelligence-based edge computing load balancing method, which belongs to the technical field of edge computing, and comprises the following steps of: obtaining an edge computing task queue, sequentially taking out a plurality of target edge computing tasks from the edge computing task queue according to a fixed number, and then aiming at the plurality of target edge computing tasks, carrying out load balancing on the target edge computing tasks; carrying out allocation relation coding on the edge computing server by adopting an intelligent coding method to obtain a plurality of task allocation codes, and carrying out load balance evaluation by adopting an artificial intelligence algorithm to determine the load balance corresponding to the task allocation codes; and finally, according to the load balance corresponding to the task allocation codes, determining a current optimal code, optimizing other task allocation codes according to the current optimal code, obtaining a final task allocation code, and on the basis of the final task allocation code, performing task allocation. And the target edge computing task is allocated to the edge computing server for processing, so that load balancing of edge computing is realized, overload of the edge computing server can be effectively avoided, and the overall data processing efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the field of edge computing technology, and specifically relates to an edge computing load balancing method based on artificial intelligence. Background Art

[0002] With the increasing popularity of Internet of Things (IoT) devices and the rapid development of 5G networks, edge computing, as a new computing model, is playing an increasingly important role in providing low-latency, high-bandwidth services. Edge computing deploys computing and data storage resources at the edge of the network, close to users and devices, significantly reducing data transmission latency and improving response speeds. However, resources in edge computing environments are limited, and how to efficiently allocate and schedule these resources to achieve load balancing and improve overall system performance and reliability has become a pressing issue. Existing technologies generally place edge computing tasks directly on the nearest edge computing server for execution. While this can improve response speed, it can also lead to a situation where a single edge computing server is overloaded with tasks, resulting in reduced processing speeds. Summary of the Invention

[0003] The present invention provides an edge computing load balancing method based on artificial intelligence to solve the technical problems existing in the prior art.

[0004] An edge computing load balancing method based on artificial intelligence, comprising: Obtain an edge computing task queue, and sequentially take out a plurality of target edge computing tasks from the edge computing task queue according to a fixed number; For multiple target edge computing tasks, an intelligent coding method is used to encode the allocation relationship of edge computing servers and obtain multiple task allocation codes; For any task assignment code, an artificial intelligence algorithm is used to evaluate the load balancing and determine the load balancing corresponding to the task assignment code; Determine the current optimal code based on the load balancing of the task assignment code, and optimize other task assignment codes based on the current optimal code to obtain the final task assignment code; Based on the final task allocation code, the target edge computing task is allocated to the edge computing server for processing to achieve load balancing of edge computing.

[0005] Furthermore, for multiple target edge computing tasks, an intelligent coding method is used to encode the allocation relationship of edge computing servers to obtain multiple task allocation codes, including: For multiple target edge computing tasks, randomly generate a value in the (0,1] range, encode the value corresponding to the target edge computing task into a vector, and determine the task assignment code; Repeat the encoding multiple times to obtain multiple task assignment codes.

[0006] Furthermore, after encoding the allocation relationship of the edge computing server using the intelligent encoding method, the method further includes: Evenly divide the (0,1] numerical range into multiple sub-target intervals; the number of sub-target intervals is the same as the number of edge computing servers; Randomly match the sub-target interval for each edge computing server and determine the sub-target interval corresponding to the edge server; wherein the edge computing server and the sub-target interval are in a one-to-one correspondence; Determine the edge computing server corresponding to the sub-target interval where the value corresponding to a target edge computing task in the task allocation code is located as the target edge computing server of the target edge computing task.

[0007] Furthermore, for any task allocation code, an artificial intelligence algorithm is used to perform load balancing evaluation to determine the load balancing corresponding to the task allocation code, including: For any task assignment code, determine the computer resource requirements corresponding to the target edge computing task corresponding to the task assignment code and the occupied computer resources corresponding to each edge computing server; Assign the task to the target edge computing task corresponding to the code value, and determine the target edge computing server corresponding to each target edge computing task; Add the computer resources required for the target edge computing task to the occupied computer resources of the corresponding target edge computing server to obtain the predicted occupied computer resources corresponding to each edge computing server; The predicted occupied computer resources of all edge computing servers are used as input to the pre-deployed artificial intelligence algorithm to obtain the load balancing corresponding to the task allocation code; The load balancing is used to characterize the inverse of the variance of computer resource usage, that is, the greater the load balancing of the edge computing server, the more balanced the load.

