A method, equipment, and media for dynamic allocation and task scheduling of edge-cloud computing power in park operation and maintenance.

Abstract

This disclosure presents a method, device, and medium for dynamic allocation and task scheduling of edge-cloud computing power in park operations and maintenance, relating to the field of park operations and maintenance technology that integrates edge computing and cloud computing. The method includes: determining a first score for a first task in each dimension, and, based on a first weight, weighting and summing the first scores for each dimension to obtain a second score for the first task; determining the task priority corresponding to the second score, and, based on the task priority and the load status of candidate execution nodes, determining the target execution node type for the first task; using a genetic algorithm, determining the target execution node for the first task from multiple candidate execution nodes of the target execution node type; and executing the first task on the target execution node based on a protocol type matching the first task. This transforms park operations and maintenance from a passive, inefficient, and high-cost model to a proactive, intelligent, and highly economical new model, achieving comprehensive improvements in technical performance, commercial value, and system reliability.

Description

Technical Field

[0001] This disclosure relates to the field of park operation and maintenance technology that integrates edge computing and cloud computing, and in particular to a method, device and medium for dynamic allocation of edge-cloud computing power and task scheduling in park operation and maintenance. Background Technology

[0002] With the large-scale development of smart parks and industrial parks, park operation and maintenance scenarios are becoming increasingly complex, involving real-time monitoring and fault handling of various types of equipment such as power distribution, air conditioning, security, and fire protection, generating a massive number of operation and maintenance tasks. Currently, park operation and maintenance mostly adopts a single mode of "centralized cloud scheduling" or "independent edge operation," which has problems such as unclear task priorities, fixed computing power allocation, disconnection between cloud and edge collaboration, instability of operation and maintenance when the network is down, and lack of cross-domain scheduling. There is an urgent need for a complete solution for the entire chain of park operation and maintenance. Summary of the Invention

[0003] This disclosure provides a method, device, and medium for dynamic allocation and task scheduling of edge-cloud computing power in park operations and maintenance, in order to solve the following technical problem: how to provide a new intelligent and cost-effective operation and maintenance model for park operations and maintenance, so as to achieve comprehensive improvement in technical performance, commercial value, and system reliability.

[0004] In a first aspect, embodiments of this disclosure provide a method for dynamic allocation and task scheduling of computing power at the park's edge-cloud operation and maintenance site, the method comprising: Determine the first score of the first task for each dimension indicator, and based on the first weight, sum the first scores of each dimension indicator to obtain the second score of the first task; Determine the task priority corresponding to the second score, and based on the task priority and the load status of the candidate execution nodes, determine the target execution node type of the first task; Based on a genetic algorithm, the target execution node for the first task is determined from multiple candidate execution nodes of the target execution node type; The first task is executed on the target execution node based on the protocol type that matches the first task.

[0005] Secondly, this disclosure also provides a dynamic allocation and task scheduling system for edge-cloud computing power in park operations and maintenance, the system comprising: The scoring module is used to determine the first score of the first task in each dimension indicator, and based on the first weight, to perform a weighted sum of the first scores of each dimension indicator to obtain the second score of the first task. The determination module is used to determine the task priority corresponding to the second score, and to determine the target execution node type of the first task based on the task priority and the load status of the candidate execution nodes; A screening module is used to determine the target execution node of the first task from multiple candidate execution nodes of the target execution node type based on a genetic algorithm; An execution module is configured to execute the first task on the target execution node based on a protocol type that matches the first task.

[0006] Thirdly, this disclosure also provides a campus operation and maintenance edge-cloud computing power dynamic allocation and task scheduling device, the device including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method as described above.

[0007] Fourthly, embodiments of this disclosure also provide a computer storage medium storing computer-executable instructions, which, when executed, implement the method as described above.

[0008] The present disclosure provides a method, device, and medium for dynamic allocation and task scheduling of edge-cloud computing power in park operations and maintenance, which has the following beneficial effects: First, the first score for each dimension of the primary task is determined. Then, the first scores for each dimension are weighted and summed according to preset weights to obtain the second score for the primary task. This generates an objective, fair, and comparable second score for the task from multiple dimensions, thereby quantifying and accurately identifying task priorities. Next, the task priority corresponding to the second score is determined. Based on the task priority and node load status, the node type for executing the task is determined, enabling intelligent partitioning and load balancing of resource domains to ensure dynamic balance of global load. Then, a genetic algorithm is used to determine the target execution node for the primary task from multiple candidate execution nodes of the target execution node type. By finding a suitable target execution node for the task through the genetic algorithm, the optimization of task scheduling decisions and multi-objective balancing are achieved. Finally, the primary task is executed on the target execution node according to the protocol type matched to the primary task. Matching the most suitable communication protocol to the task based on task priority ensures that instructions for high-priority tasks arrive at the execution node in a timely manner and are executed, thereby achieving differentiated services and optimized utilization of network resources. Attached Figure Description

[0009] The accompanying drawings, which are included to provide a further understanding of this disclosure and form part of this disclosure, illustrate exemplary embodiments of the present disclosure and are used to explain the disclosure, but do not constitute an undue limitation of the disclosure. In the drawings: Figure 1A flowchart of a method for dynamic allocation of edge-cloud computing power and task scheduling in park operation and maintenance, provided in an embodiment of this disclosure; Figure 2 This is a schematic diagram of the internal structure of a device provided in an embodiment of this disclosure. Detailed Implementation

[0010] To make the objectives, technical solutions, and advantages of this disclosure clearer, the technical solutions of this disclosure will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.

[0011] It is understood that in the embodiments of this disclosure, data related to user information (such as user accounts) is involved. When the embodiments of this disclosure are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with relevant laws, regulations and standards.

[0012] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. The terminology used herein is for the purpose of describing embodiments of this disclosure only and is not intended to be limiting of this disclosure.

