Power distribution network automatic calculation task scheduling method based on cloud edge collaboration, terminal equipment and storage medium

Abstract

The invention discloses a power distribution network automatic calculation task scheduling method based on cloud edge collaboration, terminal equipment and a storage medium, and belongs to the technical field of power distribution networks. The method comprises the following steps: calculating a scheduling decision index value of a task according to a calculation complexity index, a real-time requirement index, a data processing amount index and a data security requirement index of the task; and performing distribution value cloud processing or local processing on the task according to the scheduling decision index value, so that by implementing the method, the scheduling between cloud computing resources and edge computing can be realized, the real-time performance of the system is improved, the cloud and edge computing resource distribution is dynamically adjusted, the computing efficiency is improved, and the computing cost is reduced. In addition, excessive configuration and waste of resources are avoided, and the technical problems that the real-time requirement is difficult to guarantee and computing power resource configuration is unbalanced in the prior art can be solved.

Description

Technical Field

[0001] The present invention relates to the field of distribution network technology, and in particular to a distribution network automated computing task scheduling method based on cloud-edge collaboration, a terminal device, and a storage medium. Background Art

[0002] As core infrastructure for ensuring the safe and economical operation of power networks, the performance of distribution network automation systems directly impacts the stability and reliability of power supply. Traditional distribution network automation technologies rely on cloud computing centers to centrally handle all tasks, which makes it difficult to guarantee real-time performance and leads to an imbalance in computing resource allocation. For example, when a distribution network failure occurs, centralized scheduling can lead to delayed response due to network latency, potentially expanding the scope of the outage. Therefore, to address the issue of relying solely on cloud computing centers to handle all tasks, edge computing can be introduced to assist in task processing. Therefore, a task scheduling strategy based on cloud-edge collaboration is currently needed to enable task scheduling between cloud computing and edge computing resources. Summary of the Invention

[0003] The present invention provides a distribution network automation computing task scheduling method, terminal device and storage medium based on cloud-edge collaboration, which can solve the problem that the existing technology relies on cloud computing centers to centrally process all tasks, resulting in difficulty in ensuring real-time requirements and imbalance in computing resource allocation.

[0004] An embodiment of the present invention provides a method for scheduling automated computing tasks in a distribution network based on cloud-edge collaboration, including:

[0005] Obtain a task set and a task feature vector of each task in the task set; wherein the task feature vector includes: a computational complexity index, a real-time requirement index, a data processing volume index, and a data security requirement index; the computational complexity index is used to measure the amount of computing resources required for the task, and the more computing resources the task requires, the higher the computational complexity; the real-time requirement index is used to measure the urgency of the task completion deadline, and the tighter the task completion deadline, the higher the real-time requirement index; the data processing volume index is used to measure the amount of data that the task needs to process, and the larger the amount of data that the task needs to process, the higher the data processing volume index; the data security requirement index is used to measure the degree of data security requirement of the task, and the higher the data security requirement of the task, the higher the data security requirement index;

[0006] For each task, the scheduling decision index value of the task is calculated based on the task feature vector of the task and the calculation formula of the scheduling decision index; wherein the scheduling decision index is directly proportional to the computational complexity index, the data processing volume index, and the data security requirement index, and is inversely proportional to the real-time requirement index;

[0007] Determine whether the scheduling decision indicator value is greater than a preset scheduling decision indicator threshold; if so, mark the task as a task to be uploaded;

[0008] Upload all tasks marked as needing to be uploaded to the cloud for collaborative processing by the cloud, and process the remaining tasks locally.

[0009] Furthermore, the calculation formula of the scheduling decision index is:

[0010]

[0011] Where g(f(t)) represents the scheduling decision index value; t represents the task; f(t) represents the task feature vector of task t; C(t) represents the computational complexity index of task t; R(t) represents the real-time requirement index of task t; D(t) represents the data processing volume index of task t; S(t) represents the data security requirement index of task t; w1 is the weight coefficient of the computational complexity index; w2 is the weight coefficient of the real-time requirement index; w3 is the weight coefficient of the data processing volume index; and w4 is the weight coefficient of the data security requirement index.

[0012] Furthermore, the distribution network automation computing task scheduling method based on cloud-edge collaboration also includes:

[0013] Obtaining a node characteristic vector for each node in the distribution network; wherein the nodes include: cloud nodes and edge nodes; the node characteristic vector includes: a node comprehensive performance index and a node load index; wherein the node comprehensive performance index is used to measure the quality of the node comprehensive performance; the better the node comprehensive performance, the higher the node comprehensive performance index; the node load index is used to measure the level of the node load; the higher the node load, the higher the node load index;

[0014] For each node, the task assignment probability value of the node is calculated based on the node feature vector of the node using the task assignment probability calculation formula; wherein the task assignment probability value is used to measure the probability of the node being assigned a task; the task assignment probability value is directly proportional to the node's comprehensive performance index and inversely proportional to the node's load index;

[0015] For each task, if the task does not need to be executed across nodes, the task will be assigned to the node with the highest task assignment probability value; if the task needs to be executed across nodes, the task will be split into several subtasks; according to the number of subtasks, a corresponding number of nodes are selected in descending order of task assignment probability values as target nodes; and the tasks are evenly distributed to the target nodes.

[0016] Furthermore, the calculation formula for the task assignment probability is:

[0017]

[0018] Where i represents a node; P i represents the task assignment probability value of node i; W i represents the comprehensive performance index of node i; t i represents the node load index of node i; n represents the total number of nodes; λ1 is the first smoothing factor, which is used to adjust the influence of the node load index on the task allocation probability.

