Method and system for dynamic scheduling of intelligent converged terminal resources for network access

Abstract

The application is suitable for power dispatching technology field of power distribution network, especially relates to network access intelligent fusion terminal resource dynamic scheduling method and system, the method includes determining resource demand information corresponding to each data processing task according to the power grid data processing task set, and generating task resource demand set; determining task execution order according to the task resource demand set and current resource state of intelligent fusion terminal, and generating task scheduling queue; according to the task scheduling queue, the internal computing resources of intelligent fusion terminal are allocated, and the corresponding data processing task is executed; the application carries out unified analysis on the resource demand corresponding to each data processing task, carries out clustering on the resource demand, determines the task execution order according to the correlation between different categories of resource demand, makes the data processing tasks with similar resource demand execute continuously, reduces the frequent switching of processor, storage resource and communication resource, and reduces resource reconstruction overhead.

Description

Technical Field

[0001] This invention relates to the field of power dispatching technology for power distribution networks, and in particular to a method and system for dynamic scheduling of intelligent converged terminal resources for network access. Background Technology

[0002] In modern smart distribution network systems, smart converged terminals, as core distribution equipment on the edge of the power grid, undertake key functions such as power grid operation data acquisition, power status monitoring, and distribution task processing. With the surge in the number of connected devices in the distribution network, the power data processing tasks and power grid status judgment tasks carried by the terminals have exploded. To meet this demand, traditional distribution terminals mostly adopt fixed task scheduling strategies. However, this fixed strategy cannot dynamically adjust the task execution order according to the real-time power grid load, the terminal's remaining computing power, communication bandwidth, and storage resources. When the power grid operation fluctuates or data tasks are concentrated and sudden, problems such as delays in high-priority power monitoring tasks, terminal resource congestion, and lag in power grid data processing can easily occur, resulting in untimely perception of the distribution network operation status and delayed power dispatch response, thus reducing the stability of the distribution network operation and the accuracy of distribution control.

[0003] Therefore, how to provide a dynamic scheduling scheme to improve the efficiency of dynamic resource scheduling and solve problems such as the lag in power grid data processing is the technical problem that the present invention aims to solve. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for dynamic scheduling of intelligent converged terminal resources for network access, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for dynamic scheduling of intelligent converged terminal resources for network access, the method comprising:

[0007] Collect power grid operation data uploaded by network access devices, and generate a power grid data processing task set based on the power grid operation data;

[0008] Based on the power grid data processing task set, the resource requirement information corresponding to each data processing task is determined, and a task resource requirement set is generated; wherein, there is a one-to-one correspondence between data processing tasks and task resource requirements;

[0009] The task execution order is determined based on the task resource requirement set and the current resource status of the intelligent fusion terminal, and a task scheduling queue is generated.

[0010] The intelligent fusion terminal allocates its internal computing resources according to the task scheduling queue, executes the corresponding data processing tasks, and generates a task execution result set.

[0011] Update the resource status of the intelligent fusion terminal based on the task execution result set, and recursively adjust the subsequent task scheduling queue.

[0012] As a further aspect of the present invention, the step of collecting power grid operation data uploaded by the network access device and generating a power grid data processing task set based on the power grid operation data includes:

[0013] Receive power grid operation data uploaded by network access devices and sort them according to the collection time;

[0014] The power grid operation data is matrixed to obtain a power grid data volume; the matrix dimension of the power grid data volume is a preset value.

[0015] The task receiving port is opened in real time to receive task information. Based on the task information, data elements are extracted from the power grid data body. The extracted data elements are compared to determine the data characteristics corresponding to the task information. The task information includes the task type and generation time.

[0016] Based on the statistical data characteristics of task information, a data feature table is constructed, and an independent task prediction process is built based on the data feature table.

[0017] A set of power grid data processing tasks is generated synchronously based on the task receiving port and the task prediction process.

[0018] As a further aspect of the present invention, the steps of receiving task information through a real-time open task receiving port, extracting data elements from the power grid data body based on the task information, comparing the extracted data elements, and determining the data characteristics corresponding to the task information include:

[0019] The task receiving port is opened in real time, allowing you to receive task types and generation times.

[0020] The extraction interval is determined based on the generation time, and sub-data volumes are extracted from the power grid data volume based on the extraction interval as data elements;

[0021] The extracted data elements are classified according to the task type. The element positions of the same type of data elements are selected sequentially, and all elements at the element positions are extracted to obtain an element array.

[0022] Calculate the mean and standard deviation of the array of elements, and mark the positions of elements whose standard deviation is less than a preset standard deviation threshold;

[0023] The positions of the marked elements and their corresponding mean values ​​are used to construct data features, which serve as data features for the task type.

[0024] As a further aspect of the present invention, the step of determining the resource requirement information corresponding to each data processing task based on the power grid data processing task set and generating a task resource requirement set includes:

[0025] Read each data processing task in the power grid data processing task set and query the priority of the task.

[0026] Determine the required computing resource type based on the data processing rules corresponding to the data processing task;

[0027] The statistics show the processor time, storage space, and communication bandwidth required to execute the corresponding data processing tasks.

[0028] Record the priority, computing resource type, processor time, storage space, and communication bandwidth for each data processing task;

[0029] Store all recorded information according to task number and generate a task resource requirement set.

[0030] As a further aspect of the present invention, the step of determining the task execution order based on the task resource requirement set and the current resource status of the intelligent fusion terminal, and generating a task scheduling queue, includes:

[0031] Read and compare the task resource requirements of each task in the task resource requirement set, cluster the task resource requirements, and simultaneously determine the mean requirement information of each category of task resource requirements.