[0013] In the following description, the terms “first, second, ...” are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that “first, second, ...” may be interchanged in a specific order or sequence where permitted, so that the embodiments of this disclosure described herein can be implemented in an order other than that illustrated or described herein.

[0014] The technical solutions proposed in the embodiments of this disclosure will be described in detail below with reference to the accompanying drawings.

[0015] Figure 1 This document presents a flowchart illustrating a method for dynamic allocation of edge-cloud computing power and task scheduling in park operations and maintenance, provided in one or more embodiments. This method can be applied to various types of park operations and maintenance scenarios, such as smart parks, industrial parks, and commercial complexes. Certain input parameters or intermediate results in the process can be manually adjusted to help improve accuracy.

[0016] This disclosure provides a method for dynamic allocation and task scheduling of edge-cloud computing power in park operations and maintenance. It should be noted that the execution entity in these embodiments can be a server or any terminal device with data processing capabilities. For example, the server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. The terminal device can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, in-vehicle terminal, etc., but is not limited to these.

[0017] The technical solutions proposed in the embodiments of this disclosure will be described in detail below with reference to the accompanying drawings.

[0018] like Figure 1 As shown in the figure, the present disclosure provides a method for dynamic allocation and task scheduling of edge-cloud computing power in park operations and maintenance, which specifically includes the following steps: Step 101: Determine the first score for each dimension indicator of the first task.

[0019] It should be noted that the first task can be any independent operation and maintenance work unit in the park that needs to be processed by the system, such as voltage abnormality alarm in the power distribution room, shutdown of the air conditioning system in production workshop A area, or fire equipment inspection reminder in the office building. The dimensional indicators refer to a set of multi-angle evaluation standards that are set up in advance to scientifically and quantitatively assess the urgency and importance of the first task. The dimensional indicators may include real-time performance (task response time limit), fault level (severity of equipment failure), scope of impact (number of equipment and personnel covered), handling timeliness (risk of fault spread), equipment importance (core equipment or ordinary equipment), data scale (amount of data processed by the task), etc. The first score refers to a quantitative and standardized score given for each dimensional indicator of the first task.

[0020] As an example, assuming the first task is a voltage anomaly in the power distribution room, according to the preset scoring rules, for each dimension indicator, the first score corresponding to the real-time dimension indicator is 100 points, the first score corresponding to the fault level dimension indicator is 90 points, the first score corresponding to the impact scope dimension indicator is 80 points, the first score corresponding to the handling timeliness dimension indicator is 85 points, the first score corresponding to the equipment importance dimension indicator is 95 points, and the first score corresponding to the data scale dimension indicator is 20 points.

[0021] Step 102: Based on the first weight, the first scores of each dimension indicator are weighted and summed to obtain the second score of the first task.

[0022] It should be noted that the first weight can be a preset value or dynamically determined according to the actual scenario, and no specific limitation is made here; the second score is a comprehensive quantitative result of the first task and is a direct basis for judging the task priority.

[0023] Continuing with the example above, assuming the first weight for the real-time dimension indicator is 0.3, the first weight for the fault level dimension indicator is 0.25, the first weight for the impact scope dimension indicator is 0.2, the first weight for the handling timeliness dimension indicator is 0.1, the first weight for the equipment importance dimension indicator is 0.1, and the first weight for the data scale dimension indicator is 0.05, based on the first weight, the first score of each dimension indicator is weighted and summed to obtain the second score of the first task as 87.5 points.

[0024] Step 103: Determine the task priority corresponding to the second score.

[0025] It should be noted that task priority is the final classification of the urgency, importance and processing registration of the first task in the entire park operation and maintenance system. It is used to guide the system to allocate resources differently to ensure that the most critical tasks can be processed with the highest priority, the most sufficient and the fastest speed. The task priority of the corresponding task can be determined according to the preset score threshold range.

[0026] Continuing with the example above, suppose the scoring rules for determining task priorities are: "Task priority: P1 level, scoring range: 90-100; Task priority: P2 level, scoring range: 80-89; Task priority: P3 level, scoring range: 70-79; Task priority: P4 level, scoring range: 50-69; Task priority: P2 level, scoring range: 0-49". The second score for the first task is 87.5 points. According to the preset scoring rule range, the task priority of the first task can be determined to be P1 level.

[0027] Step 104: Based on the task priority and the load status of the candidate execution nodes, determine the target execution node type of the first task.

[0028] It should be noted that the load status of the candidate execution node refers to a real-time quantitative description of the busyness of the available resource pool (edge ​​node cluster or cloud node cluster) at the current moment; the type of the target execution node can include edge nodes and cloud nodes.

[0029] In some embodiments, step 104 described above can be implemented as follows: in response to the task priority being a first priority and the load state of the candidate execution node being less than a preset load threshold, an edge node is selected as the target execution node type of the first task; in response to the task priority being a first priority and the load state of the candidate execution node being greater than a preset load threshold, both the edge node and the cloud node are selected as the target execution node types of the first task; in response to the task priority being a second priority and the load state of the candidate execution node being less than a preset load threshold, the edge node is selected as the target execution node type of the first task; in response to the task priority being a second priority and the load state of the candidate execution node being greater than a preset load threshold, the cloud node is selected as the target execution node type of the first task.

[0030] Thus, when a task has a high priority and the edge nodes are idle, the task is assigned to an edge node for execution. When a task has a high priority and the edge nodes are busy, the task is assigned to both edge nodes and cloud nodes. This ensures that high-priority tasks are processed quickly while making the load on each node more reasonable, achieving a dynamic balance between the rapid processing of important tasks and the overall system load. When a task has a low priority and the edge nodes are idle, the task is assigned to an edge node for execution. When a task has a low priority and the edge nodes are busy, the task is assigned to a cloud node for execution. This allows low-priority tasks to be transferred from busy edge nodes to cloud nodes, ensuring that tasks are completed while reserving more edge computing power for high-priority tasks, preventing the entire system from crashing due to overload, thereby achieving a significant optimization of global resource utilization.

