AI data processing method and system based on edge calculation and storage medium

By screening and optimizing edge computing carrier devices, the problems of response delay and system load in AI data processing are solved, and the efficiency and sustainability of data processing are improved.

CN119005354AActive Publication Date: 2024-11-22HUAYUAN CLOUD (SHENZHEN) NETWORK TECHNOLOGY CO LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202410927043.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-11
Publication Date
2024-11-22
Estimated Expiration
2044-07-11

AI Technical Summary

Technical Problem

The existing AI data processing methods have great delays in response speed, which leads to the inability to respond to emergencies in a timely manner and increases the system load when facing a large amount of data, affecting the data processing efficiency.

Method used

By acquiring multiple carrier devices that can perform edge computing, the optimal device is selected for data processing using the screening and optimization method, and the carrier device is replaced in real time with the data processing results to improve response speed and reduce system load.

Benefits of technology

It realizes improving response speed in AI data processing, preventing emergency processing failures caused by delays, and maintaining low latency when facing large amounts of data, reducing system load and ensuring the sustainability of data processing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119005354A_ABST
    Figure CN119005354A_ABST
Patent Text Reader

Abstract

The invention discloses an AI data processing method and system based on edge calculation and a storage medium, and relates to the technical field of AI data processing, and the method comprises the steps: obtaining a plurality of carrier devices ZT capable of carrying out edge calculation; screening the carrier equipment ZT, and optimizing the screening process; carrying out data processing on the basis of edge calculation by using the screened carrier equipment ZT through AI, and replacing the carrier equipment ZT which carries out the edge calculation in real time on the basis of the data processing; the method is used for solving the problems that in the aspect of the response speed of AI data processing, large response delay exists, the large response delay can cause that AI data processing cannot cope with emergency situations in time, and meanwhile the response delay of the AI during data processing can be increased and the load borne by an AI system can be increased when a large amount of data is processed.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of AI data processing technology, and specifically to an AI data processing method, system and storage medium based on edge computing. Background Art

[0002] AI data processing is a method that uses artificial intelligence technology to efficiently process and analyze large-scale data, aiming to extract valuable information from massive data and help companies make wise decisions. This method can automatically identify and organize structured and unstructured data, dig out hidden patterns and associations, and provide more accurate and comprehensive data insights.

[0003] Existing methods for AI data processing are usually optimized in terms of actual data analysis. For example, by improving the data processing of airborne intelligent sensor AI, the collection accuracy of drones during long-distance collection and the optimization of the collection path are improved. Although this improvement method can optimize the collection process of drones, there is still a large response delay in the response speed of AI data processing. This will result in that when the drone is far away from the terminal and there is an emergency such as entering a tunnel or misidentifying the collection object, the large response delay will cause the AI ​​data processing to be unable to respond to the emergency in time, causing the drone to be damaged and fall or erroneously collected. For example, in the Chinese patent application publication number Cn118072559A, a method for processing AI data based on drone airborne intelligent sensors is disclosed. This solution is to solve the problems of low accuracy of existing drone long-distance data transmission and unreasonable planning of collection routes, which leads to many invalid flight paths of drones. Other improvements for AI data processing include blurring the image through AI data processing in actual image processing to increase the image retention time. Although this improvement method can increase the image retention time, when a large number of pictures are received in a short period of time, the AI ​​data processing process for each picture needs to be processed through cloud computing, which will increase the response delay of AI when processing a large number of pictures and increase the load on the AI ​​system, resulting in the inability to convert the picture in time or even picture damage. In view of this, it is necessary to improve the existing AI data processing. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems in the prior art to a certain extent. By proposing an AI data processing method, system and storage medium based on edge computing, it is used to solve the problem that there is a large response delay in the response speed of AI data processing in the prior art. The large response delay will cause the AI ​​data processing to be unable to respond to emergencies in a timely manner. At the same time, facing a large amount of processing data, the response delay of AI during data processing will increase and the load on the AI ​​system will be increased.

[0005] To achieve the above objectives, in a first aspect, the present application provides an AI data processing method based on edge computing, comprising the following steps:

[0006] Acquire multiple carrier devices ZT capable of edge computing based on the location environment of AI;

[0007] Screen multiple carrier device ZTs using a carrier screening method, and optimize the screening process after each screening using a screening optimization method;

[0008] The carrier device ZT obtained through AI screening performs data processing based on edge computing, and the carrier device ZT performing edge computing is replaced in real time based on the data processing results.

[0009] Furthermore, obtaining multiple carrier devices ZT capable of edge computing based on the location environment of the AI ​​includes:

[0010] A three-dimensional rectangular coordinate system is established, denoted as the carrier acquisition coordinate system, where the units of the X-axis, Y-axis, and Z-axis of the three-dimensional rectangular coordinate system are all meters; the position of the electronic device where the AI ​​is located is fixed at the coordinate origin of the carrier acquisition coordinate system, and a sphere is made with the coordinate origin as the center and the radius as the standard scene radius, denoted as the carrier search sphere; the devices that can be used as edge computing carriers in the actual scene corresponding to the AI ​​acquisition carrier search sphere are denoted as carrier devices ZT1 to carrier devices ZT n , among which, the device that can serve as a carrier for edge computing is a terminal with computing capabilities.

[0011] Further, using the carrier screening method to screen multiple carrier devices ZT, and using the screening optimization method to optimize the screening process after each screening includes:

[0012] Screening multiple carrier devices ZT using a carrier screening method;

[0013] After each screening run, the screening process was optimized using the screening optimization method.