[0032] By comparing the average resource requirements of different types of tasks, the correlation of resource requirements for each type of task can be obtained.

[0033] Read the resource capacity of the intelligent converged terminal, and construct a processing process based on the resource capacity and priority;

[0034] For any processing process, tasks are selected sequentially based on clustering results and relevance, the task execution order is determined, and a task scheduling queue is generated.

[0035] As a further aspect of the present invention, the step of updating the resource status of the intelligent fusion terminal based on the task execution result set and recursively adjusting the subsequent task scheduling queue includes:

[0036] For tasks that have already been executed, query the historical task execution results;

[0037] The anomaly degree of the current task execution result is obtained based on the historical task execution results; the anomaly degree is proportional to the data distance.

[0038] When the anomaly level reaches the preset anomaly threshold, query the resource requirements of a type of task corresponding to the task and adjust its priority.

[0039] Update the task scheduling queue based on the adjusted priority.

[0040] The present invention also provides a dynamic scheduling system for intelligent converged terminal resources for network access, the system comprising:

[0041] The task set generation module is used to collect power grid operation data uploaded by network access devices and generate a power grid data processing task set based on the power grid operation data.

[0042] The demand set generation module is used to determine the resource demand information corresponding to each data processing task based on the power grid data processing task set, and generate a task resource demand set; wherein, there is a one-to-one correspondence between data processing tasks and task resource demands.

[0043] The task queue generation module is used to determine the task execution order based on the task resource requirement set and the current resource status of the intelligent fusion terminal, and to generate a task scheduling queue.

[0044] The result set acquisition module is used to allocate the internal computing resources of the intelligent fusion terminal according to the task scheduling queue, execute the corresponding data processing tasks, and generate a task execution result set.

[0045] The recursive adjustment module is used to update the resource status of the intelligent fusion terminal based on the task execution result set and recursively adjust the subsequent task scheduling queue.

[0046] As a further aspect of the present invention, the task set generation module includes:

[0047] The data sorting unit is used to receive power grid operation data uploaded by network access devices and sort them according to the collection time.

[0048] A data volume generation unit is used to perform matrix processing on the power grid operation data to obtain a power grid data volume; the matrix dimension of the power grid data volume is a preset value.

[0049] The data element extraction unit is used to open the task receiving port in real time, receive task information, extract data elements from the power grid data body based on the task information, compare the extracted data elements, and determine the data characteristics corresponding to the task information; wherein, the task information includes task type and generation time.

[0050] The prediction process construction unit is used to build a data feature table based on the statistical characteristics of task information, and to build an independent task prediction process based on the data feature table.

[0051] The task set generation unit is used to synchronously generate power grid data processing task sets based on the task receiving port and the task prediction process.

[0052] As a further embodiment of the present invention, the requirement set generation module includes:

[0053] The priority query unit is used to read each data processing task in the power grid data processing task set and query the priority of the task.

[0054] The type determination unit is used to determine the required computing resource type based on the data processing rules corresponding to the data processing task.

[0055] The data statistics unit is used to calculate the processor time, storage space, and communication bandwidth required to execute the corresponding data processing task.

[0056] The data recording unit is used to record the priority, computing resource type, processor time occupied, storage space and communication bandwidth of each data processing task.

[0057] The data storage unit is used to store information of all records according to the task number and generate a task resource requirement set.

[0058] As a further embodiment of the present invention, the task queue generation module includes:

[0059] The demand clustering unit is used to read and compare the task resource requirements of each task in the task resource requirement set, cluster the task resource requirements, and simultaneously determine the mean demand information of each type of task resource requirements.

[0060] The relevance calculation unit is used to compare the mean resource requirements of different types of tasks to obtain the relevance of resource requirements for each type of task.

[0061] The processing process construction unit is used to read the resource capacity of the intelligent fusion terminal and construct the processing process based on the resource capacity and priority.

[0062] The task order determination unit is used to select tasks sequentially based on clustering results and relevance for any processing process, determine the task execution order, and generate a task scheduling queue.

[0063] Compared with the prior art, the beneficial effects of the present invention are:

[0064] This invention performs a unified analysis of the resource requirements corresponding to each data processing task and clusters these requirements. The execution order is determined based on the correlation between different categories of resource requirements, ensuring that data processing tasks with similar resource requirements are executed consecutively. This reduces frequent switching of processor, storage, and communication resources, lowering resource reconfiguration overhead. Simultaneously, the priority of corresponding tasks is dynamically updated based on the task execution results, and the subsequent task scheduling queue is recursively adjusted. This allows the task scheduling strategy to continuously optimize in real-time as the power grid's operating status changes, avoiding the resource utilization degradation caused by long-term use of fixed scheduling strategies. Ultimately, this improves the data processing efficiency, resource utilization, and adaptive scheduling capability of the intelligent fusion terminal to changes in the power grid's operating status. Attached Figure Description

[0065] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention.

[0066] Figure 1 A flowchart illustrating the method for dynamic scheduling of intelligent converged terminal resources for network access provided in this embodiment of the invention.