[0031] It should be noted that the first priority is higher than the second priority. The first priority and the second priority can refer to a priority (for example, the first priority is P1 level and the second priority is P2 level), or they can refer to a range of priorities (for example, the first priority is P1-P2 level and the second priority is P3-P5 level). No specific limitation is made here.

[0032] As an example, suppose a non-core air conditioner in an office building malfunctions, triggering a device alarm task. Through task analysis and priority determination, the current device alarm task has a priority of P2, which is the first priority. The computing power awareness agent reports that the average CPU utilization of the current edge node cluster (the load status of candidate execution nodes) is 92%, which is greater than the preset load threshold (85%). In this case, both edge nodes and cloud nodes are used as the target execution node types for the device alarm task. That is, by coordinating edge nodes and cloud nodes, the current device alarm task can be processed in a timely and comprehensive manner. Assuming the system currently has a data statistics task, through task analysis and priority determination, the current data statistics task has a priority of P4, which is the second priority. The computing power awareness agent reports that the average CPU utilization of the current edge node cluster (the load status of candidate execution nodes) is 90%, which is greater than the preset load threshold (85%). In this case, cloud nodes are used as the target execution node type for the data statistics task, that is, cloud nodes are used as the processing nodes for the data statistics task, reserving more edge computing power for higher-priority tasks.

[0033] Step 105: Based on a genetic algorithm, determine the target execution node for the first task from multiple candidate execution nodes of the target execution node type.

[0034] It should be noted that the genetic algorithm is a search algorithm that draws on the principles of "natural selection" and "genetic evolution" in the biological world to find the optimal solution to complex problems. Instead of deriving it step by step through traditional mathematical formulas, this algorithm simulates an evolutionary process of "survival of the fittest" and allows multiple candidate solutions to gradually evolve into the optimal solution through competition and cooperation.

[0035] As an example, suppose the voltage anomaly task (task 1) in substation No. 1 is triggered. According to the decision logic, the target execution node type for this task is an edge node. Currently, there are three candidate execution nodes: Edge-1 (geographically closest to substation No. 1), Edge-2 (geographically slightly farther away, currently idle), and Edge-3 (geographically in the center, currently processing a P3-level task). First, a batch of complete allocation schemes is randomly generated to form the initial population (Scheme A: Select Edge-1, Scheme B: Select Edge-2, Scheme C: Select Edge-1). 3) Subsequently, fitness evaluation is performed on each scheme to obtain a fitness score for each scheme. The fitness score of scheme A is 95 points, that of scheme B is 92 points, and that of scheme C is 80 points. Then, based on the scores, schemes A and B are likely to be selected as "parent" schemes because of their high scores, while scheme C is likely to be eliminated. After multiple rounds of rapid iteration, individuals in the population will be highly inclined towards the schemes with the highest scores. In the last round, the scheme with the highest fitness score is selected as Edge-1 as the target execution node for the current task.

[0036] In some embodiments, step 105 described above can be implemented as follows: for each candidate execution node, perform the following processing: determine the minimum task response latency of the first task on the candidate execution node; determine the maximum computing power utilization of the candidate execution node; determine the minimum cross-domain transmission cost of the first task on the candidate execution node; determine the fitness score of the candidate execution node based on the minimum task response latency, the maximum computing power utilization, and the minimum cross-domain transmission cost; and select the candidate execution node with the highest fitness score as the target execution node of the first task.

[0037] Thus, by determining the minimum task response latency, end-to-end low latency for critical tasks can be ensured; by determining the maximum computing power utilization, resource waste can be avoided and return on investment can be maximized; by determining the minimum cross-domain transmission cost, technical optimization and commercial value can be effectively combined, and computing power costs can be effectively reduced. By integrating the data of these three indicators, an effective trade-off can be made among them to find the best balance between conflicting goals.

[0038] It should be noted that the minimum task response latency is the predicted total time required for the first task to be processed and completed on a candidate execution node from the start of scheduling; the maximum computing power utilization rate is the predicted computing power utilization efficiency that a candidate execution node can achieve after the first task is assigned to it; and the minimum cross-domain transmission cost is the predicted comprehensive cost incurred in transmitting the data of the first task from its place of origin to the execution node.

[0039] As an example, suppose the voltage anomaly task of substation No. 1 (the first task) is triggered. According to the decision logic, the target execution node type of the current task is an edge node. For Edge-A, the minimum task response latency is determined to be 22ms, the maximum computing power utilization rate is 60%, and the minimum cross-domain cost is 0. The fitness function is used to fuse the three dimensions of indicators, and the fitness score of Edge-A is 105.5. Similarly, the fitness score of Edge-B is 63.3 and the fitness score of Edge-C is 70.9. Finally, the candidate execution node with the highest fitness score is selected as the target execution node of the first task, that is, the target execution node of the current task is Edge-A.

[0040] In some embodiments, determining the minimum task response latency of the first task on the candidate execution node can be achieved as follows: determining the queuing latency of the candidate execution node based on the waiting queue length of the candidate execution node and the estimated execution time of each waiting task; mapping the performance indicators of the candidate execution node to obtain the task execution latency of the candidate execution node; determining the network transmission latency of the candidate node based on the data volume of the first task and the network parameters of the candidate execution node; and summing the queuing latency, the task execution latency, and the network transmission latency to obtain the minimum task response latency of the first task on the candidate execution node.