[0019] Furthermore, the distribution network automation computing task scheduling method based on cloud-edge collaboration also includes:

[0020] Obtain node performance indicators for each node in the distribution network and link performance indicators for each communication link in the distribution network; wherein the nodes include cloud nodes and edge nodes; the node performance indicators include: CPU usage, memory usage, network traffic, and response time; the link performance indicators include: latency, packet loss rate, bit error rate, delay, and link utilization;

[0021] For each node, a node fault identification index value of the node is calculated based on the node performance index of the node using a calculation formula for the fault identification index; wherein the node fault identification index is used to measure the probability of a node fault; the higher the probability of a node fault, the higher the node fault identification index;

[0022] Determining whether the node fault identification index value exceeds a preset node fault identification index threshold, and if so, performing fault detection on the node;

[0023] For each communication link, a link fault identification index value of the link is calculated based on the link performance indicator of the communication link using a calculation formula for the fault identification index; wherein the link fault identification index is used to measure the probability of a communication link failure; the higher the probability of a communication link failure, the higher the link fault identification index;

[0024] It is determined whether the link fault identification index value exceeds a preset link fault identification index threshold value; if so, a fault detection is performed on the communication link.

[0025] Furthermore, the calculation formula of the fault identification index is:

[0026]

[0027] Where FI represents the fault identification index value; x j represents the j-th node performance index of a node or the j-th link performance index of a communication link; μ jrepresents the normal baseline value of the j-th node performance indicator or the normal baseline value of the j-th link performance indicator; σ j represents the standard deviation of the j-th node performance indicator or the j-th link performance indicator; m represents the total number of node performance indicators or the total number of link performance indicators; λ2 is the second smoothing factor, which is used to adjust the impact of abnormal node performance indicators on the node fault identification index or the impact of abnormal link performance indicators on the link fault identification index.

[0028] Furthermore, the distribution network automation computing task scheduling method based on cloud-edge collaboration also includes:

[0029] Obtaining a fault node performance indicator for each fault node in the distribution network and a fault link performance indicator for each fault communication link in the distribution network;

[0030] For each faulty node, a node fault impact assessment coefficient value of the faulty node is calculated based on the faulty node performance indicator of the faulty node and using a calculation formula for the fault impact assessment coefficient; wherein the node fault impact assessment coefficient value is used to measure the fault severity of the faulty node; the higher the fault severity of the faulty node, the higher the node fault impact assessment coefficient;

[0031] Sort all faulty nodes in descending order according to the node fault impact assessment coefficient values, use the sorting results as the faulty node repair order, and repair the faulty nodes according to the faulty node repair order;

[0032] For each faulty communication link, a link fault impact assessment coefficient value of the faulty communication link is calculated according to the faulty link performance indicator of the faulty communication link and using a calculation formula for the fault impact assessment coefficient; wherein the link fault impact assessment coefficient value is used to measure the fault severity of the faulty communication link, and the higher the fault severity of the faulty communication link, the higher the link fault impact assessment coefficient;

[0033] All faulty communication links are sorted in descending order according to the link fault impact assessment coefficient values, the sorting result is used as the faulty communication link repair sequence, and the faulty communication links are repaired according to the faulty communication link repair sequence.

[0034] Furthermore, the calculation formula of the fault impact assessment coefficient is:

[0035]

[0036] Where E represents the fault impact assessment coefficient value.

[0037] Another embodiment of the present invention also provides a terminal device, including: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the steps of the distribution network automation computing task scheduling method based on cloud-edge collaboration as described in the present invention.

[0038] Another embodiment of the present invention also provides a computer-readable storage medium item, including: a stored computer program, which controls the device where the computer-readable storage medium is located to execute the steps of the distribution network automation computing task scheduling method based on cloud-edge collaboration according to the present invention when the computer program is running.

[0039] The following beneficial effects are achieved by implementing the present invention:

[0040] The present invention provides a method for scheduling automated computing tasks in a distribution network based on cloud-edge collaboration, which can realize scheduling between cloud computing resources and edge computing, calculate the scheduling decision index value of the task according to the task feature vector of the task, and assign the task to cloud processing or local processing according to the scheduling decision index value. Part of the computing tasks are sunk to the edge nodes, which significantly reduces the delay of data transmission and improves the real-time performance of the system. In addition, the elasticity and scalability of cloud computing and the localized processing capabilities of edge computing are utilized to dynamically adjust the computing resource allocation of the cloud and edge according to the actual load conditions in the distribution network, which not only improves the computing efficiency, but also avoids the over-allocation and waste of resources, thereby achieving effective cost control and solving the technical problems that the existing technology is difficult to guarantee the real-time requirements and the computing power resource allocation is unbalanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for use in the implementation. Obviously, the drawings described below are only some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0042] Figure 1 This is a flow chart of a method for scheduling automated computing tasks in a distribution network based on cloud-edge collaboration provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0043] To make the objectives, technical solutions, and advantages of this application more clear, the technical solutions in this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0044] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0045] In the description of the embodiments of this application, the technical terms "first" and "second" are used only to distinguish different objects and should not be understood to indicate or imply relative importance or implicitly specify the quantity, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "plurality" is more than two, unless otherwise clearly and specifically defined.

[0046] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0047] In the description of the embodiments of the present application, unless otherwise expressly specified or limited, technical terms such as "installed," "connected," "connected," and "fixed" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integration; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; internal connections between two components or interactions between two components. Those skilled in the art can understand the specific meanings of the above terms in the embodiments of the present application based on specific circumstances.