[0008] Furthermore, it also includes: For any task assignment code, determine one by one whether each target edge computing task satisfies the constraint conditions after being assigned to the corresponding target edge computing server. If so, continue to judge until each target edge computing task in the task assignment code is judged to be completed. Otherwise, the target edge computing task that does not meet the constraint conditions will be assigned to other edge computing servers until each target edge computing task in the task assignment code is judged to be completed. Among them, the constraint condition is set as: after the target edge computing task is assigned to the edge computing server, the predicted computer resources occupied by the edge computing server do not exceed its resource upper limit.

[0009] Furthermore, the target edge computing tasks that do not meet the constraints are assigned to other edge computing servers, including: Determine the edge computing server with the lowest computer resource utilization rate, and assign the target edge computing tasks that do not meet the constraints to the edge computing server with the lowest computer resource utilization rate.

[0010] Furthermore, the deployment method of the artificial intelligence algorithm includes: The historical occupied computer resources and historical computer resource utilization rates after historical edge computing tasks are assigned to edge computing servers are collected, and the historical occupied computer resources are used as optimization samples. The inverse of the variance of the historical computer resource utilization rate is used as the expected output to deploy the artificial intelligence algorithm.

[0011] Furthermore, according to the load balancing corresponding to the task allocation code, the current optimal code is determined, and other task allocation codes are optimized according to the current optimal code to obtain the final task allocation code, including: According to the load balance corresponding to the task allocation code, the task allocation code with the greatest load balance is determined as the current optimal code; Based on the current optimal code, the particle swarm algorithm is used to optimize other task allocation codes to obtain the final task allocation code.

[0012] Furthermore, based on the final task allocation code, the target edge computing task is allocated to the edge computing server for processing, including: Determine the sub-goal interval of the value corresponding to each target edge computing task in the final task allocation code, and obtain the sub-goal interval corresponding to each target edge computing task; Determine the edge computing server corresponding to the sub-target interval corresponding to the target edge computing task as the final edge computing server of the target edge computing task; The target edge computing task is assigned to the corresponding final edge computing server for processing, and each final edge computing server processes the target edge computing task in the chronological order.

[0013] Furthermore, after allocating the target edge computing task to the edge computing server for processing, the method further includes: If an edge computing server is overloaded, the overloaded edge computing server will not participate in the execution of other edge computing tasks in the next task processing cycle.

[0014] The present invention provides an artificial intelligence-based edge computing load balancing method, which obtains an edge computing task queue and sequentially takes out multiple target edge computing tasks from the edge computing task queue according to a fixed number, then uses an intelligent coding method to encode the allocation relationship of the edge computing server for the multiple target edge computing tasks, obtains multiple task allocation codes, and uses an artificial intelligence algorithm to perform load balancing evaluation to determine the load balancing corresponding to the task allocation code; finally, according to the load balancing corresponding to the task allocation code, the current optimal code is determined, and other task allocation codes are optimized according to the current optimal code to obtain the final task allocation code, and based on the final task allocation code, the target edge computing task is allocated to the edge computing server for processing, thereby realizing edge computing load balancing, effectively avoiding overload of the edge computing server, and improving the overall data processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0016] Figure 1 A flowchart of an edge computing load balancing method based on artificial intelligence provided by an embodiment of the present invention.

[0017] The above drawings illustrate specific embodiments of the present invention, which will be described in more detail below. These drawings and the accompanying description are not intended to limit the scope of the present invention in any way, but rather to illustrate the concept of the present invention to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0018] Exemplary embodiments will be described in detail herein, examples of which are illustrated in the accompanying drawings. In the following description, when referring to the drawings, like numbers in different figures represent like or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present invention. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present invention, as detailed in the appended claims.

[0019] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0020] like Figure 1 As shown, an embodiment of the present invention provides an edge computing load balancing method based on artificial intelligence, including: S1. Obtain an edge computing task queue, and sequentially take out a plurality of target edge computing tasks from the edge computing task queue according to a fixed number; When edge computing tasks arrive, they can be placed in the edge computing task queue one by one in the chronological order of the edge computing tasks. Then, each time an edge computing task is executed, multiple target edge computing tasks are taken out in a fixed number for execution.