[0067] Figure 2 A block diagram of the intelligent converged terminal resource dynamic scheduling system for network access provided in an embodiment of the present invention. Detailed Implementation

[0068] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0069] Figure 1 The flowchart illustrates a method for dynamic scheduling of intelligent converged terminal resources for network access provided by an embodiment of the present invention. The method includes:

[0070] Step S100: Collect power grid operation data uploaded by network access devices, and generate a power grid data processing task set based on the power grid operation data;

[0071] The operation data of the power grid is obtained by network access devices. These devices have connection channels with various power grid data acquisition units, allowing them to acquire the power grid operation data uploaded by the network access devices. There are many types of network access devices, such as smart meters, concentrators, and data acquisition units, as well as measurement and control devices, protection devices, fault recorders, and electronic instrument transformers. Any device capable of acquiring data and sending it to other ports can be called a network access device. In terms of data structure, different devices have at least device tags, and the acquired data itself also has time tags. A matrix is ​​constructed based on the device tags, and then statistical data is collected based on the matrix to obtain a complete data volume. During the process of statistical data collection based on the matrix, the time difference should be small and fixed; that is, a time axis is preset, and data from different times of each device is continuously read to obtain a regular data volume, but duplicate data may exist. Alternatively, another method is to use device tags as indexes to statistically analyze different power grid operation data to obtain operation arrays. Statistical analysis of each operation array yields a dataset. This method eliminates duplicates, but the data is not particularly regular.

[0072] Identifying power grid operation data can determine what tasks need to be performed. This process is essentially a rule matching process. When the power grid operation data meets certain states, what tasks need to be performed? The correspondence between states and tasks is pre-set. In practical applications, matching tasks can be found by comparing the power grid operation data with the rules. Since there is a lot of power grid operation data, the matched tasks may not be unique. Therefore, what is obtained is a set of power grid data processing tasks.

[0073] It is worth mentioning that the device tag is not just a type tag, but a specific type of device at a specific location, specifically defined by its unique identifier. Therefore, the device tag includes information such as type, installation location, and the location corresponding to the collected data.

[0074] Step S200: Determine the resource requirement information corresponding to each data processing task based on the power grid data processing task set, and generate a task resource requirement set; wherein, there is a one-to-one correspondence between data processing tasks and task resource requirements;

[0075] After the set of power grid data processing tasks is determined, the resources required for each task are queried, that is, the resource requirement information corresponding to each data processing task. The resource requirements corresponding to each power grid data processing task are combined into a set to obtain the task resource requirement set; there is a one-to-one correspondence between data processing tasks and task resource requirements.

[0076] Step S300: Determine the task execution order based on the task resource requirement set and the current resource status of the intelligent fusion terminal, and generate a task scheduling queue;

[0077] Step S400: Allocate the internal computing resources of the intelligent fusion terminal according to the task scheduling queue, execute the corresponding data processing tasks, and generate a task execution result set;

[0078] The task resource requirement set represents the data processing resources required for each power grid data processing task. Combined with the current resource status of the intelligent fusion terminal, the task execution order can be determined to obtain the task scheduling queue. Based on the task scheduling queue, the internal computing resources of the intelligent fusion terminal are allocated to continuously execute the corresponding data processing tasks, obtain the task execution results in real time, and obtain the task execution result set.

[0079] Step S500: Update the resource status of the intelligent fusion terminal according to the task execution result set, and recursively adjust the subsequent task scheduling queue;

[0080] After obtaining the task execution result, the resource status of the intelligent fusion terminal is updated according to the task execution result. Correspondingly, the task execution order will also change, and the task scheduling queue will also be updated accordingly.

[0081] In practical applications, the technical solution of this invention can be widely applied to intelligent converged terminal resource scheduling scenarios under the wind power grid-connected distribution network architecture. Wind power scenarios are characterized by strong randomness, volatility, and intermittency. The daily operation of wind farms generates massive amounts of real-time wind measurement data, unit status data, grid-connected power quality data, and fault monitoring data. Terminals need to continuously execute multiple types of power data processing tasks. In large-scale wind power grid-connected scenarios, traditional fixed scheduling methods are prone to problems such as uneven distribution of computing resources, blockage of high-priority wind power monitoring tasks, and excessive data processing latency. This invention, by dynamically identifying the resource requirements of wind power business tasks, clustering and analyzing task characteristics, and adaptively adjusting task priorities and scheduling queues, can achieve dynamic optimization and allocation of computing power, bandwidth, and storage resources of wind power distribution network edge terminals. This effectively improves the real-time performance of wind farm grid-connected monitoring and the stability of grid operation, and is suitable for power scheduling scenarios of various types of wind power access to the distribution system, such as onshore wind power and offshore wind power.

[0082] Regarding step S100, the step of collecting power grid operation data uploaded by the network access device and generating a power grid data processing task set based on the power grid operation data includes:

[0083] Receive power grid operation data uploaded by network access devices and sort them according to the collection time;

[0084] The power grid operation data is matrixed to obtain a power grid data volume; the matrix dimension of the power grid data volume is a preset value.

[0085] The task receiving port is opened in real time to receive task information. Based on the task information, data elements are extracted from the power grid data body. The extracted data elements are compared to determine the data characteristics corresponding to the task information. The task information includes the task type and generation time.

[0086] Based on the statistical data characteristics of task information, a data feature table is constructed, and an independent task prediction process is built based on the data feature table.

[0087] A set of power grid data processing tasks is generated synchronously based on the task receiving port and the task prediction process.

[0088] In one example of the technical solution of this invention, the generation process of the power grid data processing task set is described. The power grid operation data uploaded by the network access device is received and sorted according to the acquisition time. The power grid operation data is matrix-processed to obtain the power grid data body. The power grid data body itself is a multi-dimensional matrix. For any power grid operation data, labels describing its attributes are obtained, such as coordinates in different directions, the corresponding line number, acquisition time, and data type. A multi-dimensional matrix is ​​constructed based on the labels describing its attributes. The power grid operation data is inserted into the corresponding position in the multi-dimensional matrix to obtain the power grid data body.