[0014] Furthermore, the electronic device where the AI ​​is located is recorded as the AI ​​device, and the carrier screening method includes:

[0015] The response sub-test method, the capability sub-test method and the depth sub-test method are used to obtain the best response time, the best response quantity and the best response depth of each carrier device ZT respectively; the response sub-test method includes: using the AI ​​device to send ICMP response request packets to all carrier devices ZT respectively, and the time taken by the response packet corresponding to each carrier device ZT is recorded as response time HY1 to response time HY n ; For any carrier device ZT n1, the AI ​​device repeatedly sends the data to the carrier device ZT n1 Send n ICMP echo request packets and obtain the corresponding response time of n echo packets n1 , respectively denoted as MS1 to MS n , use the response fluctuation algorithm to obtain the carrier device ZT n1 The response fluctuation algorithm includes: Among them, B is the response fluctuation, bi is MS1 to MS n The i-th response time HY n1 , bp is MS1 to MS n The average value of MS1 to MS n The minimum value of is recorded as the optimal response time, and the minimum value of MS1 to MS n The maximum value is recorded as MS max ;

[0016] Get the response fluctuations and MS corresponding to ZT of all carrier devices max and the best response time; the average value of all response fluctuations is recorded as the fluctuation mean, and the carrier device ZT with the smallest best response time among the carrier devices ZT whose response fluctuations are less than the fluctuation mean is recorded as the preferred response device.

[0017] Further, the ability subtest method includes: all MS max The maximum value in is recorded as the capability test time; during the capability test time, use the AI ​​device to send ICMP response request packets to all carrier devices ZT respectively, and send them again immediately after receiving the response packets, and obtain the number of ICMP response request packets sent by the AI ​​device to all carrier devices ZT during the capability test time, which are recorded as the best response number NL1 to the best response number NL n , wherein, when the AI ​​device sends an ICMP response request packet, it only sends an ICMP response request packet to one carrier device ZT at the same time;

[0018] The best response number NL1 to the best response number NL n The carrier device ZT corresponding to the maximum value in is recorded as the capability-optimized device;

[0019] The deep sub-test method includes: using big data to randomly obtain a keyword, which is recorded as the test keyword; using AI devices to connect the carrier device ZT1 to the carrier device ZT n Then, the test keyword is searched and the search results are recorded as carrier search results SJ1 to carrier search results SJ n , establish a plane rectangular coordinate system, recorded as the depth analysis coordinate system, where the unit of the X-axis of the depth analysis coordinate system is the search time, and the unit of the Y-axis is the number of data; for any carrier search result SJ n1, based on using AI devices to connect carrier devices ZT n1 The time of the post-search test keyword and the carrier search results SJ n1 The number of searched data corresponding to the increase in search time is plotted in the depth analysis coordinate system, which is recorded as the search depth curve QX n1 , where for the search depth curve QX n1 Any point (x0, y0) in the AI ​​device is connected to the carrier device ZT n1 When the time for searching the test keyword is x0, the amount of data searched is y0; draw the search depth curves QX corresponding to all carrier connection devices ZT in the same depth analysis coordinate system, mark the vertical coordinates corresponding to all search depth curves QX when X = capability test time as the optimal response depth of the carrier device ZT corresponding to each search depth curve QX, record the minimum value of all optimal response depths as the reference data volume, and record the carrier device ZT corresponding to the search depth curve QX with the vertical coordinate equal to the reference data volume and the smallest horizontal coordinate as the depth preferred device.

[0020] Furthermore, the vector screening method also includes:

[0021] A plane rectangular coordinate system is established, which is recorded as the comprehensive screening coordinate system, wherein the coordinate points from the origin to the right in the X-axis of the comprehensive screening coordinate system are recorded as carrier device ZT1 to carrier device ZT n ,The Y-axis coordinates of the comprehensive screening coordinate system are set to time, number of responses, and number of data;

[0022] When the Y-axis coordinate of the comprehensive screening coordinate system is time, based on the optimal response time corresponding to each carrier device ZT, the coordinate points are carrier device ZT1 to carrier device ZT n Draw n bar graphs at , recorded as time bar graphs, where the width of each time bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the time bar graph is the optimal response time corresponding to each carrier device ZT;

[0023] When the coordinate of the Y axis of the comprehensive screening coordinate system is the number of responses, the coordinate points are from the carrier device ZT1 to the carrier device ZT based on the optimal number of responses NL corresponding to each carrier device ZT. n Draw n bar graphs at , which are recorded as quantity bar graphs, where the width of each quantity bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the quantity bar graph is the number of best responses corresponding to each carrier device ZT;

[0024] When the coordinate of the Y axis of the comprehensive screening coordinate system is the number of data, based on the optimal response depth NL corresponding to each carrier device ZT, the coordinate points are carrier device ZT1 to carrier device ZT nDraw n bar graphs at , recorded as depth bar graphs, where the width of each depth bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the depth bar graph is the optimal response depth corresponding to each carrier device ZT;

[0025] The sum of the areas of the time bar graph, quantity bar graph, and depth bar graph corresponding to the carrier device ZT is recorded as the comprehensive screening value, and the average value of the comprehensive screening value is recorded as the screening mean value; the carrier device ZT whose comprehensive screening value among the response preferred device, the capability preferred device, and the depth preferred device is greater than the screening mean value is recorded as the screening preferred device, where the maximum value of the screening preferred device is 3; the carrier devices ZT other than the screening preferred device are sorted from large to small based on the comprehensive screening value, and are recorded as screening spare device BY1 to screening spare device BY2 respectively. f , where f is a positive integer less than or equal to n and greater than or equal to n-3.

[0026] Furthermore, the screening optimization method includes: after each screening of the carrier device ZT, when any carrier device ZT n1 When the carrier device ZT is recorded as the preferred device, the capability preferred device, and the depth preferred device at the same time, n1 It is recorded as the screening-free device. When the carrier screening method is used again to screen the carrier device ZT, the screening-free device is not screened.