[0048] See also Figure 1 To address the existing issues of relying on cloud computing centers to centrally process all tasks, resulting in difficulty in ensuring real-time performance and imbalanced computing resource allocation, an embodiment of the present invention provides a method for scheduling automated computing tasks in a distribution network based on cloud-edge collaboration, including:

[0049] S1. Obtain a task set and a task feature vector of each task in the task set; wherein, the task feature vector includes: a computational complexity index, a real-time requirement index, a data processing volume index and a data security requirement index; the computational complexity index is used to measure the amount of computing resources required for the task, and the more computing resources the task requires, the higher the computational complexity; the real-time requirement index is used to measure the urgency of the task completion deadline, and the tighter the task completion deadline, the higher the real-time requirement index; the data processing volume index is used to measure the amount of data that the task needs to process, and the larger the amount of data that the task needs to process, the higher the data processing volume index; the data security requirement index is used to measure the degree of data security requirements of the task, and the higher the data security requirements of the task, the higher the data security requirement index.

[0050] In one embodiment, obtaining a task set and a task feature vector of each task in the task set includes:

[0051] Obtaining distribution network operation data, equipment status information, and optimization targets; wherein the operation data includes voltage, current, power, and energy consumption;

[0052] Preprocessing the operating data; wherein the preprocessing includes filtering, calibration, outlier detection and data formatting;

[0053] Calculate the accuracy score, real-time score and reliability score based on the pre-processed operation data;

[0054] Determine the task type according to the accuracy score, the real-time score, and the reliability score, and form a task set;

[0055] Determine business needs based on pre-processed operating data, equipment status information, and optimization goals;

[0056] According to the business requirements, a task feature vector of each task is determined.

[0057] It's important to note that operational data forms the basis for subsequent analysis and monitoring of the distribution network's status. This data is collected in real time through sensors and measuring devices. These sensors and measuring devices include electricity meters, voltage sensors, current sensors, and power quality analyzers. Smart meters measure energy consumption and provide information on voltage, current, and power. Voltage sensors measure voltage levels in the distribution network. Current sensors measure current flowing through conductors. Power quality analyzers analyze power quality, including power factor and harmonic content.

[0058] Optimization objectives refer to a series of goals set in the distribution network automation computing task scheduling system to improve system performance, resource utilization efficiency, and operational stability. The optimization objectives are designed to meet the requirements of high efficiency, economy, reliability, and security in system operation. The optimization objectives are set based on the following aspects: task execution efficiency, resource utilization efficiency, system stability, cost control, and data security. Task execution efficiency ensures that tasks can be completed quickly and efficiently. Resource utilization efficiency rationally allocates computing, storage, and network resources to avoid resource waste. System stability reduces the occurrence of failures and ensures long-term stable operation of the system. Cost control reduces system operation and maintenance costs while meeting performance requirements. Data security ensures the security and integrity of data during transmission and processing.

[0059] There are two ways to obtain the optimization target:

[0060] (1) Directly obtain the optimization goals set by humans based on the demand analysis during the system design phase, such as: task execution time requires that all tasks be completed within the specified time, resource utilization requires that the utilization of resources such as CPU and memory does not exceed a certain threshold, and system availability requires that the system failure time does not exceed a certain proportion;

[0061] (2) Dynamically adjust the optimization target based on the system operation data: The data sources are operation data, equipment status information, performance indicators and task characteristics; by analyzing historical data, the normal operation status and performance bottlenecks of the system are determined, and by real-time monitoring of the system operation status, the optimization target is dynamically adjusted. The optimization target is adjusted according to changes in business needs, and the optimization target can also be adjusted according to fault detection and analysis results.

[0062] Preprocessing operations remove noise and interference in the data through filtering to improve the accuracy and reliability of the data. The collected data is corrected according to the calibration parameters of the equipment to ensure the accuracy of the data. Outliers in the data are identified and processed to avoid their adverse effects on subsequent analysis. The data is converted into a unified format for subsequent processing and analysis.

[0063] Business requirements are determined based on the operating conditions and optimization goals of the distribution network. It can monitor current to detect overload or short-circuit anomalies, monitor voltage to ensure it remains within a safe and prescribed range, and analyze power quality to assess the efficiency and stability of the power system.

[0064] The process of generating business requirements is as follows: first, real-time monitoring and analysis of operating status data is carried out to identify abnormal conditions in the data (voltage fluctuations, current overloads, power anomalies, etc.), and the health status and operating efficiency of the equipment are evaluated in combination with equipment status information and performance indicators. Based on these analysis results and combined with optimization goals (improving task execution efficiency, optimizing resource utilization, and ensuring system stability), the type of task to be performed is determined (real-time data monitoring, data quality analysis, fault warning, data analysis, predictive maintenance, fault diagnosis, etc.); then, a task feature vector is defined for each task, including key information such as the computational complexity of the task, real-time requirements, the amount of data to be processed, and the level of data security requirements. The feature vector reflects the resource requirements and priority of the task, which is used for subsequent task scheduling and resource allocation decisions. In this way, the system dynamically generates business requirements based on the current operating status and optimization goals.

[0065] A task set is a series of tasks that need to be processed based on operational data and business needs. Task types include: real-time data monitoring, data quality analysis, fault warning, predictive maintenance, and fault diagnosis.

[0066] The accuracy score is calculated based on the deviation between the calibrated data and the standard value, indicating the relative deviation between the calibrated data and the standard value. The smaller the deviation, the higher the accuracy score.

[0067]

[0068] The real-time score is calculated based on the difference between the data collection time and the current time. The smaller the difference, the higher the real-time score.

[0069]

[0070] The reliability score is calculated based on the results of outlier detection and the continuity of the data. It represents the proportion of outliers in the data. The smaller the proportion, the higher the reliability score.

[0071]

[0072] The process of determining the type of task is as follows:

[0073] When the real-time score is high, the real-time data monitoring task is triggered. If the real-time score is greater than the preset threshold (0.8), it is considered that the real-time requirement of the data is high and real-time data monitoring is required.