[0089] Based on this, the task receiving port is opened in real time to receive task information. This process is the preprocessing stage. The task information is determined by the staff based on the situation. For the technical solution of this invention, only the receiving action needs to be performed. The task information itself is data to be received. In practical applications, the task receiving port is also open in real time, allowing staff to actively set task information at any time. For the received task information, different dimensional intervals are determined in the power grid data body according to the task time, task type, and task location. These intervals include the equipment type interval, equipment location interval, and time interval. The time interval is generally calculated by counting backwards from the current time. The location interval is the location range related to the task execution process. The equipment type interval is generally the indicator that needs to be obtained in the task execution scenario. The data within the interval is called a data element. The data element is actually a part of the content in the power grid data body. Then, the extracted data elements are compared to determine the data characteristics corresponding to the task information. After obtaining the data characteristics, a data feature table is constructed based on the statistical data characteristics of the task information. An independent task prediction process is constructed based on the data feature table.

[0090] Specifically, the process of comparing the extracted data elements to determine the data features corresponding to the task information is the conventional feature extraction process. For example, for the same task, multiple extracted data elements are queried, and the intersection of the data elements is calculated. This intersection is the data feature. It is worth mentioning that when calculating the intersection, it is necessary to compare the data at the same matrix position. If their data difference is less than the preset data difference threshold, they can be considered as the same data. In addition, in the process of constructing the power grid data volume, the data can be normalized first, that is, dimensionless, to convert it into dimensionless values. This is a conventional numerical processing process and will not be elaborated here. In fact, there are many feature extraction schemes. For those skilled in the art, feature extraction is not complicated based on the data elements already extracted in the technical solution of this invention.

[0091] Furthermore, regarding the construction of an independent task prediction process based on a data feature table, the data feature table provides the data features corresponding to each task information. In practical applications, the power grid data volume can be traversed based on the data features (a higher-dimensional convolutional recognition process). If a data item is matched, the corresponding task information is queried. This is the prediction process, which requires no human intervention and has an extremely fast response speed. Different task information can be located in a large amount of power grid data volume. This is the prediction process.

[0092] Finally, a set of power grid data processing tasks is generated synchronously based on the task receiving port and the task prediction process. This process means that the task receiving port and the task prediction process are carried out synchronously. The task receiving port means that the staff can issue tasks in real time according to the situation. The task prediction process is the main body of this method and can quickly identify global data and identify tasks that may need to be executed. Of course, the tasks received by the task receiving port can be used to update the data feature table (the more tasks, the more complete the data feature table), and the higher the accuracy of the prediction results. Then, the results of the task prediction process can be used as a reference for the staff. At this time, the task receiving port can have a built-in confirmation process. The output of the task prediction process is sent to the task receiving port and synchronously displayed to the staff. When the staff's confirmation instruction is received, the execution is issued.

[0093] As a preferred embodiment of the technical solution of the present invention, the steps of receiving task information through a real-time open task receiving port, extracting data elements from the power grid data body based on the task information, and comparing the extracted data elements to determine the data characteristics corresponding to the task information include:

[0094] The task receiving port is opened in real time, allowing you to receive task types and generation times.

[0095] The extraction interval is determined based on the generation time, and sub-data volumes are extracted from the power grid data volume based on the extraction interval as data elements;

[0096] The extracted data elements are classified according to the task type. The element positions of the same type of data elements are selected sequentially, and all elements at the element positions are extracted to obtain an element array.

[0097] Calculate the mean and standard deviation of the array of elements, and mark the positions of elements whose standard deviation is less than a preset standard deviation threshold;

[0098] The positions of the marked elements and their corresponding mean values ​​are used to construct data features, which serve as data features for the task type.

[0099] The data feature generation process is based on the feature extraction process of a multidimensional matrix. The task receiving port is opened in real time to receive the task type and generation time. The interception interval is determined based on the generation time. For example, taking the current time as the right endpoint, the region within a time period is obtained backward. Based on the interception interval, sub-data volumes are extracted from the power grid data volume as data elements. This process does not consider other factors such as location or type. In fact, it is a simplified solution. If conditions permit, the task receiving port can be opened in real time to receive the task type, generation time, application location and data source type. The interception interval is determined based on the generation time, application location and data source type. Sub-data volumes are extracted from the power grid data volume based on the interception interval as data elements.

[0100] After extracting the data elements, they are categorized based on task type. Within each category, element positions are selected sequentially, and all elements at those positions are extracted to obtain an element array. This process is equivalent to discretization, converting multiple data elements into a set of arrays for each element position. The mean and standard deviation of the element arrays are calculated, and element positions with a standard deviation less than a preset threshold are marked. A larger standard deviation indicates greater variation in the values ​​of different data elements at that position, while a smaller standard deviation indicates less variation. Less variation suggests that for the same task, the data at that position exhibits similar values, which can be used as feature points. Finally, the marked element positions and their corresponding means are statistically analyzed to construct data features, which serve as data features for the task type.

[0101] Regarding step S200, the step of determining the resource requirement information corresponding to each data processing task based on the power grid data processing task set and generating a task resource requirement set includes:

[0102] Read each data processing task in the power grid data processing task set and query the priority of the task.

