Intelligent matching and calling method for multi-type computing power hardware of power grid

By building mapping correlation of computing power equipment in the power grid and evaluating equipment fitness, the problem of low matching efficiency of computing power resources in the power grid is solved, and the reliability and safety of power grid operation are improved.

CN118916172APending Publication Date: 2024-11-08STATE GRID JIANGSU ELECTRIC POWER CO LTD NANTONG POWER SUPPLY BRANCH +3
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411091909.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-09
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The existing power grid computing power hardware matching calling methods cannot quickly and accurately match computing power resources, resulting in low efficiency in power grid operation tasks, affecting the reliability and security of the system.

Method used

By connecting to the power grid operation management system, obtaining information of multiple computing power equipment, building mapping associations of computing power equipment, judging the remaining computing power space, performing delay prediction and equipment trustworthiness evaluation, calculating fitness, and finding the optimal computing power equipment to perform tasks .

Benefits of technology

It achieves timely and accurate computing resource matching, improves the processing efficiency of grid operation tasks, and enhances the reliability and safety of grid.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118916172A_ABST
    Figure CN118916172A_ABST
Patent Text Reader

Abstract

The invention provides an intelligent matching and calling method for multi-type computing power hardware of a power grid, and relates to the technical field of intelligent power grids, and the method comprises the steps: judging whether the residual computing power space of initial task processing equipment meets a first computing power resource required by a first processing task or not; if not, obtaining first associated computing power equipment distribution based on computing power equipment mapping association matching; carrying out delay prediction and equipment credibility evaluation on a plurality of first computing power equipment in the first associated computing power equipment distribution; and performing fitness calculation based on the plurality of first delay predicted values, the plurality of first device credibility and the plurality of first device occupancy computing power predictions to obtain a first optimal computing power device to execute the first processing task. According to the method, the problems that the power grid operation task cannot be timely processed and the task processing efficiency is relatively low due to the fact that computing power resource matching cannot be rapidly and accurately performed can be solved, the proper computing power resource can be timely and accurately found for task processing, and the overall processing efficiency of the task is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the field of smart grid technology, and more specifically, to a method for intelligent matching and calling of multiple types of computing power hardware for a power grid. Background Art

[0002] The matching of computing resources for power grid operation refers to the reasonable allocation and dispatching of computing resources according to the needs of power grid operation to ensure the safe, stable and economical operation of the power grid. With the continuous expansion of the scale of power grids and the improvement of their intelligence level, the amount of data that needs to be processed during the operation of power grids is increasing, and the demand for computing resources is also increasing. Therefore, how to accurately match the computing resources for power grid operation in a timely manner according to real-time processing tasks is very critical to ensure the safe and stable operation of the power grid system.

[0003] The shortcomings of the existing power grid computing power hardware matching and calling method are: due to the inability to quickly and accurately match computing power resources, the power grid operation tasks cannot be processed in a timely manner, resulting in low task processing efficiency. Summary of the invention

[0004] Therefore, in order to solve the above technical problems, the technical solutions adopted by the embodiments of the present disclosure are as follows:

[0005] A method for intelligent matching and calling of multi-type computing power hardware of a power grid includes the following steps: connecting to a power grid operation and management system of a target area, obtaining information of multiple computing power devices, wherein the computing power device information has a positioning identifier; based on the positioning identifier, establishing a computing power device mapping association of the target area according to the multiple computing power device information; receiving a first processing task, and judging whether the remaining computing power space of an initial task processing device satisfies the first computing power resources required by the first processing task; when not satisfied, obtaining a first associated computing power device distribution based on the initial task processing device and the computing power device mapping association matching; performing delay prediction on multiple first computing power devices in the first associated computing power device distribution, respectively, obtaining multiple first delay prediction values, performing device credibility evaluation on multiple first computing power devices in the first associated computing power device distribution, and generating multiple first device credibility; performing fitness calculation on the multiple first computing power devices based on the multiple first delay prediction values, the multiple first device credibility and the multiple first device occupied computing power predictions, and obtaining a first optimal computing power device according to the fitness calculation result, wherein the first optimal computing power device is the first computing power device with the largest fitness value; and executing the first processing task based on the first optimal computing power device.

[0006] A system for intelligently matching and calling multiple types of computing power hardware for a power grid, comprising: a computing power equipment information acquisition module, the computing power equipment information acquisition module is used to connect to the power grid operation and management system of a target area to obtain multiple computing power equipment information, wherein the computing power equipment information has a positioning identifier; a computing power equipment mapping association building module, the computing power equipment mapping association building module is used to build a computing power equipment mapping association in the target area based on the positioning identifier and the multiple computing power equipment information; a remaining computing power space judgment module, the remaining computing power space judgment module is used to receive a first processing task and judge whether the remaining computing power space of an initial task processing device satisfies the first computing power resource required by the first processing task; a first associated computing power equipment distribution acquisition module, the first associated computing power equipment distribution acquisition module is used to obtain a first associated computing power resource based on the initial task processing device and the computing power equipment mapping association matching when the first associated computing power resource is not satisfied. A first computing power device distribution; a first computing power device analysis module, the first computing power device analysis module is used to perform delay prediction on multiple first computing power devices in the first associated computing power device distribution, obtain multiple first delay prediction values, perform device trustworthiness assessment on multiple first computing power devices in the first associated computing power device distribution, and generate multiple first device trustworthiness; a first optimal computing power device acquisition module, the first optimal computing power device acquisition module is used to perform fitness calculation on the multiple first computing power devices based on the multiple first delay prediction values, the multiple first device trustworthiness and the multiple first device occupied computing power predictions, and obtain the first optimal computing power device according to the fitness calculation result, the first optimal computing power device is the first computing power device with the largest fitness value; a first processing task execution module, the first processing task execution module is used to execute the first processing task based on the first optimal computing power device.