[0027] Furthermore, the carrier device ZT obtained by using AI for data processing based on edge computing, and the carrier device ZT for edge computing is replaced in real time based on the data processing result, including:

[0028] When using AI devices for data processing, all carrier devices ZT in the environment of the AI ​​devices are screened based on the carrier screening method, and the data processing of the AI ​​devices is optimized using the edge computing processing method. The edge computing processing method includes:

[0029] When there are screening-free devices among all carrier devices ZT, the screening is stopped and the screening-free devices are used as terminals for data processing based on edge computing;

[0030] When there is no screening-free device among all carrier devices ZT, the preferred screening device is preferentially used as the terminal for data processing based on edge computing. When the screening result does not contain the preferred screening device, the screening backup device BY is selected as the terminal based on the sequence number of the screening backup device BY from small to large, and data processing is performed based on edge time;

[0031] When the AI ​​device uses any carrier device ZT as a terminal for data processing, the AI ​​device sends an ICMP response request packet to the carrier device ZT as the terminal every capability test time. When the time taken for the response packet is greater than the capability test time, the carrier device ZT as the terminal is recorded as a high-latency terminal. The edge computing processing method is used to analyze the carrier devices ZT except the high-latency terminals and reselect the carrier device ZT as the terminal.

[0032] In the second aspect, the present application also provides an AI data processing system based on edge computing, including a carrier acquisition module, a carrier screening and optimization module, and a carrier processing module;

[0033] The carrier acquisition module is used to acquire multiple carrier devices ZT capable of edge computing based on the location environment of the AI;

[0034] The carrier screening optimization module is used to screen multiple carrier devices ZT using a carrier screening method, and to optimize the screening process after each screening using a screening optimization method;

[0035] The carrier processing module is used to perform data processing based on edge computing using the carrier device ZT screened through AI, and to replace the carrier device ZT performing edge computing in real time based on the data processing results.

[0036] In a third aspect, the present application provides a storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps in the above method are performed.

[0037] Beneficial effects of the present invention: The present invention first obtains multiple carrier devices ZT that can perform edge computing based on the location environment of the AI; then uses the carrier screening method to screen the multiple carrier devices ZT, and uses the screening optimization method to optimize the screening process after each screening. The advantage of this is that by using the carrier screening method to screen multiple carrier devices ZT, it is possible to provide carrier device support for subsequent AI data processing improvements, ensuring that the screened carrier devices are the best carrier devices within the obtainable range when edge computing is used to assist AI in data processing, thereby effectively improving the processing efficiency during edge computing;

[0038] The present invention also uses the carrier device ZT obtained by screening through AI to perform data processing based on edge computing, and replaces the carrier device ZT performing edge computing in real time based on the data processing results. The advantage of this is that by optimizing AI data processing using edge computing based on the carrier device ZT, it is possible to give the local device the ability to process data when AI performs data processing, that is, there is no need to upload AI data to cloud computing during processing, which can improve the response speed of AI data processing and prevent the problem that AI data processing cannot respond to emergencies in a timely manner due to large response delays. At the same time, the faster response speed of AI data processing can allow the AI ​​system to always maintain a low response delay and reduce the load on the AI ​​system when facing a large amount of data. In addition, by replacing the carrier device ZT performing edge computing in real time based on the data processing results, the optimal carrier device ZT can be flexibly selected during the AI ​​data processing process to achieve edge computing, so that AI data processing always maintains a faster response delay, thereby ensuring the sustainability of AI data processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a functional block diagram of the system of the present invention;

[0040] Figure 2 is a flow chart of the steps of the method of the present invention;

[0041] Figure 3 A schematic diagram of obtaining the depth optimization device of the present invention;

[0042] Figure 4 It is a schematic diagram of a comprehensive screening coordinate system in which the Y-axis coordinate of the present invention is time;

[0043] Figure 5 It is a connection block diagram of the electronic device of the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] Example 1, first aspect, please refer to Figure 1 As shown, the present application provides an AI data processing system based on edge computing, including a carrier acquisition module, a carrier screening and optimization module, and a carrier processing module;

[0046] The carrier acquisition module is used to acquire multiple carrier devices ZT capable of edge computing based on the location environment of the AI.

[0047] The carrier acquisition module includes a carrier device grabbing unit, which is configured with a carrier device grabbing strategy. The carrier device grabbing strategy includes: establishing a three-dimensional rectangular coordinate system, recorded as the carrier acquisition coordinate system, wherein the units of the X-axis, Y-axis and Z-axis of the three-dimensional rectangular coordinate system are all meters; fixing the position of the electronic device where the AI ​​is located at the coordinate origin of the carrier acquisition coordinate system, and making a sphere with the coordinate origin as the center and the radius as the standard scene radius, recorded as the carrier search sphere;

[0048] In the specific implementation process, the standard scene radius is the effective distance for edge computing. In this embodiment, the standard scene radius is set to 1000m. In actual situations, the standard scene radius can be adjusted according to the signal transmission situation in the actual scene. For example, in an environment with low signal reception efficiency, the standard scene radius can be reduced to ensure that all carrier devices ZT in the carrier search sphere can achieve good signal transmission with the AI ​​device.

[0049] Use AI to obtain the devices that can be used as carriers for edge computing in the actual scene corresponding to the carrier search ball, which are recorded as carrier devices ZT1 to ZT n Among them, the device that can serve as a carrier for edge computing is a terminal with certain computing capabilities.

[0050] In the specific implementation process, the terminal with certain computing capabilities can be a mobile phone, gateway, router, server, etc., which can process data. For example, when an AI smart car is unmanned, the mobile phone carried by the occupants can be used as a carrier to support edge computing, and the mobile phone can be recorded as the carrier device ZT;

[0051] The carrier screening optimization module is used to screen multiple carrier devices ZT using a carrier screening method, and to optimize the screening process after each screening using a screening optimization method.