[0041] Thus, based on the waiting queue length of the candidate execution nodes and the estimated execution time of each waiting task, the actual start time of the current task can be accurately predicted. Furthermore, by mapping the performance indicators of the candidate execution nodes, the task execution latency of the candidate execution nodes can be predicted, allowing for an accurate assessment of the time required for the candidate execution node to execute the current task. Simultaneously, based on the data volume of the task and the network parameters of the candidate execution nodes, the network transmission latency of the candidate execution nodes can be predicted, reflecting the impact of network conditions on data transmission in real time. Finally, by integrating data from different dimensions, the minimum task response latency of the current candidate execution node can be determined, thereby achieving scientific, accurate, and forward-looking response latency prediction, providing a solid and reliable data foundation for the entire intelligent scheduling system.

[0042] It should be noted that estimated execution time refers to the time it would take for each task already waiting in the queue to be executed independently on the current candidate execution node; queuing latency refers to the total time that the first task must wait for the candidate execution node to complete all the tasks ahead of it in its current queue after being assigned to it; the performance metrics of the candidate execution node refer to a set of quantifiable hardware and system parameters used to describe the node's computing and data processing capabilities, such as the currently available CPU frequency, the number of idle CPU cores, the amount of available memory, and disk IOPS; task execution latency refers to the time it takes for the CPU of the first task to actually start processing data and calculate the final result, excluding queuing and network transmission time; network transmission latency refers to the time it takes for the data required by the first task to be transmitted from its origin (e.g., a sensor) to the candidate execution node via the network.

[0043] As an example, suppose the voltage anomaly task in substation No. 1 (the first task) is triggered. According to the decision logic, the target execution node type for the current task is an edge node. For Edge-A, detection reveals two tasks in its CPU waiting queue. Based on historical data and node performance, the estimated execution times for these two waiting tasks are estimated to be 200ms and 50ms respectively. Calculations show that the queuing delay for these tasks on the current candidate execution node is 250ms. Subsequently, the performance metrics of the candidate execution node are obtained: Edge-A's CPU clock speed is 3.0GHz. Through mapping processing, it can be calculated that... The task execution latency on Edge-A is 83ms. The current task data size is 10MB. Edge-A's network parameters are a 1Gbps (125MB / s) LAN. The network is currently idle, with available bandwidth close to the theoretical value. The inherent network latency (ping value) is 0.5ms. Calculations show that the network transmission latency on Edge-A is 80ms. Finally, adding the queuing latency of 250ms, the task execution latency of 83ms, and the network transmission latency of 80ms, we obtain the minimum task response latency on Edge-A as 163ms.

[0044] In some embodiments, determining the minimum cross-domain transmission cost of the first task at the candidate execution node can be achieved as follows: determining the network bandwidth cost of the first task at the candidate execution node based on the data transmission type of the first task and the location information of the candidate execution node; determining the data security cost of the first task at the candidate execution node based on the task priority of the first task and the security attributes of the candidate execution node; determining the resource preemption cost of the first task at the candidate execution node based on the real-time status of the network link occupied by the task data transmission; and summing the network bandwidth cost, the data security cost, and the resource preemption cost to obtain the minimum cross-domain transmission cost of the first task at the candidate execution node.

[0045] In this way, network bandwidth costs are determined based on data transmission type and node location information, aligning scheduling decisions with computing power costs. This ensures that the final decision takes computing power costs into account. Simultaneously, data security costs are calculated based on task priorities and node security attributes, effectively considering task execution security and preventing catastrophic losses due to data leaks. Furthermore, resource preemption costs for nodes are calculated based on real-time network link status, making task scheduling decisions more socially oriented and ensuring the smooth operation of the entire park's business. By combining network bandwidth costs, data security costs, and resource preemption costs, park operations and maintenance are elevated from simple performance optimization to a new level of comprehensive value optimization that includes economic considerations and risk management.

[0046] It should be noted that network bandwidth cost refers to the monetary value of network resources directly consumed in transmitting the data of the first task from its origin to the candidate execution node; the security attributes of the execution node refer to a set of quantifiable characteristics describing the security protection capabilities and levels of the node and its network environment; data security cost refers to the estimated loss caused by potential risks such as data leakage, tampering, or destruction when transmitting the data of the first task to the candidate execution node for processing; and resource preemption cost refers to the estimated loss caused by the network link bandwidth occupied by the data transmission of the first task, which may cause performance degradation or interruption to other services (such as video conferencing and remote control) on the same link.

[0047] As an example, assuming the voltage anomaly task in power distribution room 1 (the first task) is triggered, the target execution node type for the current task is determined to be an edge node based on the decision logic. For Edge-A, the data transmission type of the current task is a 1GB large file data packet, and the location information of the candidate execution node is from the local campus to the remote cloud, which is a cross-domain transmission. The 5G private network is billed by traffic at a unit price of 5 yuan / GB. Through calculation, the network bandwidth cost of the current task in Edge-A can be determined to be 5 yuan. Subsequently, the priority of the current task is obtained as P1 level, and the node security attributes of Edge-A are "using the national cryptographic algorithm IPSec VPN, data is stored in encrypted form (AES-256), and physically isolated in the data center". It can be determined that the data security cost of Edge-A is 10. Then, the bandwidth utilization rate of the current network link is obtained to be 90%, which is very busy. Through evaluation and analysis, the resource preemption cost of Edge-A can be determined to be 2000. Finally, by adding the network bandwidth cost, data security cost, and resource preemption cost, the minimum cross-domain transmission cost of Edge-A can be obtained as 2015.

[0048] Step 106: Execute the first task on the target execution node based on the protocol type that matches the first task.

[0049] It should be noted that the protocol types include Transmission Control Protocol (TCP) and User Datagram Protocol (UDP). TCP is a reliable, one-to-one, connection-oriented communication protocol that provides a reliable communication connection for applications, enabling the error-free transmission of byte streams from one computer to other computers on the network. Data communication systems with high reliability requirements often use TCP to transmit data. UDP, on the other hand, is an unreliable, connectionless communication protocol that can implement many-to-one, one-to-many, and one-to-one connections. It is suitable for application environments that transmit only small amounts of data at a time and have low reliability requirements.