[0007] Due to the adoption of the above technical method, the present disclosure has achieved the following technical advances compared with the prior art:

[0008] The invention can solve the technical problem that the existing power grid computing power hardware matching calling method cannot quickly and accurately match computing power resources, resulting in the inability to timely process power grid operation tasks, resulting in low task processing efficiency. By connecting to the power grid operation management system of the target area, multiple computing power equipment information is obtained, wherein the computing power equipment information has a positioning identifier; based on the positioning identifier, a computing power equipment mapping association of the target area is established according to the multiple computing power equipment information; a first processing task is received, and it is determined whether the remaining computing power space of the initial task processing device meets the first computing power resources required by the first processing task; when not satisfied, based on the initial task processing device and the computing power equipment mapping The first associated computing power device distribution is obtained by association matching; delay prediction is performed on multiple first computing power devices in the first associated computing power device distribution to obtain multiple first delay prediction values; device credibility assessment is performed on multiple first computing power devices in the first associated computing power device distribution to generate multiple first device credibility; fitness calculation is performed on the multiple first computing power devices based on the multiple first delay prediction values, the multiple first device credibility and the multiple first device occupied computing power predictions, and the first optimal computing power device is obtained according to the fitness calculation result, and the first optimal computing power device is the first computing power device with the largest fitness value; the first processing task is executed based on the first optimal computing power device. It is possible to find suitable computing power resources for task processing in a timely and accurate manner, thereby improving the overall processing efficiency of power grid operation tasks, thereby further improving the reliability and safety of power grid operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings required for describing the embodiments are briefly introduced below.

[0010] Figure 1 A flowchart of a method for intelligently matching and calling multiple types of computing power hardware in a power grid is provided for this application;

[0011] Figure 2 A schematic diagram of a process for obtaining the distribution of first associated computing power equipment in a method for intelligent matching and calling of multi-type computing power hardware in a power grid is provided for this application;

[0012] Figure 3 A structural diagram of a system for intelligently matching and calling multiple types of computing power hardware in a power grid is provided for this application.

[0013] Explanation of the accompanying drawings: computing power equipment information acquisition module 01, computing power equipment mapping association building module 02, remaining computing power space judgment module 03, first associated computing power equipment distribution acquisition module 04, first computing power equipment analysis module 05, first optimal computing power equipment acquisition module 06, first processing task execution module 07. DETAILED DESCRIPTION

[0014] The following will be combined with the drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0015] Based on the above description, if Figure 1 As shown, the present disclosure provides a method for intelligent matching and calling of multi-type computing power hardware for a power grid, comprising:

[0016] The method provided in this application is used to optimize the matching and calling method of multiple types of computing power hardware in the power grid, so as to achieve the purpose of timely and accurately finding suitable computing power resources for task processing, thereby improving the overall processing efficiency of power grid operation tasks, thereby further improving the reliability and safety of power grid operation. The method is specifically implemented in a smart matching and calling system for multiple types of computing power hardware in the power grid.

[0017] Connecting to the power grid operation management system of the target area to obtain information of multiple computing devices, wherein the computing device information has a positioning identifier;

[0018] In an embodiment of the present application, first, the power grid operation and management system of the target area is connected, wherein the power grid operation and management system refers to a system for monitoring and managing the operation of the power grid, mainly including functions such as power grid data collection, operation data monitoring, and computing power resource scheduling; the target area can be set according to actual conditions by technical personnel in this field. According to the power grid operation and management system, multiple computing power equipment information is obtained, and the computing power equipment information has a positioning identifier, wherein the computing power equipment information includes the computing power equipment type, computing power resource capacity, data processing security level, equipment connection relationship, etc., wherein the computing power equipment type includes cloud computing equipment, servers, edge computing equipment, etc. The equipment connection relationship can be set according to actual conditions, for example: cloud computing equipment is connected to a primary server, a primary server is connected to a secondary server, etc.; the positioning identifier refers to the specific location coordinates of the computing power equipment in the target area. By obtaining multiple computing power equipment information, data support is provided for the next step of building a computing power equipment mapping association.

[0019] Based on the positioning identifier, establishing a computing power device mapping association of the target area according to the information of the multiple computing power devices;

[0020] In the embodiment of the present application, based on the positioning identifier, the device connection is performed according to the device connection relationship in the plurality of computing power device information, and the computing power device mapping association of the target area is constructed. By constructing the computing power device mapping association, support is provided for the next step of associated computing power device matching.

[0021] Receive a first processing task, and determine whether the remaining computing power space of the initial task processing device meets the first computing power resources required by the first processing task;

[0022] In an embodiment of the present application, a first processing task is received, and the first processing task is any one of a plurality of power grid operation tasks to be processed, wherein the first processing task includes a task processing-associated device and a first computing power resource required for task processing, and then a determination is made as to whether the remaining computing power space of the initial task processing device satisfies the first computing power resources required for the first processing task, wherein the initial task processing device is a computing power device that originally executes the first processing task.

[0023] When the condition is not satisfied, obtaining a first associated computing power device distribution based on the mapping association matching between the initial task processing device and the computing power device;

[0024] In an embodiment of the present application, when the remaining computing power space of the initial task processing device is less than the first computing power resources required for the first processing task, it is considered not met; when the remaining computing power space of the initial task processing device does not meet the first computing power resources required for the first processing task, the first associated computing power device distribution is obtained based on the mapping association matching between the initial task processing device and the computing power device.