[0021] S2. For multiple target edge computing tasks, use an intelligent coding method to encode the allocation relationship of edge computing servers and obtain multiple task allocation codes; Traditional methods for allocating edge computing tasks generally use binary coding combined with genetic algorithms for optimization, or optimization through reinforcement learning. Although these methods can achieve load balancing to a certain extent, they are difficult to achieve optimal load balancing or require time to gradually achieve load balancing. Therefore, the embodiments of the present invention use intelligent coding methods to encode the allocation relationship of edge computing servers, which can achieve continuous optimization of task allocation coding and ultimately find the optimal allocation method.

[0022] S3. For any task assignment code, use an artificial intelligence algorithm to evaluate the load balancing and determine the load balancing corresponding to the task assignment code; Using artificial intelligence algorithms to assess load balancing can directly collect features for identification, enabling rapid data recognition. Load uniformity is used to indicate whether the load across the entire edge computing server cluster is balanced. A higher load uniformity indicates a more balanced load.

[0023] S4. Determine the current optimal code based on the load balancing corresponding to the task assignment code, and optimize other task assignment codes based on the current optimal code to obtain a final task assignment code; The task allocation code with the greatest load balancing is the current optimal code. Since the embodiment of the present invention adopts an intelligent coding method to perform the allocation relationship coding of the edge computing server, the allocation relationship coding can be continuously optimized and combined with the current optimal coding to search in the high-dimensional solution space to determine the final task allocation code, which can effectively improve the load uniformity and release the overall performance of the edge computing server cluster.

[0024] S5. Based on the final task allocation code, the target edge computing task is allocated to the edge computing server for processing to achieve load balancing of edge computing.

[0025] After decoding the final task assignment code, the edge computing server corresponding to each target edge computing task can be determined, and then the target edge computing task can be assigned to the edge computing server for processing, thereby effectively improving the load uniformity of the edge computing server and ultimately improving the overall data processing efficiency of the edge computing server.

[0026] In an embodiment of the present invention, for multiple target edge computing tasks, an intelligent coding method is used to encode the allocation relationship of edge computing servers to obtain multiple task allocation codes, including: For multiple target edge computing tasks, a value is randomly generated in the (0, 1] numerical range, and the value corresponding to the target edge computing task is encoded into a vector to determine the task assignment code; the encoding is repeated multiple times to obtain multiple task assignment codes.

[0027] For example, there are five target edge computing tasks. These five target edge computing tasks are arranged in chronological order as A1, A2, A3, A4, and A5. At this time, the order of the five target edge computing tasks is fixed. Then, a numerical value is assigned to each target edge computing task, and the numerical value is encoded as a vector. Assuming that the obtained task assignment is encoded as [0.12, 0.25, 0.65, 0.33, 0.95], then A1, A2, A3, A4, and A5 correspond to 0.12, 0.25, 0.65, 0.33, and 0.95 respectively. The subsequent decoding process can be decoded according to this correspondence.

[0028] In an embodiment of the present invention, after encoding the allocation relationship of the edge computing server using the intelligent encoding method, the method further includes: Evenly divide the (0,1] numerical range into multiple sub-target intervals; the number of sub-target intervals is the same as the number of edge computing servers; Randomly match the sub-target interval for each edge computing server and determine the sub-target interval corresponding to the edge server; wherein the edge computing server and the sub-target interval are in a one-to-one correspondence; Determine the edge computing server corresponding to the sub-target interval where the value corresponding to a target edge computing task in the task allocation code is located as the target edge computing server of the target edge computing task.

[0029] For example, if there are four edge computing servers, four sub-target intervals should be divided, and the sub-target intervals corresponding to the four edge computing servers are (0, 0.25], (0.25, 0.5], (0.5, 0.75], and (0.75, 1]. By converting the allocation process into continuous numerical optimization, the optimization efficiency and effect can be effectively improved.

[0030] In an embodiment of the present invention, for any task allocation code, an artificial intelligence algorithm is used to perform load balancing evaluation to determine the load balancing corresponding to the task allocation code, including: For any task assignment code, determine the computer resource requirements corresponding to the target edge computing task corresponding to the task assignment code and the occupied computer resources corresponding to each edge computing server; Assign the task to the target edge computing task corresponding to the code value, and determine the target edge computing server corresponding to each target edge computing task; Add the computer resources required for the target edge computing task to the occupied computer resources of the corresponding target edge computing server to obtain the predicted occupied computer resources corresponding to each edge computing server; The predicted occupied computer resources of all edge computing servers are used as input to the pre-deployed artificial intelligence algorithm to obtain the load balancing corresponding to the task allocation code; The load balancing is used to characterize the inverse of the variance of computer resource usage, that is, the greater the load balancing of the edge computing server, the more balanced the load.