[0052] The carrier screening and optimization module is configured with a carrier device screening unit and a carrier screening and optimization unit; the carrier device screening unit is configured with a carrier device screening strategy, and the carrier device screening strategy includes:

[0053] Use the carrier screening method to screen multiple carrier devices ZT, and record the electronic device where AI is located as AI device. The carrier screening method includes:

[0054] The response sub-test method, the capability sub-test method and the depth sub-test method are used to obtain the best response time, the best response quantity and the best response depth of each carrier device ZT respectively.

[0055] The response sub-test method includes: using the AI ​​device to send ICMP response request packets to all carrier devices ZT respectively, and the time taken for the response packet corresponding to each carrier device ZT is recorded as response time HY1 to response time HY n ; For any carrier device ZT n1 , the AI ​​device repeatedly sends the data to the carrier device ZT n1 Send n ICMP echo request packets and obtain the corresponding response time of n echo packets n1 , respectively denoted as MS1 to MS n , use the response fluctuation algorithm to obtain the carrier device ZT n1 The response fluctuation algorithm includes: Among them, B is the response fluctuation, bi is MS1 to MS n The i-th response time HY n1 , bp is MS1 to MS n The average value of MS1 to MS n The minimum value of is recorded as the optimal response time, and the minimum value of MS1 to MS n The maximum value is recorded as MS max , sending ICMP response request packets is a test method for testing response speed. The shorter the time taken for the response packets after sending the ICMP response request packets, the lower the delay between the two. For example, in actual applications, when the AI ​​smart car is unmanned, the number of carrier devices ZT is 3 and they are all mobile phones. The AI ​​system sends 3 ICMP response request packets to 3 mobile phones and obtains the time taken for the response packets corresponding to the 3 mobile phones. Through data analysis, the 3 response times HY of the first mobile phone are 2ms, 1ms and 3ms respectively, the 3 response times HY of the second mobile phone are 20ms, 20ms and 20ms respectively, and the 3 response times HY of the third mobile phone are 10ms, 9ms and 11ms respectively. Then, the response fluctuations of the 3 mobile phones are calculated to be 0.67, 0 and 0.67 respectively; by obtaining the response time HY and response fluctuations of the same carrier device ZT multiple times, the network fluctuation of the carrier device ZT can be obtained. By selecting the carrier device ZT with smaller network fluctuations, it can be ensured that the signal transmission during edge computing is more stable;

[0056] Get the response fluctuations and MS corresponding to ZT of all carrier devices max and the best response time; the average value of all response fluctuations is recorded as the fluctuation mean, and the carrier device ZT with the smallest best response time among the carrier devices ZT whose response fluctuations are less than the fluctuation mean is recorded as the preferred response device.

[0057] The ability subtest method includes: all MS maxThe maximum value in is recorded as the capability test time; during the capability test time, use the AI ​​device to send ICMP response request packets to all carrier devices ZT respectively, and send them again immediately after receiving the response packets, and obtain the number of ICMP response request packets sent by the AI ​​device to all carrier devices ZT during the capability test time, which are recorded as the best response number NL1 to the best response number NL n , wherein, when the AI ​​device sends an ICMP response request packet, it only sends an ICMP response request packet to one carrier device ZT at the same time. The response capability of the carrier device ZT can be tested through the capability sub-test method. For example, some carrier devices ZT will only have a faster response speed when responding for the first time. If multiple response tests are frequently performed, the response will be slow. Selecting such carrier devices ZT for edge computing of AI data processing will cause AI data processing to be slow or even stuck when facing a large amount of data; for example, in actual application, when the AI ​​smart car is unmanned, the carrier device ZT is 3 mobile phones, and the capability test time is 50ms. Then, the AI ​​device can be used to send ICMP response request packets to the 3 mobile phones respectively within 3 50ms, and send them again immediately after receiving the response packets, so as to obtain the best response number NL. The larger the best response number NL, the more frequent and multiple responses and transmissions the mobile phone can make when transmitting data with the AI ​​device;

[0058] The best response number NL1 to the best response number NL n The carrier device ZT corresponding to the maximum value in is recorded as the capability-optimized device.

[0059] The deep sub-test method includes: using big data to randomly obtain a keyword, which is recorded as the test keyword; using AI devices to connect the carrier device ZT1 to the carrier device ZT n Then, the test keyword is searched and the search results are recorded as carrier search results SJ1 to carrier search results SJ n , establish a plane rectangular coordinate system, recorded as the depth analysis coordinate system, where the unit of the X-axis of the depth analysis coordinate system is the search time, and the unit of the Y-axis is the number of data; for any carrier search result SJ n1 , based on using AI devices to connect carrier devices ZT n1 The time of the post-search test keyword and the carrier search results SJ n1 The number of searched data corresponding to the increase in search time is plotted in the depth analysis coordinate system, which is recorded as the search depth curve QX n1 , where for the search depth curve QX n1 Any point (x0, y0) in the AI ​​device is connected to the carrier device ZT n1When the time for searching the test keyword is x0, the amount of data searched is y0; draw the search depth curves QX corresponding to all carrier connection devices ZT in the same depth analysis coordinate system, mark the vertical coordinates corresponding to all search depth curves QX when X = capability test time as the optimal response depth of the carrier device ZT corresponding to each search depth curve QX, record the minimum value of all optimal response depths as the reference data volume, and record the carrier device ZT corresponding to the search depth curve QX with the vertical coordinate equal to the reference data volume and the smallest horizontal coordinate as the depth preferred device.