[0103] Determine the required computing resource type based on the data processing rules corresponding to the data processing task;

[0104] The statistics show the processor time, storage space, and communication bandwidth required to execute the corresponding data processing tasks.

[0105] Record the priority, computing resource type, processor time, storage space, and communication bandwidth for each data processing task;

[0106] Store all recorded information according to task number and generate a task resource requirement set.

[0107] In one example of the technical solution of this invention, the process of generating a task resource requirement set is described. Each data processing task in the power grid data processing task set is read, and the priority of the task is queried. In a power grid management architecture, the types of data processing tasks are limited, and each task has a different priority. In fact, when task information is received, the task receiving port can also receive the priority to assess the importance of the task. The required computing resource type is determined according to the data processing rules corresponding to the data processing task, and then the corresponding demand is queried. The required computing resource types include processor time, storage space, and communication bandwidth, etc. Then, each data is statistically analyzed according to the unique label of the data processing task, and the statistical results are stored to obtain the task resource requirement set. The above process is actually a data query and statistical process. Priorities, data processing rules, and computing resource types are all known data, and can be directly read according to the corresponding relationships.

[0108] In the above content, regarding the priority of the tasks, for the technical solution of this invention, it is assumed to be known data. As for the determination process, a feasible solution is to classify the power grid data processing tasks hierarchically according to the power business specifications, dividing the tasks into four business levels: fault monitoring, load statistics, status inspection, and ledger statistics, each with a preset basic priority score. At the same time, the score is dynamically adjusted based on the real-time operating conditions of the distribution network. Specifically, dynamic weighting coefficients are generated based on the current power grid load fluctuation amplitude, the number of equipment operation anomalies, and the frequency of real-time data collection. The basic priority score and the dynamic weighting coefficient are weighted and integrated to obtain the real-time priority value of each task, and divided into high, medium, and low priorities according to the value. Among them, the fault emergency power monitoring task has the highest priority.

[0109] Similarly, the required computing resource types are obtained by matching based on a pre-set power data processing rule base. This rule base is pre-built and fixed based on the computational characteristics of various power services of the distribution network terminal. First, for the business attributes of different power grid data processing tasks, the corresponding standardized data processing rules are retrieved. The rules clearly record the exclusive resource requirement types corresponding to various power tasks, mainly including four core types: intensive computing power resources, high-frequency communication bandwidth resources, large-capacity storage resources, and real-time computing cache resources. Then, the data computation characteristics, data transmission volume, and storage retention requirements of the current task are analyzed and compared with the adaptation thresholds of each resource type in the rule base. If the task is a high-intensity computational business such as instantaneous analysis of power grid faults or waveform calculations, then intensive computing power resources are matched; if it is a real-time data upload of the power grid or equipment status interaction business, then high-frequency communication bandwidth resources are matched. Through the precise mapping of the rule base, the automated and accurate determination of computing resource types is achieved, adapting to the differentiated task processing needs of the distribution terminal.

[0110] Regarding step S300, the step of determining the task execution order based on the task resource requirement set and the current resource status of the intelligent fusion terminal, and generating a task scheduling queue, includes:

[0111] Read and compare the task resource requirements of each task in the task resource requirement set, cluster the task resource requirements, and simultaneously determine the mean requirement information of each category of task resource requirements.

[0112] By comparing the average resource requirements of different types of tasks, the correlation of resource requirements for each type of task can be obtained.

[0113] Read the resource capacity of the intelligent converged terminal, and construct a processing process based on the resource capacity and priority;

[0114] For any processing process, tasks are selected sequentially based on clustering results and relevance, the task execution order is determined, and a task scheduling queue is generated.

[0115] In one example of the technical solution of this invention, the process of generating the task scheduling queue is described. The task resource requirements of each task in the task resource requirement set are read and compared, the task resource requirements are clustered, and the average requirement information of each type of task resource requirement is determined synchronously. This process is used to group similar tasks together. When processing, similar tasks are processed in order, which can reduce the preprocessing time. If different tasks are executed, there are more preprocessing processes. For example, different types of tasks may involve different libraries. When executing a new task, it is necessary to establish a connection channel with the new library, which requires more time. If they are similar tasks, at least there is an overlap between the libraries they use, which can save some time.

[0116] Then, the average resource requirements of different task categories are compared to obtain the relevance of resource requirements for each category. This is the difference between categories, which is used to determine the order. If there is no priority, a task category is randomly selected and executed. After that, the next task category with the highest relevance is selected and processed in this order. The preprocessing process will be very smooth and efficient.

[0117] However, task priority is a significant issue. The resource capacity of the intelligent fusion terminal is read, and processing processes are constructed based on this capacity and priority. Resource capacity represents the terminal's capabilities. The average resource requirement for each task is queried; this average requirement and resource capacity are parameters with the same data structure, allowing for comparison and generally representing the required processing resources. Within the resource capacity limit, several task types are randomly selected. Due to the large number of task types, the selection condition only needs to be within the resource capacity, resulting in numerous random selection methods (a combination problem). The random selection process is iteratively executed to obtain a preset number of random selection schemes. Ideally, for each scheme, the task types should have low correlation. In subsequent processes, the number of processes should be independently set for each selected task type. The main objective of this method is to ensure maximum independence between processes. For example, assuming A, B, ... If there are four types of tasks, C and D, and only A and B have high correlation between them, then after A is processed, B is likely to be processed soon, and B will also be processed quickly. However, C and D will not be executed quickly. By setting B, C, and D as independent processes, C and D, or similar tasks, can be executed faster. In a random state, although the overall efficiency is similar, tasks with high correlation to A may be executed continuously, leading to an unbalanced processing. The way to make the processes as independent as possible is to optimize the random selection scheme. For any random selection scheme, query the correlation between each type of task and find the maximum correlation value. The smaller the maximum value, the smaller the correlation between each pair of tasks (the maximum values ​​are all very small). Therefore, the maximum value can be used as the optimization index, and the random selection scheme with the smallest maximum value is selected as the final scheme.