[0025] like Figure 2 As shown, in one embodiment, the method further includes:

[0026] Obtaining a first safety requirement coefficient for the first processing task;

[0027] Obtaining a plurality of first initial associated computing power devices based on the mapping association matching between the initial task processing device and the computing power device;

[0028] Obtaining a first remaining computing power space and a first safety processing coefficient of the plurality of first initially associated computing power devices;

[0029] The multiple first initially associated computing power devices are screened based on the first computing power resources and the first security requirement, and the first initially associated computing power devices whose first remaining computing power space meets the first computing power resources and whose first security processing coefficient meets the first security requirement coefficient are extracted, and the first associated computing power device distribution is obtained in combination with the positioning identifier.

[0030] In the embodiment of the present application, first, a first safety requirement coefficient of the first processing task is obtained, and the first safety requirement coefficient is set based on the data processing safety requirement of the first processing task, wherein the greater the data processing safety requirement, the greater the first safety requirement coefficient. Then, according to the computing power device mapping association, the initial task processing device is matched with the device association to obtain multiple first initial associated computing power devices, and the first remaining computing power space and the first safety processing coefficient of the multiple first initial associated computing power devices are obtained.

[0031] The first remaining computing power space and the first security processing coefficient of the multiple first initially associated computing power devices are judged according to the first computing power resource and the first security requirement, and the first initially associated computing power devices whose first remaining computing power space meets the first computing power resource and whose first security processing coefficient meets the first security requirement coefficient are extracted to obtain multiple first associated computing power devices. The multiple first associated computing power devices are distributed based on the positioning identifier to obtain the distribution of the first associated computing power devices. By obtaining the distribution of the first associated computing power devices, support is provided for the next step of the associated computing power device matching analysis.

[0032] Performing delay prediction on each of the plurality of first computing power devices in the first associated computing power device distribution to obtain a plurality of first delay prediction values, and performing device credibility evaluation on each of the plurality of first computing power devices in the first associated computing power device distribution to generate a plurality of first device credibility;

[0033] In an embodiment of the present application, first, delay prediction is performed on each of the multiple first computing power devices in the first associated computing power device distribution to obtain multiple first delay prediction values, wherein the first delay prediction values ​​and the first computing power devices have a one-to-one correspondence; device trustworthiness assessment is performed on each of the multiple first computing power devices in the first associated computing power device distribution to generate multiple first device trustworthiness.

[0034] In one embodiment, the method further comprises:

[0035] Based on the digital twin technology, twin modeling is performed according to the information of the multiple computing power devices to obtain multiple device twin sub-models;

[0036] Based on the positioning identifier, the multiple device twin sub-models are connected in series to fuse and construct a twin model of the power grid equipment in the target area;

[0037] Randomly selecting a first device in the computing power device mapping association, and performing multiple delay simulations on multiple computing power associated devices of the first device through the power grid device twin model to obtain multiple computing power delay sets;

[0038] Calculate the average of the plurality of computing power delay sets to obtain the average of the computing power delays of the plurality of computing power associated devices;

[0039] Building a device delay matching table based on the first device, the multiple computing power associated devices, and the corresponding multiple computing power delay averages;

[0040] Based on the initial task processing device, delay matching is performed on multiple first computing power devices in the first associated computing power device distribution according to the device delay matching table to obtain multiple first delay prediction values.

[0041] In the embodiment of the present application, the digital twin is a simulation process that makes full use of physical models, sensor updates, operation history and other data, integrates multiple disciplines, multiple physical quantities, multiple scales, and multiple probabilities, completes mapping in the virtual space, and thus reflects the entire life cycle of the corresponding physical equipment.

[0042] First, based on the digital twin technology, twin simulation modeling is performed in a visual simulation platform according to the information of the multiple computing devices to obtain multiple device twin sub-models, wherein the visual simulation platform is a simulation modeling method commonly used by technicians in this field and will not be described in detail here. Then, according to the positioning identifier and the device connection relationship, the multiple device twin sub-models are connected in series and fused to obtain the power grid device twin model of the target area.

[0043] In the computing power device mapping association, a computing power device is randomly selected as the first device, and multiple power grid operation tasks are set, wherein the technical personnel in the field of the power grid operation tasks can customize the settings according to the actual situation. Then, based on the multiple power grid operation tasks, the multiple computing power associated devices of the first device are subjected to N delay simulations through the power grid device twin model, wherein N is an integer greater than or equal to 10, and the value of N can be set according to actual needs. The higher the actual demand accuracy, the larger the value of N, and N computing power delay sets are obtained.

[0044] Calculate the mean of the N computing power delay sets in sequence to obtain the computing power delay mean of the multiple computing power associated devices. Establish a first mapping association among the first device, the multiple computing power associated devices, and the corresponding multiple computing power delay mean values, and analyze in sequence based on the first mapping association to obtain multiple mapping associations of multiple devices in the computing power device mapping association. Based on the binary tree principle, take the computing power device as a child node, take the multiple computing power associated devices corresponding to the computing power device and the multiple computing power delay mean values ​​as the leaf nodes of the child node, and construct a device delay matching table based on the multiple mapping associations.

[0045] The initial task processing device and the first associated computing power device distribution are input into the device delay matching table for matching, and the first delay averages of multiple first computing power devices are obtained, and the first delay averages are used as the first delay prediction values ​​to obtain multiple first delay prediction values. By constructing a device delay matching table based on the binary tree principle to match device delay prediction values, the accuracy of delay prediction value matching can be improved, thereby improving the accuracy of obtaining the first delay prediction value, and further improving the accuracy of computing power device fitness evaluation.