[0074] When the accuracy score or reliability score is low, the data quality analysis task is triggered. If the accuracy score or reliability score is less than the preset threshold (0.7), it is considered that there may be problems with the data quality and data quality analysis is required.

[0075] When any one or more of the accuracy score, real-time score, and reliability score is lower than the preset threshold, the fault warning task is triggered in combination with specific data features (such as voltage anomaly, current overload).

[0076] Examples of task types and their triggering conditions are as follows:

[0077] (1) Real-time data monitoring

[0078] Trigger condition: Triggered when the real-time score (value B) is high;

[0079] Application scenarios: Suitable for scenarios with high requirements for data real-time performance, such as real-time display and dynamic monitoring of power grid operation status;

[0080] (2) Data quality analysis

[0081] Trigger condition: Triggered when the accuracy score (value A) or reliability score (value C) is low;

[0082] Application scenarios: Suitable for scenarios where data quality may be problematic, such as when there are deviations or a large number of outliers during data collection;

[0083] (3) Fault warning

[0084] Trigger condition: When any one or more of the accuracy score (value A), real-time score (value B), or reliability score (value C) falls below the preset threshold, and is combined with specific data characteristics (voltage anomaly, current overload, etc.);

[0085] Application scenarios: Applicable to scenarios where data anomalies may cause system failures, such as voltage fluctuations and current overloads;

[0086] (4) Predictive maintenance

[0087] Trigger condition: triggered when the system needs to make equipment maintenance plans based on data analysis results;

[0088] Application scenarios: Applicable to equipment maintenance management scenarios, such as predicting maintenance cycles based on equipment operation data;

[0089] (5) Fault diagnosis

[0090] Trigger condition: Triggered when the system detects a fault;

[0091] Application scenarios: Applicable to fault handling scenarios, such as when the system detects voltage anomalies or equipment failures.

[0092] S2. For each task, based on the task feature vector of the task, the scheduling decision index value of the task is calculated using the calculation formula of the scheduling decision index; wherein the scheduling decision index is directly proportional to the computational complexity index, the data processing volume index and the data security requirement index, and is inversely proportional to the real-time requirement index.

[0093] In a preferred embodiment, the calculation formula of the scheduling decision index is:

[0094]

[0095] Where g(f(t)) represents the scheduling decision index value; t represents the task; f(t) represents the task feature vector of task t; C(t) represents the computational complexity index of task t; R(t) represents the real-time requirement index of task t; D(t) represents the data processing volume index of task t; S(t) represents the data security requirement index of task t; w1 is the weight coefficient of the computational complexity index; w2 is the weight coefficient of the real-time requirement index; w3 is the weight coefficient of the data processing volume index; and w4 is the weight coefficient of the data security requirement index.

[0096] It should be noted that w1+w2+w3+w4=1.

[0097] S3. Determine whether the scheduling decision indicator value is greater than a preset scheduling decision indicator threshold. If so, mark the task as a task to be uploaded.

[0098] S4. Upload all tasks marked as needing to be uploaded to the cloud for collaborative processing by the cloud, and the remaining tasks are processed locally.

[0099] In a preferred embodiment, the distribution network automation computing task scheduling method based on cloud-edge collaboration further includes:

[0100] Obtaining a node characteristic vector for each node in the distribution network; wherein the nodes include: cloud nodes and edge nodes; the node characteristic vector includes: a node comprehensive performance index and a node load index; wherein the node comprehensive performance index is used to measure the quality of the node comprehensive performance; the better the node comprehensive performance, the higher the node comprehensive performance index; the node load index is used to measure the level of the node load; the higher the node load, the higher the node load index;

[0101] For each node, the task assignment probability value of the node is calculated based on the node feature vector of the node using the task assignment probability calculation formula; wherein the task assignment probability value is used to measure the probability of the node being assigned a task; the task assignment probability value is directly proportional to the node's comprehensive performance index and inversely proportional to the node's load index;

[0102] For each task, if the task does not need to be executed across nodes, the task will be assigned to the node with the highest task assignment probability value; if the task needs to be executed across nodes, the task will be split into several subtasks; according to the number of subtasks, a corresponding number of nodes are selected in descending order of task assignment probability values as target nodes; and the tasks are evenly distributed to the target nodes.

[0103] In this embodiment, computing resources are dynamically allocated according to the task characteristics and the current load situation. The specific process is: analyzing the computational complexity, data volume and real-time requirements of the task to determine the degree of demand for computing resources of the task, monitoring the CPU, memory and storage resource usage of the edge computing device in real time, evaluating the current load situation, and dynamically allocating computing resources to each task according to the task characteristics and the current load situation. For tasks with high computational complexity and high real-time requirements, more CPU and memory resources are allocated, and for tasks with large data volumes, more storage resources are allocated. At the same time, according to the overall load situation of the system, the resource allocation strategy is dynamically adjusted to ensure balanced resource utilization and efficient execution of tasks.

[0104] According to the real-time requirements of the task and the resource evaluation results, load balancing technology is used to calculate the probability of the task being assigned to each node, and then the task request is dispersed to multiple nodes. If the task needs to be executed across nodes, the task is split and allocated. Using load balancing technology, the probability of a task being assigned to each node is calculated based on the node's performance, current load, and smoothing factor. The stronger the node's performance and the lower the load, the higher the probability of being assigned a task. The smoothing factor affects the degree to which load affects the assignment probability. If a task needs to be executed across nodes, it is split and assigned to ensure efficient task completion. The real-time requirements of the task are analyzed, and the task is preferentially assigned to the node that can meet its time requirements. The task operations such as computation and data processing are performed on the assigned nodes. The task execution status and node resource usage are monitored in real time to ensure that the task is completed according to the scheduled time and quality requirements. If insufficient resources or execution timeouts are found during task execution, task allocation and resource usage are adjusted. Based on task execution and resource usage, the collaborative scheduling module is optimized and improved to improve the flexibility and efficiency of task scheduling. After task execution is completed, the execution results of each node are summarized, quality checked and verified, and the execution results are fed back to the task requester. This enables collaborative analysis of cloud and edge resources, jointly processing batch tasks in the system, optimizing resource allocation, improving the flexibility and efficiency of task scheduling, and ensuring stable system operation.