[0118] In summary, after the above processing, the intelligent fusion terminal selects several categories of almost unrelated tasks, randomly selects one task from each category as the initial task, and obtains different processing processes. For any processing process, tasks in the same category are read sequentially. After the tasks in the same category are read, the next category of tasks is read based on relevance, and tasks in the next category are read sequentially. This is how tasks are selected sequentially based on clustering results and relevance to determine the task execution order and generate a task scheduling queue.

[0119] It is worth mentioning that the application of priority is relatively independent. For example, when selecting tasks, a random method can be used instead of selecting them according to priority order. If there are particularly high-priority tasks, one or more independent processes can be set up to execute high-priority tasks. It is even possible to introduce an interrupt operation to interrupt all processes for high-priority tasks and execute the high-priority tasks. The task scheduling queue generated by the above content is actually just the execution process within the normal priority limit.

[0120] Furthermore, regarding the relevance of resource requirements for different categories of tasks in the technical solution of this invention, the relevance is quantified using the Euclidean distance algorithm based on the average requirement information of various tasks, thereby achieving accurate determination of the similarity of power task resources. First, the average values ​​of the three core resource requirements—processor usage time, storage space, and communication bandwidth—for each type of task after clustering are normalized to eliminate the dimensional differences of different resource parameters, constructing standardized average resource feature vectors for each type of task. Then, the Euclidean distance between feature vectors of different categories is calculated pairwise. The smaller the distance value, the more similar the resource requirement structures of the two types of tasks are and the higher the relevance; the larger the distance value, the greater the difference in resource requirements and the lower the relevance. Finally, the calculated Euclidean distance is reverse-mapped and normalized to the relevance range of 0-1 to generate the final task resource requirement relevance value. Based on this quantified relevance, highly relevant power tasks can be prioritized and included in the same processing process, reducing terminal resource switching losses and improving the overall operational efficiency of distribution network terminal task scheduling.

[0121] Regarding step S400, the step of updating the resource status of the intelligent fusion terminal according to the task execution result set and recursively adjusting the subsequent task scheduling queue includes:

[0122] For tasks that have already been executed, query the historical task execution results;

[0123] The anomaly degree of the current task execution result is obtained based on the historical task execution results; the anomaly degree is proportional to the data distance.

[0124] When the anomaly level reaches the preset anomaly threshold, query the resource requirements of a type of task corresponding to the task and adjust its priority.

[0125] Update the task scheduling queue based on the adjusted priority.

[0126] In one example of the technical solution of this invention, the recursive adjustment process is described. For an executed task, the historical task execution results are queried, and the anomaly degree of the current task execution result is obtained based on the historical task execution results. The anomaly degree is proportional to the data distance. This process is to compare the current task execution result with the historical task execution results. This is equivalent to comparing the real data with the standard data in the historical data. The greater the difference (data distance), the higher the anomaly degree. When the anomaly degree reaches a preset anomaly degree threshold (high enough), the resource requirements of a type of task corresponding to the task are queried, and its priority is adjusted so that it needs to be processed earlier.

[0127] The logic behind the above process is that if there is a significant difference between the processing result of a task and the historical processing result, it can be considered that an emergency has occurred. When executing the task in the future, it should be done as early as possible. Of course, a warning message will also be generated at the same time to remind the staff to take appropriate management measures.

[0128] It should be noted that the process of obtaining the aforementioned anomaly degree generally involves calculating the data distance between the current task's resource consumption characteristics and historical standard task characteristics to achieve a quantitative assessment of power task operation anomalies. First, it retrieves historically executed power data processing tasks of the same type and under the same operating conditions, extracts their standard resource consumption parameters, and constructs a historical standard feature dataset covering core indicators such as standard computing power consumption, bandwidth usage time, and task execution time. Then, it extracts the actual resource consumption data after the current task's execution to form the current task's feature vector. Using the Manhattan data distance algorithm, it calculates the deviation distance between the current task's feature vector and the center vector of the historical standard feature dataset; this deviation distance is the task execution anomaly degree. The greater the deviation between the actual operating parameters of the task and the historical standard parameters, the larger the data distance value, and the higher the task execution anomaly degree. This quantitative method can accurately characterize the degree of deviation in the operation of distribution terminal tasks, providing precise data support for subsequent task priority adjustment and scheduling queue optimization, adapting to the dynamic scheduling needs of the power grid.

[0129] Furthermore, regarding the preset anomaly threshold mentioned above, the preset anomaly threshold in this invention is a graded dynamic threshold adapted to different operating scenarios of the distribution network. One feasible division method is three gradients: minor anomaly threshold, moderate anomaly threshold, and severe anomaly threshold, thoroughly clarifying the threshold judgment criteria. The system adaptively matches the corresponding gradient threshold according to three operating conditions: daily stable operation, peak load operation, and equipment fault early warning. Among them, a low threshold is used under stable operating conditions to accurately capture minor task operation deviations; medium and high thresholds are used under peak load and fault operating conditions to avoid ineffective scheduling and adjustment caused by high-frequency fluctuations. When the calculated anomaly value of the current task is greater than the preset gradient threshold under the corresponding operating condition, it is determined that the current task execution is abnormal, triggering the priority adjustment mechanism. At the same time, the threshold parameters can be iteratively optimized based on historical distribution operation data to achieve adaptive updating of the threshold. This refined judgment logic solves the problems of vague threshold description and unclear judgment criteria in the original system, making the terminal scheduling anomaly judgment more in line with the actual operation and control needs of the distribution network.