[0046] In one embodiment, the method further comprises:

[0047] Connecting to the power grid operation management system to retrieve the operation log of the computing power device, and obtaining multiple historical operation data sets of the multiple first computing power devices based on the operation log of the computing power device;

[0048] Calculating equipment failure probabilities based on the multiple historical operation data sets to obtain multiple first equipment failure probabilities;

[0049] Perform device credibility assessment on the multiple first computing devices according to the multiple first device failure probabilities, and obtain multiple first device credibility according to the multiple credibility assessment results.

[0050] In the embodiment of the present application, first, the power grid operation management system is connected to retrieve the operation log of the computing power equipment, and further, multiple historical operation data sets of the multiple first computing power equipment are obtained according to the operation log of the computing power equipment, wherein the historical operation data sets include data such as the total operation times and the fault operation times. Then, according to the multiple historical operation data sets, the equipment failure probability of the multiple first computing power equipment is calculated in sequence, wherein the equipment failure probability is the ratio of the equipment failure operation times to the total operation times, and multiple first equipment failure probabilities are obtained.

[0051] Perform device credibility assessment on the multiple first computing power devices according to the multiple first device failure probabilities, wherein the smaller the first device failure probability is, the higher the device credibility assessment result is, obtain multiple credibility assessment results, and finally obtain multiple first device credibility based on the multiple credibility assessment results.

[0052] In one embodiment, the method further comprises:

[0053] Acquire multiple device usage information of the multiple first computing power devices, wherein the device usage information includes device service life percentage and device operating environment information;

[0054] Extracting first device usage information from the plurality of device usage information, wherein the first device usage information includes a first device service life percentage and first device operating environment information;

[0055] Optimizing and adjusting the service life percentage of the first device based on the operating environment information of the first device to obtain an optimized service life percentage of the first device;

[0056] Based on the first equipment service life optimization percentage, a plurality of equipment service life optimization percentages are obtained in sequence;

[0057] The plurality of trustworthy evaluation results are corrected according to the plurality of device service life optimization percentages to obtain a plurality of first device trustworthinesses.

[0058] In an embodiment of the present application, first, multiple device usage information of the multiple first computing power devices is obtained, wherein the device usage information corresponds to the first computing power devices one by one, wherein the device usage information includes a device service life percentage and device operating environment information, wherein the device service life percentage is a ratio of the used device life to the expected total service life, wherein the smaller the device service life percentage is, the higher the device operation stability is; the device operating environment information includes data such as temperature, humidity, and air pressure, wherein the device operating environment will affect the device service life.

[0059] A piece of equipment usage information is randomly extracted from the multiple equipment usage information as the first equipment usage information, wherein the first equipment usage information includes the first equipment service life percentage and the first equipment operating environment information. The first equipment service life percentage is further optimized and adjusted according to the first equipment operating environment information, wherein the optimization and adjustment method can obtain a first impact deviation of the environment on the equipment life by performing a correlation impact analysis on the first equipment use environment and the first equipment service life, and correct the first equipment service life percentage according to the first impact deviation to obtain the first equipment service life optimization percentage. By correcting the equipment service life percentage according to the equipment operating environment information, the accuracy of the equipment service life percentage setting can be improved, thereby improving the accuracy of the equipment credibility assessment.

[0060] The same method as that used to obtain the first device life optimization percentage is used to analyze multiple devices in the multiple device usage information in turn to obtain multiple device life optimization percentages. Finally, a stability weight is set according to the multiple device life optimization percentages, wherein the smaller the device life optimization percentage, the larger the corresponding stability weight, and the multiple trusted evaluation results are weighted according to the stability weight, and multiple first device credibility is obtained according to the weight calculation results. By obtaining the first device credibility, a basis is provided for the next step of calculating the device fitness.

[0061] Performing fitness calculation on the multiple first computing power devices based on the multiple first delay prediction values, the multiple first device credibility and the multiple first device occupied computing power predictions, and obtaining a first optimal computing power device according to the fitness calculation result, wherein the first optimal computing power device is a first computing power device with the largest fitness value;

[0062] In an embodiment of the present application, first, fitness calculation is performed on the multiple first computing power devices according to the multiple first delay prediction values, the multiple first device trustworthinesses and the multiple first device occupancy computing power predictions to obtain multiple fitness calculation results, and then the first computing power device with the largest fitness value among the multiple fitness calculation results is taken as the first optimal computing power device to obtain the first optimal computing power device.

[0063] By performing a comprehensive fitness calculation on the first computing power device based on the delay prediction value, device credibility and device occupied computing power prediction, the rationality and accuracy of the device fitness calculation can be improved, thereby improving the accuracy of the first optimal computing power device setting.

[0064] In one embodiment, the method further comprises:

[0065] Calculating the processing time of the first processing task based on the multiple first computing devices to obtain multiple first predicted processing times;

[0066] Performing occupied computing power prediction on the plurality of first computing power devices within the plurality of first prediction processing times through an occupied computing power prediction channel to obtain occupied computing power predictions of the plurality of first devices;

[0067] In the embodiment of the present application: First, the processing time of the first processing task is calculated based on the multiple first computing power devices, where the processing time is the ratio of the task amount of the first processing task to the calculation speed of the first computing power device, and multiple first predicted processing times are obtained. An occupied computing power prediction channel is constructed, and the occupied computing power prediction is performed on the multiple first computing power devices within the multiple first predicted processing times through the occupied computing power prediction channel, and multiple first device occupied computing power predictions are output.