[0060] In the specific implementation process, when the AI ​​smart car is unmanned driving, traffic signs can be selected as keywords. By obtaining the amount of data that each carrier device ZT can search within the capability test time, the data retrieval capability of the carrier device ZT in a short time can be obtained. When the AI ​​smart car recognizes an unrecorded traffic sign during unmanned driving, the more data that can be searched in a short time, the more likely it is to retrieve the meaning of the unrecorded traffic sign, thereby avoiding vehicle accidents. For example, in a data analysis, the depth analysis coordinate system obtained can be found in Figure 3 As shown, Z1 is the test capability time, curves LL1 to LL3 are search depth curves QX1 to QX3 respectively, the ordinate Z2 of point DD1 is the reference data volume, and through analysis, it can be obtained that the search depth curve QX with the ordinate equal to Z2 and the smallest abscissa is the search depth curve QX1, and the carrier device ZT corresponding to the search depth curve QX1 is the depth optimization device;

[0061] The carrier screening method further includes: establishing a plane rectangular coordinate system, which is recorded as a comprehensive screening coordinate system, wherein the coordinate points from the origin to the right in the X-axis of the comprehensive screening coordinate system are recorded as carrier devices ZT1 to carrier devices ZT n ,The Y-axis coordinates of the comprehensive screening coordinate system are set to time, number of responses, and number of data;

[0062] When the Y-axis coordinate of the comprehensive screening coordinate system is time, based on the optimal response time corresponding to each carrier device ZT, the coordinate points are carrier device ZT1 to carrier device ZT n Draw n bar graphs at , recorded as time bar graphs, where the width of each time bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the time bar graph is the optimal response time corresponding to each carrier device ZT;

[0063] When the coordinate of the Y axis of the comprehensive screening coordinate system is the number of responses, the coordinate points are from the carrier device ZT1 to the carrier device ZT based on the optimal number of responses NL corresponding to each carrier device ZT. nDraw n bar graphs at , which are recorded as quantity bar graphs, where the width of each quantity bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the quantity bar graph is the number of best responses corresponding to each carrier device ZT;

[0064] When the coordinate of the Y axis of the comprehensive screening coordinate system is the number of data, based on the optimal response depth NL corresponding to each carrier device ZT, the coordinate points are carrier device ZT1 to carrier device ZT n Draw n bar graphs at , recorded as depth bar graphs, where the width of each depth bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the depth bar graph is the optimal response depth corresponding to each carrier device ZT;

[0065] In the specific implementation process, for example, when the AI ​​smart car is unmanned, the optimal response time of the three mobile phones as carrier devices ZT1 to ZT3 is 2ms, 10ms and 8ms respectively. The corresponding comprehensive screening coordinate system and time bar chart can be found in Figure 4 As shown, the times corresponding to C1 to C5 are 2ms, 4ms, 6ms, 8ms and 10ms respectively, and V1 to V3 are carrier devices ZT1 to ZT3;

[0066] The sum of the areas of the time bar graph, quantity bar graph, and depth bar graph corresponding to the carrier device ZT is recorded as the comprehensive screening value, and the average value of the comprehensive screening value is recorded as the screening mean value; the carrier device ZT whose comprehensive screening value among the response preferred device, the capability preferred device, and the depth preferred device is greater than the screening mean value is recorded as the screening preferred device, where the maximum value of the screening preferred device is 3; the carrier devices ZT other than the screening preferred device are sorted from large to small based on the comprehensive screening value, and are recorded as screening spare device BY1 to screening spare device BY2 respectively. f , where f is a positive integer less than or equal to n and greater than or equal to n-3.

[0067] In the specific implementation process, the larger the comprehensive screening value is, the stronger the comprehensive capability of the carrier device ZT in terms of response time, number of responses in a short time, and number of searches in a short time is. The response preferred device, capability preferred device, and depth preferred device are the carrier devices ZT with the fastest response, the most frequent response times in a short time, and the strongest search capability in a short time among all carrier devices ZT. They can be preferentially used as carriers to assist AI devices in edge computing, thereby ensuring that the data processing terminal with the fastest response, the most frequent response times in a short time, or the strongest search capability in a short time within the optional range can be used when processing AI data.

[0068] The vector screening optimization unit is equipped with a vector screening optimization strategy, which includes:

[0069] After each screening run, the screening process was optimized using the screening optimization method.

[0070] The screening optimization method includes: after each screening of the carrier device ZT, when any carrier device ZT n1 When the carrier device ZT is recorded as the preferred device, the capability preferred device, and the depth preferred device at the same time, n1 Recorded as a non-screening device. When the carrier screening method is used again to screen the carrier device ZT, the non-screening device will not be screened. When the carrier device ZT is recorded as a preferred device, a capability preferred device, and a depth preferred device at the same time, it means that the carrier device ZT is the strongest device for data processing within the optional range of the AI ​​system. In the subsequent screening process, the device can be prioritized as the data processing terminal for edge computing of the AI ​​device without screening.

[0071] The carrier processing module is used to perform data processing based on edge computing using the carrier device ZT screened through AI, and to replace the carrier device ZT performing edge computing in real time based on the data processing results.

[0072] The carrier processing module is configured with a carrier collaborative processing unit, and the carrier collaborative processing unit is configured with a carrier collaborative processing strategy. The carrier collaborative processing strategy includes: when using AI devices for data processing, all carrier devices ZT in the environment where the AI ​​devices are located are screened based on the carrier screening method, and the data processing of the AI ​​devices is optimized using the edge computing processing method. The edge computing processing method includes:

[0073] When there are screening-free devices among all carrier devices ZT, the screening is stopped and the screening-free devices are used as terminals for data processing based on edge computing;

[0074] When there is no screening-free device among all carrier devices ZT, the screening preferred device is preferentially used as the terminal for data processing based on edge computing. When the screening result does not contain the screening preferred device, the screening backup device BY is selected as the terminal based on the serial number of the screening backup device BY from small to large, and data processing is performed based on edge time.

[0075] When the AI ​​device uses any carrier device ZT as a terminal for data processing, the AI ​​device sends an ICMP response request packet to the carrier device ZT as the terminal every capability test time. When the time taken for the response packet is greater than the capability test time, the carrier device ZT as the terminal is recorded as a high-latency terminal. The edge computing processing method is used to analyze the carrier devices ZT except the high-latency terminals and reselect the carrier device ZT as the terminal.