[0130] Figure 2 This is a block diagram of a network access intelligent converged terminal resource dynamic scheduling system provided in an embodiment of the present invention. A network access intelligent converged terminal resource dynamic scheduling system 10 is provided, comprising:

[0131] The task set generation module 11 is used to collect power grid operation data uploaded by network access devices and generate a power grid data processing task set based on the power grid operation data.

[0132] The demand set generation module 12 is used to determine the resource demand information corresponding to each data processing task based on the power grid data processing task set, and generate a task resource demand set; wherein, there is a one-to-one correspondence between the data processing task and the task resource demand.

[0133] The task queue generation module 13 is used to determine the task execution order based on the task resource requirement set and the current resource status of the intelligent fusion terminal, and to generate a task scheduling queue.

[0134] The result set acquisition module 14 is used to allocate the internal computing resources of the intelligent fusion terminal according to the task scheduling queue, execute the corresponding data processing tasks, and generate a task execution result set.

[0135] The recursive adjustment module 15 is used to update the resource status of the intelligent fusion terminal according to the task execution result set and recursively adjust the subsequent task scheduling queue.

[0136] Furthermore, the task set generation module 11 includes:

[0137] The data sorting unit is used to receive power grid operation data uploaded by network access devices and sort them according to the collection time.

[0138] A data volume generation unit is used to perform matrix processing on the power grid operation data to obtain a power grid data volume; the matrix dimension of the power grid data volume is a preset value.

[0139] The data element extraction unit is used to open the task receiving port in real time, receive task information, extract data elements from the power grid data body based on the task information, compare the extracted data elements, and determine the data characteristics corresponding to the task information; wherein, the task information includes task type and generation time.

[0140] The prediction process construction unit is used to build a data feature table based on the statistical characteristics of task information, and to build an independent task prediction process based on the data feature table.

[0141] The task set generation unit is used to synchronously generate power grid data processing task sets based on the task receiving port and the task prediction process.

[0142] Specifically, the requirement set generation module 12 includes:

[0143] The priority query unit is used to read each data processing task in the power grid data processing task set and query the priority of the task.

[0144] The type determination unit is used to determine the required computing resource type based on the data processing rules corresponding to the data processing task.

[0145] The data statistics unit is used to calculate the processor time, storage space, and communication bandwidth required to execute the corresponding data processing task.

[0146] The data recording unit is used to record the priority, computing resource type, processor time occupied, storage space and communication bandwidth of each data processing task.

[0147] The data storage unit is used to store information of all records according to the task number and generate a task resource requirement set.

[0148] Furthermore, the task queue generation module 13 includes:

[0149] The demand clustering unit is used to read and compare the task resource requirements of each task in the task resource requirement set, cluster the task resource requirements, and simultaneously determine the mean demand information of each type of task resource requirements.

[0150] The relevance calculation unit is used to compare the mean resource requirements of different types of tasks to obtain the relevance of resource requirements for each type of task.

[0151] The processing process construction unit is used to read the resource capacity of the intelligent fusion terminal and construct the processing process based on the resource capacity and priority.

[0152] The task order determination unit is used to select tasks sequentially based on clustering results and relevance for any processing process, determine the task execution order, and generate a task scheduling queue.

[0153] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for dynamic scheduling of intelligent converged terminal resources for network access, characterized in that, The method includes: Collect power grid operation data uploaded by network access devices, and generate a power grid data processing task set based on the power grid operation data; specifically including: Receive power grid operation data uploaded by network access devices and sort them according to the collection time; The power grid operation data is matrixed to obtain a power grid data volume; the matrix dimension of the power grid data volume is a preset value. The task receiving port is opened in real time to receive task information. Based on the task information, data elements are extracted from the power grid data body. The extracted data elements are compared to determine the data characteristics corresponding to the task information. The task information includes the task type and generation time. Based on the statistical data characteristics of task information, a data feature table is constructed, and an independent task prediction process is built based on the data feature table. A set of power grid data processing tasks is generated synchronously based on the task receiving port and the task prediction process. Based on the power grid data processing task set, the resource requirement information corresponding to each data processing task is determined, and a task resource requirement set is generated; wherein, there is a one-to-one correspondence between data processing tasks and task resource requirements; The task execution order is determined based on the task resource requirement set and the current resource status of the intelligent fusion terminal, and a task scheduling queue is generated; specifically including: Read and compare the task resource requirements of each task in the task resource requirement set, cluster the task resource requirements, and simultaneously determine the mean requirement information of each category of task resource requirements. By comparing the average resource requirements of different types of tasks, the correlation of resource requirements for each type of task can be obtained. Read the resource capacity of the intelligent converged terminal, and construct a processing process based on the resource capacity and priority; For any processing process, tasks are selected sequentially based on clustering results and relevance, the task execution order is determined, and a task scheduling queue is generated. The intelligent fusion terminal allocates its internal computing resources according to the task scheduling queue, executes the corresponding data processing tasks, and generates a task execution result set. Update the resource status of the intelligent fusion terminal based on the task execution result set, and recursively adjust the subsequent task scheduling queue.