[0105] The process of selecting the nodes to be assigned is as follows: evaluate the computing power, memory, storage and network resources of each node, analyze the current load of the node to avoid overload, and analyze the computational complexity, data dependency, and real-time requirements of the task request to determine the degree of resource demand of the task. Use the load balancing algorithm to select the node according to the node's resources and the task's requirements. If the task is large or needs to be executed across nodes, split it into smaller subtasks and assign them to different nodes according to the requirements of the subtasks and the node's resource situation. During the task execution process, monitor the node's resources and task progress, and adjust the task allocation according to the actual situation to ensure efficient resource utilization and timely completion of the task. When the task is executed across nodes, split the task so that the split tasks are completed on time. One-to-one correspondence with nodes, the specific process is: according to the logical relationship and dependency relationship of the task, split it into several subtasks, each subtask has a clear goal, input and output, so as to facilitate independent execution and result summary, analyze the demand of each subtask for computing resources, storage resources and network resources, determine the execution priority and real-time requirements of each subtask according to the demand, select the appropriate node for each subtask according to the demand of the subtask and the resource situation of the node, ensure the load balance of subtasks on each node, record the correspondence between each subtask and the selected node through the database, execute the subtask on the selected node, monitor the execution progress and resource usage of the subtask, and summarize the results to the main node or the designated location after all subtasks are completed. The system performs quality inspection and verification on the execution results, and then feeds back the execution results to the task requester or the next processing flow; performs calculation and data processing task operations on the assigned nodes, and monitors the execution status of the task and the resource usage of the node in real time during the task execution. If it is found that there are insufficient resources or execution timeouts during the task execution, the task allocation and resource usage will be adjusted to ensure that the task can be completed according to the scheduled time requirements and quality requirements. According to the current resource situation and timeout anomalies, the task allocation and resource usage will be adjusted. The specific process is: real-time monitoring of the usage of CPU, memory, storage, network and other resources in the system, identifying resource bottlenecks (the resource usage of a node or device is too high), and adjusting the resource usage according to the resource bottlenecks. Based on the current resource situation, dynamically adjust task allocation and assign resource-intensive tasks to nodes with sufficient resources. If resources are insufficient, increase the number of servers, expand storage capacity, or increase network bandwidth. If resources are in excess, reduce resource usage. Monitor task execution progress, identify tasks that have timed out and are not completed, analyze the reasons for timeouts (insufficient resources, high task complexity), adjust the priority of timed-out tasks based on their urgency and importance, give priority to important and urgent timed-out tasks, split timed-out tasks into smaller subtasks, and reallocate them to different nodes based on the needs and resource conditions of the subtasks. For timed-out tasks due to insufficient resources, increase the corresponding resource input, such as increasing CPU, memory, or network bandwidth resources, to improve task execution speed.Use load balancing technology to evenly distribute tasks across nodes, preventing overloading of one node while leaving others idle. Dynamically adjust resource usage based on system load, increasing resources during peak loads and reducing them during low loads. Task allocation is determined based on analysis of task priorities, resource requirements, and node capabilities. A comprehensive monitoring system is established to monitor system status and task execution in real time. Task allocation and resource usage are adjusted based on the results. After task execution is complete, the execution results of each node are summarized and quality-checked and verified to ensure accuracy and reliability.

[0106] In a preferred embodiment, the calculation formula for the task assignment probability is:

[0107]

[0108] Where i represents a node; P i represents the task assignment probability value of node i; W i represents the comprehensive performance index of node i; t i represents the node load index of node i; n represents the total number of nodes; λ1 is the first smoothing factor, which is used to adjust the influence of the node load index on the task allocation probability.

[0109] It should be noted that the value of λ1 is between 0.1 and 0.3. i When the load on the node increases, will decrease, thereby reducing The value of P i Decrease means that the probability of nodes with higher load being assigned tasks will decrease. i When (the weight of the node) increases, P i It will increase, making nodes with stronger performance more likely to be assigned tasks. The smoothing factor λ affects the degree of influence of load on the allocation probability. The larger the λ value, the more significant the impact of load on the allocation probability.

[0110] In a preferred embodiment, the distribution network automation computing task scheduling method based on cloud-edge collaboration further includes:

[0111] Obtain node performance indicators for each node in the distribution network and link performance indicators for each communication link in the distribution network; wherein the nodes include cloud nodes and edge nodes; the node performance indicators include: CPU usage, memory usage, network traffic, and response time; the link performance indicators include: latency, packet loss rate, bit error rate, delay, and link utilization;

[0112] For each node, a node fault identification index value of the node is calculated based on the node performance index of the node using a calculation formula for the fault identification index; wherein the node fault identification index is used to measure the probability of a node fault; the higher the probability of a node fault, the higher the node fault identification index;

[0113] Determining whether the node fault identification index value exceeds a preset node fault identification index threshold, and if so, performing fault detection on the node;

[0114] For each communication link, a link fault identification index value of the link is calculated based on the link performance indicator of the communication link using a calculation formula for the fault identification index; wherein the link fault identification index is used to measure the probability of a communication link failure; the higher the probability of a communication link failure, the higher the link fault identification index;

[0115] It is determined whether the link fault identification index value exceeds a preset link fault identification index threshold value; if so, a fault detection is performed on the communication link.