[0076] In the specific implementation process, by sending ICMP response request packets from the AI ​​device to the carrier device ZT as the terminal every capability test time when AI is processing data, the real-time response of the carrier device ZT as the carrier of edge computing can be detected in real time. When the carrier device ZT has a large delay due to excessive use, the carrier device ZT can be replaced in time to ensure that the edge computing of the AI ​​device can always maintain low latency and high stability. For example, in the actual application process, when the AI ​​smart car is unmanned, the capability test time is 50ms. Among the three carrier devices ZT that are mobile phones for screening, mobile phone No. 1 is recorded as a non-screening device, mobile phone No. 2 is recorded as a screening preferred device, and mobile phone No. 3 is recorded as a screening backup device BY. In the process of data processing when the AI ​​smart car uses mobile phone No. 1 for edge computing, the total time spent by a response packet is 100ms. Then mobile phone No. 1 is recorded as a high-latency terminal and mobile phone No. 2 is used to continue to assist the AI ​​smart car in edge computing, so as to prevent accidents caused by excessive delay of mobile phone No. 1 during driving of the AI ​​smart car.

[0077] Example 2, second aspect, please refer to Figure 2 As shown, the present application also provides an AI data processing method based on edge computing, comprising the following steps:

[0078] Step S1, obtaining multiple carrier devices ZT capable of edge computing based on the location environment of AI.

[0079] Step S1 includes: establishing a three-dimensional rectangular coordinate system, recorded as the carrier acquisition coordinate system, wherein the units of the X-axis, Y-axis and Z-axis of the three-dimensional rectangular coordinate system are all meters; fixing the position of the electronic device where the AI ​​is located at the coordinate origin of the carrier acquisition coordinate system, and making a sphere with the coordinate origin as the center and the radius as the standard scene radius, recorded as the carrier search sphere; using the AI ​​to obtain the devices that can be used as the edge computing carrier in the actual scene corresponding to the carrier search sphere, respectively recorded as carrier devices ZT1 to carrier devices ZT n Among them, the device that can serve as a carrier for edge computing is a terminal with certain computing capabilities.

[0080] Step S2, using a carrier screening method to screen multiple carrier devices ZT, and using a screening optimization method to optimize the screening process after each screening.

[0081] Step S2 includes: Step S201, using a carrier screening method to screen multiple carrier devices ZT. The electronic device where AI is located is recorded as AI device, and the carrier screening method includes:

[0082] Step S20101, using the response sub-test method, the capability sub-test method and the depth sub-test method to respectively obtain the best response time, the best response quantity and the best response depth of each carrier device ZT.

[0083] Step S20102, the response sub-test method includes: using the AI ​​device to send ICMP response request packets to all carrier devices ZT respectively, and the time taken for the response packet corresponding to each carrier device ZT is recorded as response time HY1 to response time HY n ; For any carrier device ZT n1 , the AI ​​device repeatedly sends the data to the carrier device ZT n1 Send n ICMP echo request packets and obtain the corresponding response time of n echo packets n1 , respectively denoted as MS1 to MS n , use the response fluctuation algorithm to obtain the carrier device ZT n1 The response fluctuation algorithm includes: Among them, B is the response fluctuation, bi is MS1 to MS n The i-th response time HY n1 , bp is MS1 to MS n The average value of MS1 to MS n The minimum value of is recorded as the optimal response time, and the minimum value of MS1 to MS n The maximum value is recorded as MS max ;

[0084] Step S20103, obtain the response fluctuations and MS corresponding to all carrier devices ZT max and the best response time; the average value of all response fluctuations is recorded as the fluctuation mean, and the carrier device ZT with the smallest best response time among the carrier devices ZT whose response fluctuations are less than the fluctuation mean is recorded as the preferred response device.

[0085] Step S20104, the capability sub-test method includes: max The maximum value in is recorded as the capability test time; during the capability test time, use the AI ​​device to send ICMP response request packets to all carrier devices ZT respectively, and send them again immediately after receiving the response packets, and obtain the number of ICMP response request packets sent by the AI ​​device to all carrier devices ZT during the capability test time, which are recorded as the best response number NL1 to the best response number NL n , wherein, when the AI ​​device sends an ICMP response request packet, it only sends an ICMP response request packet to one carrier device ZT at the same time;

[0086] The best response number NL1 to the best response number NL n The carrier device ZT corresponding to the maximum value in is recorded as the capability-optimized device.

[0087] Step S20105, the depth sub-test method includes: using big data to randomly obtain a keyword, which is recorded as a test keyword; using an AI device to connect the carrier device ZT1 to the carrier device ZT n Then, the test keyword is searched and the search results are recorded as carrier search results SJ1 to carrier search results SJ n , establish a plane rectangular coordinate system, recorded as the depth analysis coordinate system, where the unit of the X-axis of the depth analysis coordinate system is the search time, and the unit of the Y-axis is the number of data; for any carrier search result SJ n1 , based on using AI devices to connect carrier devices ZT n1 The time of the post-search test keyword and the carrier search results SJ n1 The number of searched data corresponding to the increase in search time is plotted in the depth analysis coordinate system, which is recorded as the search depth curve QX n1 , where for the search depth curve QX n1 Any point (x0, y0) in the AI ​​device is connected to the carrier device ZT n1 When the time for searching the test keyword is x0, the amount of data searched is y0; draw the search depth curves QX corresponding to all carrier connection devices ZT in the same depth analysis coordinate system, mark the vertical coordinates corresponding to all search depth curves QX when X = capability test time as the optimal response depth of the carrier device ZT corresponding to each search depth curve QX, record the minimum value of all optimal response depths as the reference data volume, and record the carrier device ZT corresponding to the search depth curve QX with the vertical coordinate equal to the reference data volume and the smallest horizontal coordinate as the depth preferred device.