[0050] In some embodiments, step 106 described above can be implemented in the following ways: in response to the task priority of the first task being the third priority, the scheduling instruction of the first task is transmitted at the target execution node using the User Datagram Protocol; in response to the task priority of the first task being the fourth priority, the scheduling instruction and data synchronization information of the first task are transmitted at the target execution node using the Transmission Control Protocol.

[0051] In this way, using different protocols for instruction transmission for tasks with different priorities can maximize the utilization efficiency of network bandwidth, and making decisions through simple rule design makes the system easier to implement and maintain, thereby improving the overall stability and robustness of the system.

[0052] It should be noted that the third priority and the fourth priority can refer to a single priority (e.g., the third priority is P1 level and the fourth priority is P2 level), or they can refer to a range of priorities (e.g., the third priority is P1-P3 level and the fourth priority is P4-P5 level). No specific limitation is made here.

[0053] As an example, a temperature sensor deployed in warehouse 3 detects that the temperature continuously exceeds a preset threshold (e.g., 28°C), triggering an environmental anomaly task (the first task). Through task analysis and task priority determination, the task priority of the current device alarm task is determined to be P3, which belongs to the third priority (P1-P3). According to the protocol selection logic, it can be determined that the UDP protocol will be used to transmit the scheduling instruction. Then, the instruction { "cmd": "check_fan_status", "location": "warehouse_3"} is encapsulated into a UDP data packet and sent to Edge-W (the target execution node) with extremely low network overhead and extremely fast speed (latency ≤50ms). After receiving the instruction, Edge-W immediately starts the warehouse exhaust fan to cool it down and quickly feeds back the processing result.

[0054] In another example, the cloud-based backend service automatically completed the statistics of water, electricity, and gas usage data for the entire park last month in the early morning, generating an energy consumption monthly report task. The data needs to be synchronized to the operations and maintenance administrator's database. Through task analysis and task priority determination, the task priority of the current device alarm task is determined to be level P4, which belongs to the fourth priority (P4-P5 level). According to the protocol selection logic, it can be determined that the TCP protocol will be used to transmit scheduling instructions and report data. A reliable connection is established through TCP to Cloud-D (the target execution node) to ensure that Cloud-D is ready to receive data. Then, 50MB of energy consumption report data is transmitted completely, orderly, and error-free from the application service to Cloud-D through the reliable TCP transmission channel. After the transmission is completed, the TCP connection will be confirmed to ensure that every byte of data has arrived correctly. If there is network jitter in the middle, TCP will automatically retransmit lost data packets.

[0055] In some embodiments, after performing step 106, the following processes may also be performed: in response to the disconnection between the edge node and the cloud node, a second task within a preset priority range is determined; a preset proportion of computing power is allocated to the edge node so that the edge node can perform the second task; the business data generated by the second task is cached and stored in a hierarchical manner, and after the connection is restored, the business data is synchronized to the cloud node.

[0056] Thus, in the event of a network outage, tasks can be prioritized and non-core services can be proactively suspended or abandoned, avoiding unnecessary consumption of computing resources. At the same time, allocating a preset proportion of computing power to edge nodes can provide a stable and predictable operating environment for core business operations, thereby enhancing system robustness. Combined with hierarchical caching and data synchronization, the traceability of the execution process of critical services during network outages and the synchronizability of results can be ensured, thus guaranteeing the security of core data.

[0057] As an example, suppose a large chemical industrial park with independent production and office networks experiences a complete disconnection between the entire park's production area and the cloud data center due to a fiber optic cable being severed during municipal construction. The outage is expected to last for 2 hours. Currently executing tasks include one P1-level task (pressure anomaly alarm in reactor No. 2), one P3-level task (equipment inspection reminder in production workshop A), and one P4-level task (generation of last month's material consumption data statistical report for the park). After the outage, the local decision engine deployed on each edge node is activated and begins executing emergency procedures. The local decision engine immediately scans the task queue being processed by its node. The system's default strategy is to only guarantee tasks at levels P1-P3 (preset priority range) in emergency mode. After filtering, the second priority tasks include P1 pressure anomaly and P4-level tasks. The system performs equipment inspection tasks 3, and then allocates 70% of the computing power to the current node to execute P1 pressure anomaly and P3 equipment inspection tasks. During task execution, all generated data is intelligently classified and stored by the local decision engine. Two hours later, the network is restored, and the local decision engine of the edge node detects that the connection has been restored and immediately enters the data synchronization state. The system performs consistency verification on the backup data in the core data area and quickly synchronizes the complete data to the cloud within 10 seconds. The cloud data center immediately restores the complete fault handling record of reactor No. 2, and also synchronizes the compressed inspection record to the cloud. After all data synchronization is completed, the local decision engine releases the locked 70% computing power, resumes the execution of the P4 report task, and reconnects to the global scheduling system in the cloud. The entire park operation and maintenance system returns to normal.

[0058] In some embodiments, after performing step 106, the following processes may also be performed: collecting execution status data of the first task based on a preset frequency; transmitting the execution status data to a cloud policy center based on a protocol type matching the first task; receiving a task scheduling policy generated by the cloud policy center based on the execution status data, and adjusting the execution of the first task based on the task scheduling policy.

[0059] In this way, by collecting and transmitting task data according to a preset frequency, real-time visualization of task status can be achieved. Through differentiated protocol transmission, it can be ensured that this status data can be transmitted to the cloud efficiently and reliably, so that the cloud can have a microscopic and quantitative real-time perception of the operation status of the entire park, and determine a reasonable task scheduling strategy based on the status data, so as to make adaptive adjustments to the execution of the first task according to the task scheduling strategy, thereby making the system scheduling decision more intelligent.