[0116] In this embodiment, the operating status of each link in the system is monitored in real time, including the cloud, edge and communication links, the fault link is detected and located, the fault situation is output, the troubleshooting time is reduced, and the reliability and security of the system are improved.

[0117] It should be noted that performance indicators are mainly used to evaluate the operating efficiency and resource usage of various components in the system (cloud, edge computing nodes, communication links, etc.), including but not limited to the following data items:

[0118] (1) Computing resource performance indicators: divided into CPU usage, memory usage, GPU usage, and CPU / GPU temperature. CPU usage is the current CPU usage, reflecting the system's computing load; memory usage is the current memory usage, reflecting the system's memory load; GPU usage is the GPU usage, reflecting the graphics computing load; CPU / GPU temperature reflects the device's heat dissipation and operating stability;

[0119] (2) Storage resource performance indicators: divided into storage space utilization rate, storage read and write speed, and storage I / O latency. Storage space utilization rate is the utilization rate of the current storage device, storage read and write speed is the read and write speed of the storage device, reflecting storage performance, and storage I / O latency is the input / output latency of the storage device;

[0120] (3) Network resource performance indicators: divided into network bandwidth utilization, network traffic, network delay, and network packet loss rate. Network bandwidth utilization refers to the current network bandwidth usage rate, network traffic refers to the amount of data passing through the network per unit time, network delay refers to the delay time of data transmission in the network, and network packet loss rate refers to the proportion of packet loss in network data transmission;

[0121] (4) Task processing performance indicators: divided into task response time, task processing time, task success rate, and task queue length. Task response time is the time from task submission to task processing start, task processing time is the time required for task completion, task success rate is the ratio of tasks successfully completed, and task queue length is the number of tasks waiting to be processed;

[0122] (5) Overall system performance indicators: divided into system throughput, system resource utilization, and system failure rate. System throughput refers to the number of tasks processed by the system per unit time, system resource utilization refers to the utilization rate of the system's overall resources, and system failure rate refers to the frequency of system failures.

[0123] In a preferred embodiment, the calculation formula of the fault identification index is:

[0124]

[0125] Where FI represents the fault identification index value; x j represents the j-th node performance index of a node or the j-th link performance index of a communication link; μ j represents the normal baseline value of the j-th node performance indicator or the normal baseline value of the j-th link performance indicator; σ j represents the standard deviation of the j-th node performance indicator or the j-th link performance indicator; m represents the total number of node performance indicators or the total number of link performance indicators; λ2 is the second smoothing factor, which is used to adjust the impact of abnormal node performance indicators on the node fault identification index or the impact of abnormal link performance indicators on the link fault identification index.

[0126] It should be noted that μ j is the average value calculated based on historical data; σ j is the standard deviation of the jth performance indicator, indicating the normal fluctuation range, and is a positive number; the value of λ2 is between 0.1 and 0.3. The value range of FI starts from 0. The larger the value, the greater the deviation from the normal baseline, and the higher the possibility of failure. j (real-time performance index value) and μ j (normal baseline value) is close to, (x j -μ j ) 2 Smaller, the fault identification index FI is smaller, when x j With μj When the gap is large, (x j -μ j ) 2 Larger, and As λ increases, the fault identification index FI increases. The smoothing factor λ affects the degree of influence of outliers on the fault identification index. The larger the λ value, the more significant the influence of outliers on FI.

[0127] In a preferred embodiment, the distribution network automation computing task scheduling method based on cloud-edge collaboration further includes:

[0128] Obtaining a fault node performance indicator for each fault node in the distribution network and a fault link performance indicator for each fault communication link in the distribution network;

[0129] For each faulty node, a node fault impact assessment coefficient value of the faulty node is calculated based on the faulty node performance indicator of the faulty node and using a calculation formula for the fault impact assessment coefficient; wherein the node fault impact assessment coefficient value is used to measure the fault severity of the faulty node; the higher the fault severity of the faulty node, the higher the node fault impact assessment coefficient;

[0130] Sort all faulty nodes in descending order according to the node fault impact assessment coefficient values, use the sorting results as the faulty node repair order, and repair the faulty nodes according to the faulty node repair order;

[0131] For each faulty communication link, a link fault impact assessment coefficient value of the faulty communication link is calculated according to the faulty link performance indicator of the faulty communication link and using a calculation formula for the fault impact assessment coefficient; wherein the link fault impact assessment coefficient value is used to measure the fault severity of the faulty communication link, and the higher the fault severity of the faulty communication link, the higher the link fault impact assessment coefficient;

[0132] All faulty communication links are sorted in descending order according to the link fault impact assessment coefficient values, the sorting result is used as the faulty communication link repair sequence, and the faulty communication links are repaired according to the faulty communication link repair sequence.

[0133] In a preferred embodiment, the calculation formula of the fault impact assessment coefficient is:

[0134]

[0135] Where E represents the fault impact assessment coefficient value.

[0136] It should be noted that the value range of E is between 0 and 1. The larger the value, the more serious the fault impact. j -μ j) 2 When the actual observed value is close to the reference value, the radical part is small and the E value is small, indicating that the fault impact is small. When m increases, that is, the total number of performance indicators increases, the logarithmic function log(m) increases, the radical part decreases, and the E value decreases, indicating that as the monitoring range expands, the impact of a single fault is relatively reduced. When FI increases, that is, the fault severity increases, the exponential function Increases, the E value increases, indicating that serious faults have a greater impact on the system.