[0088] Step S20106, the carrier screening method further includes: establishing a plane rectangular coordinate system, recorded as a comprehensive screening coordinate system, wherein the coordinate points from the origin to the right in the X-axis of the comprehensive screening coordinate system are recorded as carrier devices ZT1 to carrier devices ZT n ,The Y-axis coordinates of the comprehensive screening coordinate system are set to time, number of responses, and number of data;

[0089] Step S20107: when the coordinate of the Y axis of the comprehensive screening coordinate system is time, based on the optimal response time corresponding to each carrier device ZT, the coordinate points are carrier device ZT1 to carrier device ZT n Draw n bar graphs at , recorded as time bar graphs, where the width of each time bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the time bar graph is the optimal response time corresponding to each carrier device ZT;

[0090] Step S20108: when the coordinate of the Y axis of the comprehensive screening coordinate system is the number of responses, based on the best response number NL corresponding to each carrier device ZT, the coordinate points are carrier device ZT1 to carrier device ZT n Draw n bar graphs at , which are recorded as quantity bar graphs, where the width of each quantity bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the quantity bar graph is the number of best responses corresponding to each carrier device ZT;

[0091] Step S20109: when the coordinate of the Y axis of the comprehensive screening coordinate system is the number of data, based on the best response depth NL corresponding to each carrier device ZT, the coordinate points are carrier device ZT1 to carrier device ZT n Draw n bar graphs at , recorded as depth bar graphs, where the width of each depth bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the depth bar graph is the optimal response depth corresponding to each carrier device ZT;

[0092] Step S20110, record the sum of the areas of the time bar graph, quantity bar graph and depth bar graph corresponding to the carrier device ZT as the comprehensive screening value, and record the average value of the comprehensive screening value as the screening mean; record the carrier device ZT whose comprehensive screening value among the response preferred device, the capability preferred device and the depth preferred device is greater than the screening mean as the screening preferred device, wherein the maximum value of the screening preferred device is 3; sort the carrier devices ZT other than the screening preferred device from large to small based on the comprehensive screening value, and record them as screening spare device BY1 to screening spare device BY2 respectively. f , where f is a positive integer less than or equal to n and greater than or equal to n-3.

[0093] Step S202: After each screening, the screening process is optimized using a screening optimization method.

[0094] The screening optimization method includes: after each screening of the carrier device ZT, when any carrier device ZT n1 When the carrier device ZT is recorded as the preferred device, the capability preferred device, and the depth preferred device at the same time, n1 It is recorded as the screening-free device. When the carrier screening method is used again to screen the carrier device ZT, the screening-free device is not screened.

[0095] Step S3, using the carrier device ZT obtained through AI screening to perform data processing based on edge computing, and replacing the carrier device ZT performing edge computing in real time based on the data processing result.

[0096] Step S3 includes: Step S301, when using the AI ​​device for data processing, all carrier devices ZT in the environment where the AI ​​device is located are screened based on the carrier screening method, and the data processing of the AI ​​device is optimized using the edge computing processing method, and the edge computing processing method includes:

[0097] Step S3011: when there is a screening-free device among all carrier devices ZT, the screening is stopped and the screening-free device is used as a terminal for data processing based on edge computing;

[0098] Step S3012, when there is no screening-free device among all carrier devices ZT, the screening preferred device is preferentially used as the terminal for data processing based on edge computing. When the screening result does not contain the screening preferred device, the screening spare device BY is selected as the terminal based on the serial number of the screening spare device BY from small to large and data processing is performed based on edge time.

[0099] Step S302, when the AI ​​device uses any carrier device ZT as a terminal for data processing, the AI ​​device sends an ICMP response request packet to the carrier device ZT serving as the terminal every capability test time. When the time taken for the response packet is greater than the capability test time, the carrier device ZT serving as the terminal is recorded as a high-latency terminal, and the edge computing processing method is used to analyze the carrier devices ZT other than the high-latency terminal and reselect the carrier device ZT as the terminal.

[0100] Example 3, please refer to Figure 5 As shown, the present application also provides an electronic device, which may include: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other through the communication bus. The memory stores computer-readable instructions, and the processor can call the instructions in the memory. When the computer-readable instructions are executed by the processor, the steps in the AI ​​data processing method based on edge computing are executed to achieve the following functions: first, based on the location environment of the AI, multiple carrier devices ZT capable of edge computing are obtained; then, the carrier screening method is used to screen the multiple carrier devices ZT, and the screening optimization method is used to optimize the screening process after each screening; finally, the carrier device ZT obtained by the screening is used by AI to perform data processing based on edge computing, and the carrier device ZT for edge computing is replaced in real time based on the data processing results.

[0101] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art, and the computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk.

[0102] Embodiment 4, the present application also provides a computer-readable storage medium, the present application provides a storage medium, on which a computer program is stored, and when the computer program is executed by the processor, the steps in the AI ​​data processing method based on edge computing are executed. Through the above technical solution, when the computer program is executed by the processor, the method in any optional implementation of the above embodiment is executed to achieve the following functions: first, based on the location environment of the AI, multiple carrier devices ZT capable of edge computing are obtained; then, the carrier screening method is used to screen the multiple carrier devices ZT, and the screening optimization method is used to optimize the screening process after each screening; finally, the carrier device ZT obtained by the screening is used by AI to perform data processing based on edge computing, and the carrier device ZT for edge computing is replaced in real time based on the data processing results.

[0103] Through the description of the above implementation methods, the embodiments of the present invention can be provided as methods, systems or computer program products. Based on such an understanding, the above technical solutions can be essentially or partly contributed to the prior art in the form of software products, which can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and include several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0104] In the embodiments provided in the present application, it should be understood that the disclosed system or method can be implemented in other ways. The embodiments described above are only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules or units can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of systems, modules and units can be electrical, mechanical or other forms.