[0068] In one embodiment, the method further comprises:

[0069] Based on the principle of multi-party learning integration, multiple learning operators are set, wherein the multiple learning operators at least include BP neural network, feedforward neural network, and support vector machine;

[0070] Taking the multiple learning operators as modeling directions, constructing multiple occupied computing power prediction sub-channels;

[0071] Performing supervised training on the plurality of occupied computing power prediction sub-channels to obtain a plurality of occupied computing power prediction sub-channels that meet preset training indicators, wherein each occupied computing power prediction sub-channel includes a prediction accuracy rate;

[0072] Performing trust identification on the plurality of occupied computing power prediction sub-channels according to the prediction accuracy to obtain a plurality of trust identification weights, wherein the trust identification weights and the occupied computing power prediction sub-channels have a mapping relationship;

[0073] Based on the mapping relationship, the occupied computing power prediction channel is established according to the multiple occupied computing power prediction sub-channels and the multiple trust identification weights.

[0074] In an embodiment of the present application, the method for constructing the occupied computing power prediction channel is as follows: first, multiple learning operators are set based on the principle of multi-party learning integration, wherein the multiple learning operators include at least BP neural network, feedforward neural network, and support vector machine. Technical personnel in this field can also set the number of learning operators according to actual needs, wherein the higher the required accuracy, the greater the number of operators.

[0075] Based on the multiple learning operators as modeling principles, multiple occupied computing power prediction sub-channels are constructed, wherein the occupied computing power prediction sub-channels at least include a BP neural network model, a feedforward neural network model, and a support vector machine model. Multiple sample training data are obtained based on historical equipment operation data, wherein the sample training data includes a historical time period and historical occupied computing power, and the sample training data is equally divided into first sample training data, second sample training data, and third sample training data.

[0076] The occupancy computing power prediction sub-channel constructed based on the BP neural network is supervised and trained by the first sample training data; the occupancy computing power prediction sub-channel constructed based on the feedforward neural network is supervised and trained by the second sample training data; the occupancy computing power prediction sub-channel constructed based on the support vector machine is supervised and trained by the third sample training data. The preset training index is obtained, wherein the preset training index can be set according to the actual situation, for example: iterative training 10,000 times. Multiple occupancy computing power prediction sub-channels that meet the preset training index and the corresponding prediction accuracy are obtained.

[0077] According to the prediction accuracy, the multiple occupied computing power prediction sub-channels are trust-marked to obtain multiple trust mark weights, wherein the higher the prediction accuracy, the larger the corresponding trust mark weight, wherein the trust mark weight and the occupied computing power prediction sub-channel have a one-to-one mapping relationship. Finally, based on the mapping relationship, the occupied computing power prediction channel is formed according to the multiple occupied computing power prediction sub-channels and the multiple trust mark weights, wherein the output result of the occupied computing power prediction channel is the sum of the weighted calculation results of the multiple occupied computing power prediction sub-channels. By constructing the occupied computing power prediction channel based on the principle of multi-party learning integration, the accuracy of the occupied computing power prediction can be further improved, thereby improving the accuracy of the equipment fitness calculation.

[0078] Based on the computing power fitness function, the fitness of the multiple first computing power devices is calculated according to the multiple first delay prediction values, the multiple first device credibility and the multiple first device occupied computing power predictions to obtain multiple fitness calculation results.

[0079] In an embodiment of the present application, a computing power fitness function is constructed, and then based on the computing power fitness function, fitness calculations are performed on the multiple first computing power devices respectively according to the multiple first delay prediction values, the multiple first device credibility and the multiple first device computing power occupancy predictions to obtain multiple fitness calculation results.

[0080] In one embodiment, the method further comprises:

[0081] The computing power fitness function expression is:

[0082] Among them, F i is the fitness of the i-th device among the plurality of first computing power devices, R i is the device credibility of the i-th device, Y i is the delay prediction value of the i-th device, S i is the remaining computing power space of the i-th device, s i It is the computing power prediction of the ith device.

[0083] In the embodiment of the present application, the computing power fitness function expression is: In the computing power fitness function, F i is the fitness of the i-th device among the plurality of first computing power devices, wherein the i-th device is any one of the plurality of first computing power devices; R i is the device credibility of the i-th device, where the higher the device credibility, the greater the device adaptability; Y i is the delay prediction value of the i-th device, where the smaller the delay prediction value, the greater the device fitness; S iis the real-time remaining computing power space of the i-th device, s i is the computing power prediction of the ith device, S i -s i The larger the difference, the greater the device fitness.

[0084] By constructing a computing power fitness function to calculate the fitness of the computing power device, the accuracy of the computing power device fitness calculation can be improved, thereby improving the accuracy of the first optimal computing power device setting.

[0085] The first processing task is performed based on the first optimal computing power device.

[0086] In the embodiment of the present application, the first processing task is finally executed according to the first optimal computing power device. The above method can solve the technical problem that the existing power grid computing power hardware matching calling method cannot quickly and accurately match computing power resources, resulting in the inability to timely process power grid operation tasks, resulting in low task processing efficiency. It can timely and accurately find suitable computing power resources for task processing, thereby improving the overall processing efficiency of power grid operation tasks, thereby further improving the reliability and safety of power grid operation.