[0137] In one embodiment, the faults are further divided into fault levels of different severity, namely severe fault level, moderate fault level and minor fault level. Based on the nature of the fault, the degree of impact and the affected users, a fault impact assessment coefficient is obtained in combination with a fault identification index to evaluate the severity of the fault. In combination with the value of the fault impact assessment coefficient, a corresponding impact threshold is matched for each fault level. The faults are prioritized according to their severity, and the faults with the greatest impact on the system are handled first.

[0138] Multiple fault levels correspond to multiple impact thresholds, where the impact thresholds include an upper threshold and a lower threshold. The multiple fault levels and the multiple impact thresholds satisfy the following relationship:

[0139] Severe fault level E H ≤E<1; The system operation is severely impacted, which may cause system shutdown or data loss. This has the greatest impact on the system and requires immediate attention.

[0140] Medium fault level E M ≤E <E H ; It has a certain impact on the system operation, but it will not cause the system to stop operating immediately. It has a certain impact on the system and should be handled as soon as possible after the serious fault is handled;

[0141] Minor fault level 0 <E<E M The impact on system operation is relatively small and can be temporarily monitored without immediate processing. It can be temporarily monitored and processed according to system resources and time arrangements.

[0142] Among them, E H is the lower threshold corresponding to the severe fault level and the upper threshold corresponding to the moderate fault level, E M is the lower threshold corresponding to the medium fault level and the upper threshold corresponding to the minor fault level, E H =0.75, E M =0.5.

[0143] The process of developing and implementing a troubleshooting plan includes:

[0144] Receive fault information from the fault detection module and the fault evaluation and analysis module, confirm the nature, cause, scope of impact and consequences of the fault, as well as the fault level. Determine the priority and strategy for fault handling based on the fault level and impact assessment results, give priority to handling serious faults, ensure that the main functions of the system are restored as soon as possible, and formulate corresponding handling plans for different types of faults. The handling plan includes troubleshooting steps, repair measures, required resources and time estimates. According to the handling plan, prepare the required tools, equipment, and personnel resources to ensure that all resources are in place and in an available state, and follow the steps in the handling plan to gradually troubleshoot and repair the fault. During the troubleshooting process, record relevant information for subsequent analysis and summary. During the handling process, continuously monitor the system status to ensure that the handling measures are effective. If the handling effect is not good or the problem worsens, the handling plan needs to be adjusted. After the handling is completed, conduct a comprehensive inspection of the system to verify whether the fault has been completely resolved, and then feedback the handling results to the relevant management personnel.

[0145] The present invention provides an automatic computing task scheduling system for distribution network based on cloud-edge collaboration. By sinking some computing tasks to edge nodes, the delay of data transmission is significantly reduced, the real-time performance of the system is improved, and the elasticity and scalability of cloud computing and the localized processing capabilities of edge computing are utilized to dynamically adjust the computing resource allocation of the cloud and edge in the distribution network according to the actual load conditions. This not only improves computing efficiency, but also avoids over-configuration and waste of resources, thereby achieving effective cost control.

[0146] The present invention provides a distribution network automation computing task scheduling system based on cloud-edge collaboration. By monitoring and analyzing the operating data of the distribution network, potential problems are discovered and solved, ensuring the long-term stable operation of the distribution network. According to actual needs, resource allocation is intelligently adjusted to achieve rational use of energy. This not only improves the economy and environmental protection of the distribution network, but also significantly improves operation and maintenance efficiency and reduces operation and maintenance costs.

[0147] Based on the above-mentioned method embodiment, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the distribution network automation computing task scheduling method based on cloud-edge collaboration of any embodiment of the present invention.

[0148] For example, in this embodiment, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to implement the present invention. The one or more module elements may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device.

[0149] The terminal device may be a computing device such as a desktop computer, a notebook computer, a PDA, a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0150] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, connecting various parts of the entire terminal device using various interfaces and lines.

[0151] Based on the above-mentioned method embodiments, another embodiment of the present invention provides a computer-readable storage medium, including a stored computer program, wherein, when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the distribution network automation computing task scheduling method based on cloud-edge collaboration as described in any one of the above-mentioned method embodiments of the present invention.

[0152] Wherein, the module / unit integrated in the device / terminal equipment, if implemented in the form of a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0153] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for scheduling automatic computing tasks in a distribution network based on cloud-edge collaboration, characterized in that: include: Obtain a task set and a task feature vector of each task in the task set; wherein the task feature vector includes: a computational complexity index, a real-time requirement index, a data processing volume index, and a data security requirement index; the computational complexity index is used to measure the amount of computing resources required for the task, and the more computing resources the task requires, the higher the computational complexity; the real-time requirement index is used to measure the urgency of the task completion deadline, and the tighter the task completion deadline, the higher the real-time requirement index; the data processing volume index is used to measure the amount of data that the task needs to process, and the larger the amount of data that the task needs to process, the higher the data processing volume index; the data security requirement index is used to measure the degree of data security requirement of the task, and the higher the data security requirement of the task, the higher the data security requirement index; For each task, the scheduling decision index value of the task is calculated based on the task feature vector of the task and the calculation formula of the scheduling decision index; wherein the scheduling decision index is directly proportional to the computational complexity index, the data processing volume index, and the data security requirement index, and is inversely proportional to the real-time requirement index; Determine whether the scheduling decision indicator value is greater than a preset scheduling decision indicator threshold; if so, mark the task as a task to be uploaded; Upload all tasks marked as needing to be uploaded to the cloud for collaborative processing by the cloud, and process the remaining tasks locally.