[0060] As an example, suppose a smart park with multiple edge nodes and a cloud policy center is configured. A P1-level motor overheating task on the production line of workshop A is triggered and scheduled by the cloud policy center to edge node Edge-A for execution. While executing the P1 task, Edge-A's built-in computing power awareness agent starts working, collecting the execution status data of the P1 task every 100ms and sending it to the cloud policy center via UDP. Upon receiving the status data stream from Edge-A, the cloud policy center determines that Edge-A may have experienced a hardware failure or a network attack, and re-runs the genetic algorithm, removing Edge-A from the candidate node list. The algorithm then finds the optimal replacement node, Edge-B (idle and with a stable network), among the remaining edge nodes. Simultaneously, the cloud policy center generates a new scheduling policy: { "cmd": "migrate_task", "task_id": "P1_Motor_Overheat", "from": "Edge-A", "to": "Edge-B", "sync_state": true Upon receiving the new scheduling policy, the system immediately suspends Edge-A's execution of task P1 and packages the current execution status data (18% progress, calculation results in memory, etc.) and sends it to Edge-B so that Edge-B can continue executing task P1.

[0061] The following will describe an exemplary application of the embodiments of this disclosure in a practical application scenario.

[0062] With the large-scale development of smart parks and industrial parks, park operation and maintenance scenarios are becoming increasingly complex, involving real-time monitoring and fault handling of various types of equipment such as power distribution, air conditioning, security, and fire protection, resulting in a massive number of operation and maintenance tasks. Currently, park operation and maintenance mostly adopts a single mode of "centralized cloud scheduling" or "independent edge operation," which presents the following technical pain points: (1) Vague task priority: Traditional solutions often use a “simple classification of fault types” to determine task priority, without forming a quantitative evaluation system. This results in critical faults (such as power distribution faults) competing with ordinary inspection tasks for computing resources, and the response delay of core tasks is too high.

[0063] (2) Fixed computing power allocation: The computing power allocation between the edge and the cloud adopts a static configuration mode, which cannot be dynamically adjusted according to the real-time task load and computing power status. The computing power utilization rate of edge nodes is generally only 60%-70%, while the cloud often suffers from overload during peak hours and idle resources during off-peak hours.

[0064] (3) Cloud-edge collaboration gap: There is a lack of efficient collaboration mechanism between cloud and edge. There is a delay of 3-10 seconds between the issuance of scheduling instructions and the feedback of execution. Furthermore, the policy adjustment is not linked with the actual operation scenario, resulting in "disconnect between policy and execution" and failing to adapt to the dynamic changes in park operation and maintenance requirements.

[0065] (4) Network outage and unstable operation and maintenance: The existing system relies too much on network connectivity. Once the network is interrupted, the edge nodes cannot receive instructions from the cloud, resulting in the complete cessation of operation and maintenance tasks, interruption of core equipment monitoring and fault handling, and serious security risks.

[0066] (5) Lack of cross-domain scheduling: In the case of group-based multi-park operation and maintenance, each park operates independently and cannot achieve global allocation of computing resources. When some parks are short of computing power, the idle computing power of other parks cannot be filled, resulting in high overall operation and maintenance costs.

[0067] To address the core pain points of existing cloud-edge collaborative systems for park operations and maintenance, such as "fuzzy task priorities, fixed computing power allocation, disconnected cloud-edge collaboration, and unstable operations and maintenance when the network is down," this disclosure provides a low-latency, highly available, and low-cost method for dynamic allocation of computing power and task scheduling between the edge and cloud for park operations and maintenance.

[0068] In some embodiments, the park operation and maintenance edge-cloud computing power scheduling system is deployed in an industrial park (covering an area of ​​500 acres, including 10 production workshops, 3 power distribution rooms, and 2 office buildings), with the following specific configuration: Terminal layer: 500 operation and maintenance sensors (temperature, humidity, current, voltage sensors, etc.), 30 monitoring cameras, and 50 device controllers are deployed. Operation and maintenance personnel are equipped with 20 smart APP terminals, which access the park's intranet via LoRa / Wi-Fi / 5G; Edge layer: 10 edge nodes are deployed (using Intel Core i7-12700 processors, 32GB of memory, and 1TB of SSD storage), each node covering one production workshop or key area, and the nodes are interconnected via industrial Ethernet; the edge nodes deploy a computing power perception agent (developed based on Prometheus), a local decision engine (developed based on the Python Flask framework), and a hierarchical caching module (implemented based on Redis+SSD); Cloud layer: 6 RH2288H servers are deployed. The V5 servers form a cluster, with 3 serving as computing power scheduling nodes (running an improved genetic algorithm scheduler), 2 serving as data storage nodes (using a MySQL master-slave architecture + MinIO object storage), and 1 serving as a cross-domain collaboration node (deploying VPN service and encrypted transmission module). The cluster is connected to the Internet via a 10Gbps fiber optic cable. At the cross-domain layer, the regional dispatch center deploys 2 SR860 servers (dual-machine hot standby) and runs a resource management platform (developed based on Kubernetes) to integrate the computing power resources of this industrial park and two surrounding commercial parks, connecting to each park via a 5G private network.