[0031] Computer resources can be CPU resources, RAM resources and other resources related to data processing performance. Each target edge computing task has corresponding computer resource requirements. By adding the computer resource requirements to the occupied computer resources of the corresponding target edge computing server, the changes in the occupied computer resources of the target edge computing server after the target edge computing task is assigned to the target edge computing server can be simulated, which can be used to evaluate the rationality of the allocation of the target edge computing task.

[0032] Optionally, in order to achieve multi-objective optimization, indicators such as processing time and processing energy consumption can be included in the evaluation scope and integrated into the load balancing as part of the denominator, thereby achieving more comprehensive optimization.

[0033] In an embodiment of the present invention, the following further comprises: For any task assignment code, determine one by one whether each target edge computing task satisfies the constraint conditions after being assigned to the corresponding target edge computing server. If so, continue to judge until each target edge computing task in the task assignment code is judged to be completed. Otherwise, the target edge computing task that does not meet the constraint conditions will be assigned to other edge computing servers until each target edge computing task in the task assignment code is judged to be completed. Among them, the constraint condition is set as: after the target edge computing task is assigned to the edge computing server, the predicted computer resources occupied by the edge computing server do not exceed its resource upper limit.

[0034] In an embodiment of the present invention, allocating target edge computing tasks that do not meet the constraint conditions to other edge computing servers includes: Determine the edge computing server with the lowest computer resource utilization rate, and assign the target edge computing tasks that do not meet the constraints to the edge computing server with the lowest computer resource utilization rate.

[0035] Through the above constraints, it can be effectively guaranteed that the target edge computing task can be executed smoothly after being assigned to the edge computing server.

[0036] In an embodiment of the present invention, the deployment method of the artificial intelligence algorithm includes: The historical occupied computer resources and historical computer resource utilization rates after historical edge computing tasks are assigned to edge computing servers are collected, and the historical occupied computer resources are used as optimization samples. The inverse of the variance of the historical computer resource utilization rate is used as the expected output to deploy the artificial intelligence algorithm.

[0037] For example, a convolutional neural network algorithm or a long short-term memory neural network algorithm can be deployed as an artificial intelligence algorithm to quickly evaluate load uniformity.

[0038] In an embodiment of the present invention, the current optimal code is determined based on the load balancing corresponding to the task allocation code, and other task allocation codes are optimized based on the current optimal code to obtain the final task allocation code, including: According to the load balance corresponding to the task allocation code, the task allocation code with the greatest load balance is determined as the current optimal code; Based on the current optimal code, the particle swarm algorithm is used to optimize other task allocation codes to obtain the final task allocation code.

[0039] Based on the current optimal code, the particle swarm algorithm is used to optimize other task allocation codes to obtain the final task allocation code, including: a1. Based on the current optimal encoding, the task allocation encoding is adaptively optimized to determine the task allocation encoding after adaptive optimization:

[0040]

[0041]

[0042]

[0043]

[0044]

[0045] in, Indicates thet During the optimization process i A task assignment code, Indicates the i The task allocation code after adaptive optimization, i =1,2,…,NP, NP represents the total number of task assignment codes, Indicates the t +1 training session i The update speed of the task allocation code, Indicates the t +1 training session i The update speed of the task allocation code, represents the inertia weight, represents the first learning factor, represents the second learning factor, represents the current optimal encoding, Indicates the i The historical optimal value of the task allocation encoding, Indicates the t +1 first search control factor during training, Indicates the t +1 second search control factor in the training process, mod represents the remainder function, cos represents the cosine function, Indicates the t +1 first chaos parameter during training, Indicates the t The first chaotic parameter in the training process, represents a natural constant, Indicates the t +1 second chaos parameter during training, Indicates the t The second chaotic parameter in the training process, and at the initial moment, the first search control factor, the second search control factor, the first chaotic parameter and the second chaotic parameter are all randomly generated between (0, 1); Through the above updates, the diversity of the algorithm can be effectively improved, helping the algorithm to escape from the local optimum faster, while also being able to learn better information and improve the overall performance of the algorithm.

[0046] a2. Arrange the task allocation codes after adaptive optimization in descending order of load balancing, and perform chain updates based on the arranged task allocation codes. The task allocation codes after chain updates are obtained as follows:

[0047]

[0048] in, Indicates the t During the optimization process j The task assignment code after sorting, Indicates the j The task allocation code after the chain update, j =1,2,…,NP, Represents a random number between (0,1), represents the optimal search control coefficient, represents a random number between (0,1), T represents the maximum number of optimizations, represents pi; The above chain update method can enable all task allocation codes to conduct joint search, and multiple search paths gradually form a network, which is not only conducive to improving the global search capability, but also conducive to improving the ability to search for better solutions.