[0087] In one embodiment, Figure 3 The invention provides a system for intelligent matching and calling of multi-type computing power hardware for power grid, including: computing power equipment information acquisition module 01, computing power equipment mapping association building module 02, remaining computing power space judgment module 03, first associated computing power equipment distribution acquisition module 04, first computing power equipment analysis module 05, first optimal computing power equipment acquisition module 06, first processing task execution module 07, wherein:

[0088] A computing power equipment information acquisition module 01, wherein the computing power equipment information acquisition module 01 is used to connect to the power grid operation management system of the target area to acquire a plurality of computing power equipment information, wherein the computing power equipment information has a positioning identifier;

[0089] A computing power device mapping association building module 02, wherein the computing power device mapping association building module 02 is used to build a computing power device mapping association of the target area based on the positioning identifier and the information of the plurality of computing power devices;

[0090] The remaining computing power space determination module 03 is used to receive the first processing task and determine whether the remaining computing power space of the initial task processing device meets the first computing power resources required by the first processing task;

[0091] A first associated computing power device distribution obtaining module 04, wherein the first associated computing power device distribution obtaining module 04 is used to obtain a first associated computing power device distribution based on the mapping association matching between the initial task processing device and the computing power device when the condition is not satisfied;

[0092] The first computing power device analysis module 05 is used to perform delay prediction on multiple first computing power devices in the first associated computing power device distribution, obtain multiple first delay prediction values, and perform device trustworthiness evaluation on multiple first computing power devices in the first associated computing power device distribution, and generate multiple first device trustworthiness;

[0093] The first optimal computing power device obtaining module 06 is used to calculate the fitness of the multiple first computing power devices based on the multiple first delay prediction values, the multiple first device credibility and the multiple first device occupied computing power predictions, and obtain the first optimal computing power device according to the fitness calculation result, wherein the first optimal computing power device is the first computing power device with the largest fitness value;

[0094] The first processing task execution module 07 is used to execute the first processing task based on the first optimal computing power device.

[0095] In one embodiment, the system further comprises:

[0096] A first safety requirement coefficient acquisition module, the first safety requirement coefficient acquisition module is used to acquire a first safety requirement coefficient of the first processing task;

[0097] A first initial associated computing power device acquisition module, the first initial associated computing power device acquisition module is used to obtain a plurality of first initial associated computing power devices based on the mapping association matching between the initial task processing device and the computing power device;

[0098] An information acquisition module, the information acquisition module is used to obtain a first remaining computing power space and a first safety processing coefficient of the plurality of first initially associated computing power devices;

[0099] The first associated computing power device distribution obtaining module is used to screen the multiple first initial associated computing power devices based on the first computing power resources and the first security requirement, extract the first initial associated computing power devices whose first remaining computing power space meets the first computing power resources and whose first security processing coefficient meets the first security requirement coefficient, and obtain the first associated computing power device distribution in combination with the positioning identifier.

[0100] In one embodiment, the system further comprises:

[0101] A device twin submodel obtaining module, wherein the device twin submodel obtaining module is used to perform twin modeling based on the multiple computing power device information based on the digital twin technology to obtain multiple device twin submodels;

[0102] A power grid equipment twin model construction module, wherein the power grid equipment twin model construction module is used to perform series fusion of the multiple equipment twin sub-models based on the positioning identifier to construct a power grid equipment twin model of the target area;

[0103] A computing power delay set obtaining module, wherein the computing power delay set obtaining module is used to randomly select a first device in the computing power device mapping association, and perform multiple delay simulations on multiple computing power associated devices of the first device through the power grid device twin model to obtain multiple computing power delay sets;

[0104] A computing power delay average value obtaining module, wherein the computing power delay average value obtaining module is used to calculate the average of the plurality of computing power delay sets to obtain the computing power delay average of the plurality of computing power associated devices;

[0105] A device delay matching table construction module, wherein the device delay matching table construction module is used to construct a device delay matching table based on the first device, the multiple computing power associated devices, and the corresponding multiple computing power delay averages;

[0106] The first delay prediction value acquisition module is used to perform delay matching on multiple first computing power devices in the first associated computing power device distribution based on the initial task processing device according to the device delay matching table to obtain multiple first delay prediction values.

[0107] In one embodiment, the system further comprises:

[0108] A historical operation data set acquisition module, the historical operation data set acquisition module is used to connect to the power grid operation management system to retrieve the operation log of the computing power device, and acquire multiple historical operation data sets of the multiple first computing power devices based on the operation log of the computing power device;

[0109] an equipment failure probability calculation module, the equipment failure probability calculation module being used to calculate the equipment failure probability based on the multiple historical operation data sets to obtain multiple first equipment failure probabilities;

[0110] A device trustworthy evaluation module is used to perform device trustworthiness evaluation on the multiple first computing devices according to the multiple first device failure probabilities, and obtain multiple first device trustworthiness according to the multiple trustworthy evaluation results.

[0111] In one embodiment, the system further comprises:

[0112] A device usage information acquisition module, the device usage information acquisition module is used to obtain a plurality of device usage information of the plurality of first computing power devices, wherein the device usage information includes a device service life percentage and device operating environment information;

[0113] A first device usage information extraction module, the first device usage information extraction module is used to extract first device usage information from the plurality of device usage information, wherein the first device usage information includes a first device service life percentage and first device operating environment information;

[0114] An optimization and adjustment module, the optimization and adjustment module is used to optimize and adjust the service life percentage of the first device based on the operating environment information of the first device to obtain an optimization percentage of the service life of the first device;

[0115] A module for obtaining an optimized percentage of equipment service life, wherein the module is used to obtain a plurality of optimized percentages of equipment service life in sequence based on the first optimized percentage of equipment service life;

[0116] A trustworthy evaluation result correction module is used to correct the plurality of trustworthy evaluation results according to the plurality of device service life optimization percentages to obtain a plurality of first device reliabilities.