2. The method for scheduling automatic computing tasks in a distribution network based on cloud-edge collaboration according to claim 1, characterized in that: The calculation formula of the scheduling decision index is: Where g(f(t)) represents the scheduling decision index value; t represents the task; f(t) represents the task feature vector of task t; C(t) represents the computational complexity index of task t; R(t) represents the real-time requirement index of task t; D(t) represents the data processing volume index of task t; S(t) represents the data security requirement index of task t; w1 is the weight coefficient of the computational complexity index; w2 is the weight coefficient of the real-time requirement index; w3 is the weight coefficient of the data processing volume index; and w4 is the weight coefficient of the data security requirement index.

3. The method for scheduling distribution network automation computing tasks based on cloud-edge collaboration according to claim 1, characterized in that: Also includes: Obtaining a node characteristic vector for each node in the distribution network; wherein the nodes include: cloud nodes and edge nodes; the node characteristic vector includes: a node comprehensive performance index and a node load index; wherein the node comprehensive performance index is used to measure the quality of the node comprehensive performance; the better the node comprehensive performance, the higher the node comprehensive performance index; the node load index is used to measure the level of the node load; the higher the node load, the higher the node load index; For each node, the task assignment probability value of the node is calculated based on the node feature vector of the node using the task assignment probability calculation formula; wherein the task assignment probability value is used to measure the probability of the node being assigned a task; the task assignment probability value is directly proportional to the node's comprehensive performance index and inversely proportional to the node's load index; For each task, if the task does not need to be executed across nodes, the task will be assigned to the node with the highest task assignment probability value; if the task needs to be executed across nodes, the task will be split into several subtasks; according to the number of subtasks, a corresponding number of nodes are selected in descending order of task assignment probability values as target nodes; and the tasks are evenly distributed to the target nodes.

4. The method for scheduling distribution network automation computing tasks based on cloud-edge collaboration according to claim 3 is characterized in that: The calculation formula of the task assignment probability is: Where i represents a node; P i represents the task assignment probability value of node i; W i represents the comprehensive performance index of node i; t i represents the node load index of node i; n represents the total number of nodes; λ1 is the first smoothing factor, which is used to adjust the influence of node load index on task allocation probability.

5. The method for scheduling distribution network automation computing tasks based on cloud-edge collaboration according to claim 1, characterized in that: Also includes: Obtain node performance indicators for each node in the distribution network and link performance indicators for each communication link in the distribution network; wherein the nodes include cloud nodes and edge nodes; the node performance indicators include: CPU usage, memory usage, network traffic, and response time; the link performance indicators include: latency, packet loss rate, bit error rate, delay, and link utilization; For each node, a node fault identification index value of the node is calculated based on the node performance index of the node using a calculation formula for the fault identification index; wherein the node fault identification index is used to measure the probability of a node fault; the higher the probability of a node fault, the higher the node fault identification index; Determining whether the node fault identification index value exceeds a preset node fault identification index threshold, and if so, performing fault detection on the node; For each communication link, a link fault identification index value of the link is calculated based on the link performance indicator of the communication link using a calculation formula for the fault identification index; wherein the link fault identification index is used to measure the probability of a communication link failure; the higher the probability of a communication link failure, the higher the link fault identification index; It is determined whether the link fault identification index value exceeds a preset link fault identification index threshold value; if so, a fault detection is performed on the communication link.

6. The method for scheduling distribution network automation computing tasks based on cloud-edge collaboration according to claim 5, characterized in that: The calculation formula of the fault identification index is: Where FI represents the fault identification index value; x j represents the j-th node performance index of a node or the j-th link performance index of a communication link; μ j represents the normal baseline value of the j-th node performance indicator or the normal baseline value of the j-th link performance indicator; σ j represents the standard deviation of the j-th node performance indicator or the j-th link performance indicator; m represents the total number of node performance indicators or the total number of link performance indicators; λ2 is the second smoothing factor, which is used to adjust the impact of abnormal node performance indicators on the node fault identification index or the impact of abnormal link performance indicators on the link fault identification index.

7. The method for scheduling distribution network automation computing tasks based on cloud-edge collaboration according to claim 5, characterized in that: Also includes: Obtaining a fault node performance indicator for each fault node in the distribution network and a fault link performance indicator for each fault communication link in the distribution network; For each faulty node, a node fault impact assessment coefficient value of the faulty node is calculated based on the faulty node performance indicator of the faulty node and using a calculation formula for the fault impact assessment coefficient; wherein the node fault impact assessment coefficient value is used to measure the fault severity of the faulty node; the higher the fault severity of the faulty node, the higher the node fault impact assessment coefficient; Sort all faulty nodes in descending order according to the node fault impact assessment coefficient values, use the sorting results as the faulty node repair order, and repair the faulty nodes according to the faulty node repair order; For each faulty communication link, a link fault impact assessment coefficient value of the faulty communication link is calculated according to the faulty link performance indicator of the faulty communication link and using a calculation formula for the fault impact assessment coefficient; wherein the link fault impact assessment coefficient value is used to measure the fault severity of the faulty communication link, and the higher the fault severity of the faulty communication link, the higher the link fault impact assessment coefficient; All faulty communication links are sorted in descending order according to the link fault impact assessment coefficient values, the sorting result is used as the faulty communication link repair sequence, and the faulty communication links are repaired according to the faulty communication link repair sequence.

8. The method for scheduling distribution network automation computing tasks based on cloud-edge collaboration according to claim 7, characterized in that: The calculation formula of the fault impact assessment coefficient is: Where E represents the fault impact assessment coefficient value.

9. A terminal device, characterized in that: It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the distribution network automation computing task scheduling method based on cloud-edge collaboration as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that include: A stored computer program, wherein, when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the distribution network automation computing task scheduling method based on cloud-edge collaboration as described in any one of claims 1-8.

Images