[0049] a3. Perform a global search on the task allocation code after the chain update, and obtain the task allocation code after the global search as:

[0050] in, Indicates the t During the optimization process n The reverse solution of the task allocation encoding after chain updates, Indicates the n The task assignment encoding after a global search, n =1,2,…,NP, represents the third learning factor, Represents a random number between (0,1), Indicates the task allocation encoding with minimum load balancing; Optionally, a simulated annealing algorithm can be used to control the global search process to ensure the optimization speed of the algorithm. This global search allows the algorithm to escape from the local optimal solution, thereby finding the best task allocation code and improving the balance of edge computing task allocation.

[0051] a4. Determine whether the current number of optimizations has reached the maximum number of optimizations. If so, re-obtain the current optimal code based on the task allocation code after the global search, and use the re-obtained current optimal code as the task allocation code after the global search. Otherwise, return to step a1.

[0052] Although the existing particle swarm algorithm has high optimization accuracy and fast speed, it is easy to fall into local optimality. Therefore, the embodiment of the present invention improves the particle swarm algorithm to enhance the optimization effect, find the best task allocation solution, and improve the overall work efficiency of the edge computing server cluster.

[0053] In an embodiment of the present invention, based on the final task allocation code, the target edge computing task is allocated to the edge computing server for processing, including: Determine the sub-goal interval of the value corresponding to each target edge computing task in the final task allocation code, and obtain the sub-goal interval corresponding to each target edge computing task; Determine the edge computing server corresponding to the sub-target interval corresponding to the target edge computing task as the final edge computing server of the target edge computing task; The target edge computing task is assigned to the corresponding final edge computing server for processing, and each final edge computing server processes the target edge computing task in the chronological order.

[0054] In an embodiment of the present invention, after allocating the target edge computing task to the edge computing server for processing, the method further includes: If an edge computing server is overloaded, the overloaded edge computing server will not participate in the execution of other edge computing tasks in the next task processing cycle.

[0055] Edge computing server overload means that the resource usage of the edge computing server has reached its allowed upper limit. At this time, it is unable to perform more edge computing tasks on its own. In the next task processing cycle, it can not participate in task execution. In the case of excessive access, it can ensure the smooth execution of edge computing tasks and avoid jamming.

[0056] The present invention provides an artificial intelligence-based edge computing load balancing method, which obtains an edge computing task queue and sequentially takes out multiple target edge computing tasks from the edge computing task queue according to a fixed number, then uses an intelligent coding method to encode the allocation relationship of the edge computing server for the multiple target edge computing tasks, obtains multiple task allocation codes, and uses an artificial intelligence algorithm to perform load balancing evaluation to determine the load balancing corresponding to the task allocation code; finally, according to the load balancing corresponding to the task allocation code, the current optimal code is determined, and other task allocation codes are optimized according to the current optimal code to obtain the final task allocation code, and based on the final task allocation code, the target edge computing task is allocated to the edge computing server for processing, thereby realizing edge computing load balancing, effectively avoiding overload of the edge computing server, and improving the overall data processing efficiency.

[0057] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. The present invention is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art not disclosed herein. It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and variations can be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. An edge computing load balancing method based on artificial intelligence, characterized in that: include: Obtain an edge computing task queue, and sequentially take out a plurality of target edge computing tasks from the edge computing task queue according to a fixed number; For multiple target edge computing tasks, an intelligent coding method is used to encode the allocation relationship of edge computing servers and obtain multiple task allocation codes; For any task assignment code, an artificial intelligence algorithm is used to evaluate the load balancing and determine the load balancing corresponding to the task assignment code; Determine the current optimal code based on the load balancing of the task assignment code, and optimize other task assignment codes based on the current optimal code to obtain the final task assignment code; Based on the final task allocation code, the target edge computing task is allocated to the edge computing server for processing to achieve load balancing of edge computing.

2. The edge computing load balancing method based on artificial intelligence according to claim 1 is characterized in that: For multiple target edge computing tasks, an intelligent coding method is used to encode the allocation relationship of edge computing servers and obtain multiple task allocation codes, including: For multiple target edge computing tasks, randomly generate a value in the (0,1] range, encode the value corresponding to the target edge computing task into a vector, and determine the task assignment code; Repeat the encoding multiple times to obtain multiple task assignment codes.