[0117] In one embodiment, the system further comprises:

[0118] a processing time calculation module, the processing time calculation module being used to calculate the processing time of the first processing task based on the plurality of first computing devices to obtain a plurality of first predicted processing times;

[0119] An occupied computing power prediction module, the occupied computing power prediction module is used to perform occupied computing power prediction on the multiple first computing power devices within the multiple first prediction processing times through an occupied computing power prediction channel to obtain occupied computing power predictions of the multiple first devices;

[0120] A module for obtaining fitness calculation results is used to perform fitness calculations on the multiple first computing power devices based on a computing power fitness function, according to the multiple first delay prediction values, the multiple first device credibility and the multiple first device occupancy computing power predictions, to obtain multiple fitness calculation results.

[0121] In one embodiment, the system further comprises:

[0122] A learning operator setting module, wherein the learning operator setting module is used to set a plurality of learning operators based on the multi-party learning integration principle, wherein the plurality of learning operators at least include a BP neural network, a feedforward neural network, and a support vector machine;

[0123] An occupied computing power prediction sub-channel construction module, wherein the occupied computing power prediction sub-channel construction module is used to construct a plurality of occupied computing power prediction sub-channels with the plurality of learning operators as modeling directions;

[0124] An occupied computing power prediction sub-channel training module, wherein the occupied computing power prediction sub-channel training module is used to perform supervised training on the multiple occupied computing power prediction sub-channels to obtain multiple occupied computing power prediction sub-channels that meet preset training indicators, wherein each occupied computing power prediction sub-channel includes a prediction accuracy rate;

[0125] A trust identification module, the trust identification module is used to trust identify the multiple occupied computing power prediction sub-channels according to the prediction accuracy, and obtain multiple trust identification weights, wherein the trust identification weights and the occupied computing power prediction sub-channels have a mapping relationship;

[0126] The occupied computing power prediction channel forming module is used to form the occupied computing power prediction channel based on the mapping relationship, according to the multiple occupied computing power prediction sub-channels and the multiple trust identification weights.

[0127] In one embodiment, the system further comprises:

[0128] The computing power fitness function module refers to the computing power fitness function expression as follows:

[0129] Function parameter module, wherein the function parameter module refers to a module in which F i is the fitness of the i-th device among the plurality of first computing power devices, R i is the device credibility of the i-th device, Y i is the delay prediction value of the i-th device, S i is the remaining computing power space of the i-th device, s i It is the computing power prediction of the ith device.

[0130] In summary, compared with the prior art, the embodiments of the present disclosure have the following technical effects:

[0131] (1) By obtaining the first optimal computing power device to execute the first task, it is possible to timely and accurately find suitable computing power resources for task processing, thereby improving the overall processing efficiency of power grid operation tasks, thereby further improving the reliability and safety of power grid operation.

[0132] (2) By performing a comprehensive fitness calculation on the first computing power device based on the delay prediction value, device credibility and device occupied computing power prediction, the rationality and accuracy of the device fitness calculation can be improved, thereby improving the accuracy of the first optimal computing power device setting.

[0133] (3) By constructing a computing power prediction channel based on the principle of multi-party learning integration, the accuracy of computing power prediction can be further improved, thereby improving the accuracy of device fitness calculation.

[0134] The above-described embodiments only express several implementation methods of the present disclosure, but they cannot be understood as limiting the scope of the invention patent. Therefore, without departing from the scope of the present disclosure concept as defined by the attached claims, a person of ordinary skill in the art may make various types of substitutions, modifications and changes, and these substitutions, modifications and changes all belong to the protection scope of the present disclosure.

Claims

1. A method for intelligent matching and calling of multi-type computing power hardware in a power grid, characterized in that: The method comprises: Connecting to the power grid operation management system of the target area to obtain information of multiple computing devices, wherein the computing device information has a positioning identifier; Based on the positioning identifier, establishing a computing power device mapping association of the target area according to the information of the multiple computing power devices; Receiving a first processing task, and determining whether the remaining computing power space of the initial task processing device satisfies the first computing power resources required by the first processing task; When the condition is not satisfied, obtaining a first associated computing power device distribution based on the mapping association matching between the initial task processing device and the computing power device; Performing delay prediction on each of the plurality of first computing power devices in the first associated computing power device distribution to obtain a plurality of first delay prediction values, and performing device credibility evaluation on each of the plurality of first computing power devices in the first associated computing power device distribution to generate a plurality of first device credibility; Performing fitness calculation on the multiple first computing power devices based on the multiple first delay prediction values, the multiple first device credibility and the multiple first device occupied computing power predictions, and obtaining a first optimal computing power device according to the fitness calculation result, wherein the first optimal computing power device is a first computing power device with the largest fitness value; The first processing task is performed based on the first optimal computing power device.

2. The method according to claim 1, characterized in that The obtaining of a first associated computing power device distribution based on the mapping association matching between the initial task processing device and the computing power device also includes: Obtaining a first safety requirement coefficient for the first processing task; Obtaining a plurality of first initial associated computing power devices based on the mapping association matching between the initial task processing device and the computing power device; Obtaining a first remaining computing power space and a first safety processing coefficient of the plurality of first initially associated computing power devices; The multiple first initially associated computing power devices are screened based on the first computing power resources and the first security requirement, and the first initially associated computing power devices whose first remaining computing power space meets the first computing power resources and whose first security processing coefficient meets the first security requirement coefficient are extracted, and the first associated computing power device distribution is obtained in combination with the positioning identifier.