[0069] In some embodiments, taking the "voltage anomaly in power distribution room No. 1" (P1 level task) in the industrial park as an example, the terminal layer voltage sensor collects voltage deviation threshold data (10kV→8.5kV) and uploads it to the edge node; the priority modeling module scores based on a six-dimensional model: real-time performance (30 points, response time ≤10s), fault level (25 points, severe fault), impact range (20 points, covering 1 power distribution room + 3 production workshops), handling timeliness (10 points, may cause power outage within 10 minutes), equipment importance (10 points, core power distribution equipment), and data scale (5 points, small data volume), with a total score of 90 points, which is judged as a P1 level task; subsequently, the edge node computing power perception agent collects that the current edge load is 70% (30% idle computing power), which meets the requirements of the P1 level task; after the improved genetic algorithm makes a decision, 20% of the edge node's computing power is allocated to voltage anomaly data processing and fault location. 0% of computing power is used for command issuance and device control; then the edge node performs fault location (takes 30ms) and feeds back the fault information to the cloud policy center via UDP protocol (transmission delay 40ms); the cloud generates a "adjust transformer voltage regulation level" command and sends it to the edge node via UDP protocol (transmission delay 35ms); after the edge node executes the command, it provides real-time feedback on the execution result (execution successful, voltage restored to 9.8kV), the entire collaborative process takes 105ms; finally, during the fault handling process, a network outage in the park is simulated (fiber optic connection is disconnected); the edge node determines the network outage within 180ms through heartbeat detection (no heartbeat received for 3 consecutive times), switches to emergency mode, and locks 70% of computing power to ensure voltage monitoring and stable control; within 5 minutes of network outage, the edge node continuously monitors the voltage status and caches data; after the network is restored, the fault handling data and monitoring logs are synchronized to the cloud within 10 seconds.

[0070] Through the above-described embodiments, the time taken for a P1-level task from detection to completion is 8 minutes, which is 50% shorter than the existing system (16 minutes); the edge computing power utilization rate is increased to 85%, with no computing power wastage; the task continues to execute during network outages, and core data is not lost. This can provide a new intelligent and highly economical operation and maintenance model for park operations and maintenance, achieving comprehensive improvements in technical performance, commercial value, and system reliability.

[0071] The above are embodiments of the methods proposed in this disclosure. Additionally, embodiments of the present invention also include a dynamic allocation and task scheduling system for edge-cloud computing power in park operations and maintenance, specifically comprising: a scoring module, a determination module, a filtering module, and an execution module.

[0072] The scoring module is used to determine the first score of the first task in each dimension indicator, and to perform a weighted summation of the first scores of each dimension indicator based on a first weight to obtain the second score of the first task; the determination module is used to determine the task priority corresponding to the second score, and to determine the target execution node type of the first task based on the task priority and the load status of the candidate execution nodes; the filtering module is used to determine the target execution node of the first task from multiple candidate execution nodes of the target execution node type based on a genetic algorithm; and the execution module is used to execute the first task on the target execution node based on a protocol type that matches the first task.

[0073] In some embodiments, the determining module is further configured to: in response to the task priority being a first priority and the load state of the candidate execution node being less than a preset load threshold, designate an edge node as the target execution node type of the first task; in response to the task priority being a first priority and the load state of the candidate execution node being greater than a preset load threshold, designate both the edge node and the cloud node as target execution node types of the first task; in response to the task priority being a second priority and the load state of the candidate execution node being less than a preset load threshold, designate the edge node as the target execution node type of the first task; and in response to the task priority being a second priority and the load state of the candidate execution node being greater than a preset load threshold, designate the cloud node as the target execution node type of the first task.

[0074] In some embodiments, the filtering module is further configured to perform the following processes for each candidate execution node: determine the minimum task response latency of the first task on the candidate execution node; determine the maximum computing power utilization of the candidate execution node; determine the minimum cross-domain transmission cost of the first task on the candidate execution node; determine the fitness score of the candidate execution node based on the minimum task response latency, the maximum computing power utilization, and the minimum cross-domain transmission cost; and select the candidate execution node with the highest fitness score as the target execution node of the first task.

[0075] In some embodiments, the filtering module is further configured to: determine the queuing delay of the candidate execution node based on the waiting queue length of the candidate execution node and the estimated execution time of each waiting task; map the performance indicators of the candidate execution node to obtain the task execution delay of the candidate execution node; determine the network transmission delay of the candidate execution node based on the data volume of the first task and the network parameters of the candidate execution node; and sum the queuing delay, the task execution delay, and the network transmission delay to obtain the minimum task response delay of the first task on the candidate execution node.

[0076] In some embodiments, the filtering module is further configured to: determine the network bandwidth cost of the first task at the candidate execution node based on the data transmission type of the first task and the location information of the candidate execution node; determine the data security cost of the first task at the candidate execution node based on the task priority of the first task and the security attributes of the candidate execution node; determine the resource preemption cost of the first task at the candidate execution node based on the real-time status of the network link occupied by the task data transmission; and sum the network bandwidth cost, the data security cost, and the resource preemption cost to obtain the minimum cross-domain transmission cost of the first task at the candidate execution node.

[0077] In some embodiments, the execution module is further configured to, in response to the task priority of the first task being a third priority, transmit the scheduling instruction of the first task at the target execution node using the User Datagram Protocol; and in response to the task priority of the first task being a fourth priority, transmit the scheduling instruction and data synchronization information of the first task at the target execution node using the Transmission Control Protocol.

[0078] In some embodiments, the execution module is further configured to, in response to the disconnection between the edge node and the cloud node, determine a second task within a preset priority range; allocate a preset proportion of computing power to the edge node so that the edge node executes the second task; perform hierarchical caching and storage of the business data generated by the second task; and synchronize the business data to the cloud node after the connection is restored.

[0079] In some embodiments, the execution module is further configured to collect execution status data of the first task based on a preset frequency; transmit the execution status data to a cloud policy center based on a protocol type matching the first task; and receive a task scheduling policy generated by the cloud policy center based on the execution status data, so as to adjust the execution of the first task based on the task scheduling policy.

[0080] The above are embodiments of the method proposed in this disclosure. Based on the same inventive concept, embodiments of this disclosure also provide a device, the structure of which is as follows: Figure 2 As shown.

[0081] Figure 2 This is a schematic diagram of the internal structure of a device provided in an embodiment of this disclosure. For example... Figure 2 As shown, the device includes: At least one processor 201; And a memory 202 that is communicatively connected to at least one processor; The memory 202 stores instructions that can be executed by at least one processor. The instructions are executed by at least one processor 201 to enable at least one processor 201 to perform the steps of the method corresponding to any of the above embodiments.