3. The artificial intelligence-based edge computing load balancing method according to claim 2, characterized in that: After using the intelligent coding method to encode the allocation relationship of the edge computing server, it also includes: Evenly divide the (0,1] numerical range into multiple sub-target intervals; the number of sub-target intervals is the same as the number of edge computing servers; Randomly match the sub-target interval for each edge computing server and determine the sub-target interval corresponding to the edge server; wherein the edge computing server and the sub-target interval are in a one-to-one correspondence; Determine the edge computing server corresponding to the sub-target interval where the value corresponding to a target edge computing task in the task allocation code is located as the target edge computing server of the target edge computing task.

4. The artificial intelligence-based edge computing load balancing method according to claim 3, characterized in that: For any task allocation code, an artificial intelligence algorithm is used to evaluate the load balancing performance and determine the load balancing performance corresponding to the task allocation code, including: For any task assignment code, determine the computer resource requirements corresponding to the target edge computing task corresponding to the task assignment code and the occupied computer resources corresponding to each edge computing server; Assign the task to the target edge computing task corresponding to the code value, and determine the target edge computing server corresponding to each target edge computing task; Add the computer resources required for the target edge computing task to the occupied computer resources of the corresponding target edge computing server to obtain the predicted occupied computer resources corresponding to each edge computing server; The predicted occupied computer resources of all edge computing servers are used as input to the pre-deployed artificial intelligence algorithm to obtain the load balancing corresponding to the task allocation code; The load balancing is used to characterize the inverse of the variance of computer resource usage, that is, the greater the load balancing of the edge computing server, the more balanced the load.

5. The artificial intelligence-based edge computing load balancing method according to claim 4, characterized in that: Also includes: For any task assignment code, determine one by one whether each target edge computing task satisfies the constraint conditions after being assigned to the corresponding target edge computing server. If so, continue to judge until each target edge computing task in the task assignment code is judged to be completed. Otherwise, the target edge computing task that does not meet the constraint conditions will be assigned to other edge computing servers until each target edge computing task in the task assignment code is judged to be completed. Among them, the constraint condition is set as: after the target edge computing task is assigned to the edge computing server, the predicted computer resources occupied by the edge computing server do not exceed its resource upper limit.

6. The artificial intelligence-based edge computing load balancing method according to claim 5, characterized in that: Assign target edge computing tasks that do not meet the constraints to other edge computing servers, including: Determine the edge computing server with the lowest computer resource utilization rate, and assign the target edge computing tasks that do not meet the constraints to the edge computing server with the lowest computer resource utilization rate.

7. The artificial intelligence-based edge computing load balancing method according to claim 4, characterized in that: The deployment method of the artificial intelligence algorithm includes: The historical occupied computer resources and historical computer resource utilization rates after historical edge computing tasks are assigned to edge computing servers are collected, and the historical occupied computer resources are used as optimization samples. The inverse of the variance of the historical computer resource utilization rate is used as the expected output to deploy the artificial intelligence algorithm.

8. The edge computing load balancing method based on artificial intelligence according to claim 1 is characterized in that: Determine the current optimal code based on the load balancing corresponding to the task allocation code, and optimize other task allocation codes based on the current optimal code to obtain the final task allocation code, including: According to the load balance corresponding to the task allocation code, the task allocation code with the greatest load balance is determined as the current optimal code; Based on the current optimal code, the particle swarm algorithm is used to optimize other task allocation codes to obtain the final task allocation code.

9. The edge computing load balancing method based on artificial intelligence according to claim 1 is characterized in that: Based on the final task assignment code, the target edge computing task is assigned to the edge computing server for processing, including: Determine the sub-goal interval of the value corresponding to each target edge computing task in the final task allocation code, and obtain the sub-goal interval corresponding to each target edge computing task; Determine the edge computing server corresponding to the sub-target interval corresponding to the target edge computing task as the final edge computing server of the target edge computing task; The target edge computing task is assigned to the corresponding final edge computing server for processing, and each final edge computing server processes the target edge computing task in the chronological order.

10. The edge computing load balancing method based on artificial intelligence according to claim 1, characterized in that: After allocating the target edge computing task to the edge computing server for processing, the method further includes: If an edge computing server is overloaded, the overloaded edge computing server will not participate in the execution of other edge computing tasks in the next task processing cycle.

Images