2. The method for dynamic scheduling of intelligent converged terminal resources for network access according to claim 1, characterized in that, The steps of receiving task information through the real-time open task receiving port, extracting data elements from the power grid data body based on the task information, and comparing the extracted data elements to determine the data characteristics corresponding to the task information include: The task receiving port is opened in real time, allowing you to receive task types and generation times. The extraction interval is determined based on the generation time, and sub-data volumes are extracted from the power grid data volume based on the extraction interval as data elements; The extracted data elements are classified according to the task type. The element positions of the same type of data elements are selected sequentially, and all elements at the element positions are extracted to obtain an element array. Calculate the mean and standard deviation of the array of elements, and mark the positions of elements whose standard deviation is less than a preset standard deviation threshold; The positions of the marked elements and their corresponding mean values ​​are used to construct data features, which serve as data features for the task type.

3. The method for dynamic scheduling of intelligent converged terminal resources for network access according to claim 1, characterized in that, The step of determining the resource requirement information corresponding to each data processing task based on the power grid data processing task set and generating a task resource requirement set includes: Read each data processing task in the power grid data processing task set and query the priority of the task. Determine the required computing resource type based on the data processing rules corresponding to the data processing task; The statistics show the processor time, storage space, and communication bandwidth required to execute the corresponding data processing tasks. Record the priority, computing resource type, processor time, storage space, and communication bandwidth for each data processing task; Store all recorded information according to task number and generate a task resource requirement set.

4. The method for dynamic scheduling of intelligent converged terminal resources for network access according to claim 1, characterized in that, The step of updating the resource status of the intelligent fusion terminal based on the task execution result set and recursively adjusting the subsequent task scheduling queue includes: For tasks that have already been executed, query the historical task execution results; The anomaly degree of the current task execution result is obtained based on the historical task execution results; the anomaly degree is proportional to the data distance. When the anomaly level reaches the preset anomaly threshold, query the resource requirements of a type of task corresponding to the task and adjust its priority. Update the task scheduling queue based on the adjusted priority.

5. A dynamic scheduling system for intelligent converged terminal resources in network access, characterized in that, The system includes: The task set generation module is used to collect power grid operation data uploaded by network access devices and generate a power grid data processing task set based on the power grid operation data; specifically, it includes: The data sorting unit is used to receive power grid operation data uploaded by network access devices and sort them according to the collection time. A data volume generation unit is used to perform matrix processing on the power grid operation data to obtain a power grid data volume; the matrix dimension of the power grid data volume is a preset value. The data element extraction unit is used to open the task receiving port in real time, receive task information, extract data elements from the power grid data body based on the task information, compare the extracted data elements, and determine the data characteristics corresponding to the task information; wherein, the task information includes task type and generation time. The prediction process construction unit is used to build a data feature table based on the statistical characteristics of task information, and to build an independent task prediction process based on the data feature table. The task set generation unit is used to synchronously generate a power grid data processing task set based on the task receiving port and the task prediction process. The demand set generation module is used to determine the resource demand information corresponding to each data processing task based on the power grid data processing task set, and generate a task resource demand set; wherein, there is a one-to-one correspondence between data processing tasks and task resource demands. The task queue generation module is used to determine the task execution order based on the task resource requirement set and the current resource status of the intelligent fusion terminal, and to generate a task scheduling queue; specifically, it includes: Read and compare the task resource requirements of each task in the task resource requirement set, cluster the task resource requirements, and simultaneously determine the mean requirement information of each category of task resource requirements. By comparing the average resource requirements of different types of tasks, the correlation of resource requirements for each type of task can be obtained. Read the resource capacity of the intelligent converged terminal, and construct a processing process based on the resource capacity and priority; For any processing process, tasks are selected sequentially based on clustering results and relevance, the task execution order is determined, and a task scheduling queue is generated. The result set acquisition module is used to allocate the internal computing resources of the intelligent fusion terminal according to the task scheduling queue, execute the corresponding data processing tasks, and generate a task execution result set. The recursive adjustment module is used to update the resource status of the intelligent fusion terminal based on the task execution result set and recursively adjust the subsequent task scheduling queue.

6. The intelligent converged terminal resource dynamic scheduling system for network access according to claim 5, characterized in that, The requirement set generation module includes: The priority query unit is used to read each data processing task in the power grid data processing task set and query the priority of the task. The type determination unit is used to determine the required computing resource type based on the data processing rules corresponding to the data processing task. The data statistics unit is used to calculate the processor time, storage space, and communication bandwidth required to execute the corresponding data processing task. The data recording unit is used to record the priority, computing resource type, processor time occupied, storage space and communication bandwidth of each data processing task. The data storage unit is used to store information of all records according to the task number and generate a task resource requirement set.

7. The intelligent converged terminal resource dynamic scheduling system for network access according to claim 5, characterized in that, The task queue generation module includes: The demand clustering unit is used to read and compare the task resource requirements of each task in the task resource requirement set, cluster the task resource requirements, and simultaneously determine the mean demand information of each type of task resource requirements. The relevance calculation unit is used to compare the mean resource requirements of different types of tasks to obtain the relevance of resource requirements for each type of task. The processing process construction unit is used to read the resource capacity of the intelligent fusion terminal and construct the processing process based on the resource capacity and priority. The task order determination unit is used to select tasks sequentially based on clustering results and relevance for any processing process, determine the task execution order, and generate a task scheduling queue.

Images