3. The method according to claim 1, characterized in that The step of respectively predicting the delay of the plurality of first computing power devices in the first associated computing power device distribution to obtain a plurality of first delay prediction values ​​further includes: Based on the digital twin technology, twin modeling is performed according to the information of the multiple computing power devices to obtain multiple device twin sub-models; Based on the positioning identifier, the multiple device twin sub-models are connected in series to fuse and construct a twin model of the power grid equipment in the target area; Randomly selecting a first device in the computing power device mapping association, and performing multiple delay simulations on multiple computing power associated devices of the first device through the power grid device twin model to obtain multiple computing power delay sets; Calculate the average of the plurality of computing power delay sets to obtain the average of the computing power delays of the plurality of computing power associated devices; Building a device delay matching table based on the first device, the multiple computing power associated devices, and the corresponding multiple computing power delay averages; Based on the initial task processing device, delay matching is performed on multiple first computing power devices in the first associated computing power device distribution according to the device delay matching table to obtain multiple first delay prediction values.

4. The method according to claim 1, characterized in that The performing device credibility evaluation on the plurality of first computing power devices in the first associated computing power device distribution to generate a plurality of first device credibility also includes: Connecting to the power grid operation management system to retrieve the operation log of the computing power device, and obtaining multiple historical operation data sets of the multiple first computing power devices based on the operation log of the computing power device; Calculating equipment failure probabilities based on the multiple historical operation data sets to obtain multiple first equipment failure probabilities; Perform device credibility assessment on the multiple first computing devices according to the multiple first device failure probabilities, and obtain multiple first device credibility according to the multiple credibility assessment results.

5. The method according to claim 4, characterized in that The step of obtaining a plurality of first device credibility according to a plurality of credibility evaluation results also includes: Acquire multiple device usage information of the multiple first computing power devices, wherein the device usage information includes device service life percentage and device operating environment information; Extracting first device usage information from the plurality of device usage information, wherein the first device usage information includes a first device service life percentage and first device operating environment information; Optimizing and adjusting the service life percentage of the first device based on the operating environment information of the first device to obtain an optimized service life percentage of the first device; Based on the first equipment service life optimization percentage, a plurality of equipment service life optimization percentages are obtained in sequence; The plurality of trustworthy evaluation results are corrected according to the plurality of device service life optimization percentages to obtain a plurality of first device trustworthinesses.

6. The method according to claim 1, characterized in that The step of calculating the fitness of the plurality of first computing power devices based on the plurality of first delay prediction values, the plurality of first device credibility and the plurality of first device occupied computing power predictions further includes: Calculating the processing time of the first processing task based on the multiple first computing devices to obtain multiple first predicted processing times; Performing occupied computing power prediction on the plurality of first computing power devices within the plurality of first prediction processing times through an occupied computing power prediction channel to obtain occupied computing power predictions of the plurality of first devices; Based on the computing power fitness function, the fitness of the multiple first computing power devices is calculated according to the multiple first delay prediction values, the multiple first device credibility and the multiple first device occupied computing power predictions to obtain multiple fitness calculation results.

7. The method according to claim 6, characterized in that The method further comprises: Based on the principle of multi-party learning integration, multiple learning operators are set, wherein the multiple learning operators at least include BP neural network, feedforward neural network, and support vector machine; Taking the multiple learning operators as modeling directions, constructing multiple occupied computing power prediction sub-channels; Performing supervised training on the plurality of occupied computing power prediction sub-channels to obtain a plurality of occupied computing power prediction sub-channels that meet preset training indicators, wherein each occupied computing power prediction sub-channel includes a prediction accuracy rate; Performing trust identification on the plurality of occupied computing power prediction sub-channels according to the prediction accuracy to obtain a plurality of trust identification weights, wherein the trust identification weights and the occupied computing power prediction sub-channels have a mapping relationship; Based on the mapping relationship, the occupied computing power prediction channel is established according to the multiple occupied computing power prediction sub-channels and the multiple trust identification weights.

8. The method according to claim 6, characterized in that The method further comprises: The computing power fitness function expression is: Among them, F i is the fitness of the i-th device among the plurality of first computing power devices, R i is the device credibility of the i-th device, Y i is the delay prediction value of the i-th device, S i is the remaining computing power space of the i-th device, s i It is the computing power prediction of the ith device.

9. An intelligent matching and calling system for multi-type computing hardware in power grids, characterized in that: The system is used to perform the steps of any one of the methods for intelligent matching and calling of multi-type computing power hardware for power grids described in claims 1-8, comprising: A computing power equipment information acquisition module, the computing power equipment information acquisition module is used to connect to the power grid operation management system of the target area to obtain a plurality of computing power equipment information, wherein the computing power equipment information has a positioning identifier; A computing power device mapping association building module, the computing power device mapping association building module is used to build a computing power device mapping association of the target area based on the positioning identifier and the information of the multiple computing power devices; A remaining computing power space determination module, the remaining computing power space determination module is used to receive the first processing task, and determine whether the remaining computing power space of the initial task processing device meets the first computing power resources required by the first processing task; A first associated computing power device distribution obtaining module, wherein the first associated computing power device distribution obtaining module is used to obtain a first associated computing power device distribution based on the initial task processing device and the computing power device mapping association matching when the condition is not satisfied; A first computing power device analysis module, the first computing power device analysis module is used to perform delay prediction on multiple first computing power devices in the first associated computing power device distribution, obtain multiple first delay prediction values, and perform device trustworthiness evaluation on multiple first computing power devices in the first associated computing power device distribution, and generate multiple first device trustworthiness; A first optimal computing power device obtaining module, the first optimal computing power device obtaining module is used to perform fitness calculation on the multiple first computing power devices based on the multiple first delay prediction values, the multiple first device credibility and the multiple first device occupied computing power predictions, and obtain the first optimal computing power device according to the fitness calculation result, wherein the first optimal computing power device is the first computing power device with the largest fitness value; A first processing task execution module, wherein the first processing task execution module is used to execute the first processing task based on the first optimal computing power device.