[0082] Some embodiments of this disclosure provide corresponding to Figure 1 A non-volatile computer storage medium stores computer-executable instructions configured to perform the steps of the method corresponding to any of the above embodiments.

[0083] The various embodiments in this disclosure are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for IoT devices and media are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0084] The systems, media, and methods provided in this disclosure are one-to-one correspondences. Therefore, the systems and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the systems and media will not be repeated here.

[0085] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0086] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0087] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0088] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0089] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0090] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0091] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0092] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0093] The above description is merely an embodiment of this disclosure and is not intended to limit the scope of this disclosure. Various modifications and variations can be made to this disclosure by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the scope of the claims of this disclosure.

Claims

1. A method for dynamic allocation and task scheduling of edge-cloud computing power in park operation and maintenance, characterized in that, The method includes: Determine the first score of the first task for each dimension indicator, and based on the first weight, sum the first scores of each dimension indicator to obtain the second score of the first task; Determine the task priority corresponding to the second score, and based on the task priority and the load status of the candidate execution nodes, determine the target execution node type of the first task; Based on a genetic algorithm, the target execution node for the first task is determined from multiple candidate execution nodes of the target execution node type; The first task is executed on the target execution node based on the protocol type that matches the first task.

2. The method according to claim 1, characterized in that, Determining the target execution node type for the first task based on the task priority and the load status of the candidate execution nodes includes: In response to the task priority being first priority and the load status of the candidate execution node being less than a preset load threshold, the edge node is selected as the target execution node type for the first task; In response to the task priority being first priority and the load status of the candidate execution node being greater than a preset load threshold, the edge node and the cloud node are designated as the target execution node types for the first task; In response to the task priority being the second priority and the load status of the candidate execution node being less than a preset load threshold, the edge node is designated as the target execution node type for the first task; In response to the task priority being the second priority and the load status of the candidate execution node being greater than a preset load threshold, the cloud node is designated as the target execution node type for the first task.

3. The method according to claim 1, characterized in that, The step of determining the target execution node for the first task from multiple candidate execution nodes of the target execution node type based on a genetic algorithm includes: For each candidate execution node, perform the following processing: Determine the minimum task response latency of the first task on the candidate execution node; Determine the maximum computing power utilization of the candidate execution nodes; Determine the minimum cross-domain transmission cost of the first task at the candidate execution node; The fitness score of the candidate execution node is determined based on the minimum task response latency, the maximum computing power utilization, and the minimum cross-domain transmission cost. The candidate execution node with the highest fitness score is selected as the target execution node for the first task.

4. The method according to claim 3, characterized in that, Determining the minimum task response latency of the first task on the candidate execution node includes: Based on the waiting queue length of the candidate execution nodes and the estimated execution time of each waiting task, the queuing delay of the candidate execution nodes is determined; The performance metrics of the candidate execution nodes are mapped to obtain the task execution latency of the candidate execution nodes; Based on the data volume of the first task and the network parameters of the candidate execution node, the network transmission delay of the candidate node is determined; The queuing delay, the task execution delay, and the network transmission delay are summed to obtain the minimum task response delay of the first task at the candidate execution node.

5. The method according to claim 3, characterized in that, Determining the minimum cross-domain transmission cost of the first task on the candidate execution node includes: Based on the data transmission type of the first task and the location information of the candidate execution node, determine the network bandwidth cost of the first task at the candidate execution node; Based on the task priority of the first task and the security attributes of the candidate execution node, determine the data security cost of the first task at the candidate execution node; Based on the real-time status of the network link occupied by the task data transmission, the resource preemption cost of the first task at the candidate execution node is determined. The network bandwidth cost, the data security cost, and the resource preemption cost are summed to obtain the minimum cross-domain transmission cost of the first task at the candidate execution node.

6. The method according to claim 1, characterized in that, The method further includes: In response to the disconnection between the edge node and the cloud node, a second task within a preset priority range is determined; A preset proportion of computing power is allocated to the edge nodes so that the edge nodes can perform the second task; The business data generated by the second task is cached and stored in a hierarchical manner, and after the connection is restored, the business data is synchronized to the cloud node.

7. The method according to claim 1, characterized in that, The method further includes: The execution status data of the first task is collected based on a preset frequency; Based on the protocol type matching the first task, the execution status data is transmitted to the cloud policy center; The system receives a task scheduling policy generated by the cloud-based policy center based on the execution status data, and adjusts the execution of the first task based on the task scheduling policy.

8. The method according to claim 1, characterized in that, The step of executing the first task on the target execution node based on a protocol type matching the first task includes: In response to the first task having a task priority of third priority, the scheduling instruction for the first task is transmitted at the target execution node using the User Datagram Protocol. In response to the first task having a priority of fourth priority, the target execution node transmits the scheduling instructions and data synchronization information of the first task using the transmission control protocol.

9. A dynamic allocation and task scheduling system for edge-cloud computing power in park operation and maintenance, characterized in that, The system includes: The scoring module is used to determine the first score of the first task in each dimension indicator, and based on the first weight, to perform a weighted summation of the first scores of each dimension indicator to obtain the second score of the first task. The determination module is used to determine the task priority corresponding to the second score, and to determine the target execution node type of the first task based on the task priority and the load status of the candidate execution nodes; A screening module is used to determine the target execution node of the first task from multiple candidate execution nodes of the target execution node type based on a genetic algorithm; An execution module is configured to execute the first task on the target execution node based on a protocol type that matches the first task.

10. A device for dynamic allocation and task scheduling of edge-cloud computing power in park operation and maintenance, characterized in that, The device includes: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method as described in any one of claims 1-8.

Images