[0105] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. The AI ​​data processing method based on edge computing is characterized by: The steps include: Acquire multiple carrier devices ZT capable of edge computing based on the location environment of AI; Using a carrier screening method to screen multiple carrier device ZTs, and using a screening optimization method to optimize the screening process after each screening; The carrier device ZT obtained through AI screening performs data processing based on edge computing, and the carrier device ZT performing edge computing is replaced in real time based on the data processing results.

2. The AI ​​data processing method based on edge computing according to claim 1, characterized in that: Based on the location environment of AI, multiple carrier devices ZT capable of edge computing are obtained, including: A three-dimensional rectangular coordinate system is established, denoted as the carrier acquisition coordinate system, where the units of the X-axis, Y-axis, and Z-axis of the three-dimensional rectangular coordinate system are all meters; the position of the electronic device where the AI ​​is located is fixed at the coordinate origin of the carrier acquisition coordinate system, and a sphere is made with the coordinate origin as the center and the radius as the standard scene radius, denoted as the carrier search sphere; the devices that can be used as edge computing carriers in the actual scene corresponding to the AI ​​acquisition carrier search sphere are denoted as carrier devices ZT1 to carrier devices ZT n , among which, the device that can serve as a carrier for edge computing is a terminal with computing capabilities.

3. The AI ​​data processing method based on edge computing according to claim 2, characterized in that: Use the carrier screening method to screen multiple carrier device ZTs, and use the screening optimization method to optimize the screening process after each screening, including: Screening multiple carrier devices ZT using a carrier screening method; After each screening run, the screening process was optimized using the screening optimization method.

4. The AI ​​data processing method based on edge computing according to claim 3 is characterized in that: The electronic device where AI is located is recorded as AI device. The carrier screening method includes: The response sub-test method, the capability sub-test method and the depth sub-test method are used to obtain the best response time, the best response quantity and the best response depth of each carrier device ZT respectively; the response sub-test method includes: using the AI ​​device to send ICMP response request packets to all carrier devices ZT respectively, and the time taken by the response packet corresponding to each carrier device ZT is recorded as response time HY1 to response time HY n ; For any carrier device ZT n1 , the AI ​​device repeatedly sends the data to the carrier device ZT n1 Send n ICMP echo request packets and obtain the corresponding response time of n echo packets n1 , respectively denoted as MS1 to MS n , use the response fluctuation algorithm to obtain the carrier device ZT n1 The response fluctuation algorithm includes: Among them, B is the response fluctuation, bi is MS1 to MS n The i-th response time HY n1 , bp is MS1 to MS n The average value of MS1 to MS n The minimum value of is recorded as the optimal response time, and the minimum value of MS1 to MS n The maximum value is recorded as MS max ; Get the response fluctuations and MS corresponding to ZT of all carrier devices max and the best response time; the average value of all response fluctuations is recorded as the fluctuation mean, and the carrier device ZT with the smallest best response time among the carrier devices ZT whose response fluctuations are less than the fluctuation mean is recorded as the preferred response device.

5. The AI ​​data processing method based on edge computing according to claim 4 is characterized in that: The ability subtest method includes: all MS max The maximum value in is recorded as the capability test time; during the capability test time, use the AI ​​device to send ICMP response request packets to all carrier devices ZT respectively, and send them again immediately after receiving the response packets, and obtain the number of ICMP response request packets sent by the AI ​​device to all carrier devices ZT during the capability test time, which are recorded as the best response number NL1 to the best response number NL n , wherein, when the AI ​​device sends an ICMP response request packet, it only sends an ICMP response request packet to one carrier device ZT at the same time; The best response number NL1 to the best response number NL n The carrier device ZT corresponding to the maximum value in is recorded as the capability-optimized device; The deep sub-test method includes: using big data to randomly obtain a keyword, which is recorded as the test keyword; using AI devices to connect the carrier device ZT1 to the carrier device ZT n Then, the test keyword is searched and the search results are recorded as carrier search results SJ1 to carrier search results SJ n , establish a plane rectangular coordinate system, recorded as the depth analysis coordinate system, where the unit of the X-axis of the depth analysis coordinate system is the search time, and the unit of the Y-axis is the number of data; for any carrier search result SJ n1 , based on using AI devices to connect carrier devices ZT n1 The time of the post-search test keyword and the carrier search results SJ n1 The number of searched data corresponding to the increase in search time is plotted in the depth analysis coordinate system, which is recorded as the search depth curve QX n1 , where for the search depth curve QX n1 Any point (x0, y0) in the AI ​​device is connected to the carrier device ZT n1 When the time for searching the test keyword is x0, the amount of data searched is y0; draw the search depth curves QX corresponding to all carrier connection devices ZT in the same depth analysis coordinate system, mark the vertical coordinates corresponding to all search depth curves QX when X = capability test time as the optimal response depth of the carrier device ZT corresponding to each search depth curve QX, record the minimum value of all optimal response depths as the reference data volume, and record the carrier device ZT corresponding to the search depth curve QX with the vertical coordinate equal to the reference data volume and the smallest horizontal coordinate as the depth preferred device.

6. The AI ​​data processing method based on edge computing according to claim 5, characterized in that: Vector screening also includes: A plane rectangular coordinate system is established, which is recorded as the comprehensive screening coordinate system, wherein the coordinate points from the origin to the right in the X-axis of the comprehensive screening coordinate system are recorded as carrier device ZT1 to carrier device ZT n ,The Y-axis coordinates of the comprehensive screening coordinate system are set to time, number of responses, and number of data; When the Y-axis coordinate of the comprehensive screening coordinate system is time, based on the optimal response time corresponding to each carrier device ZT, the coordinate points are carrier device ZT1 to carrier device ZT n Draw n bar graphs at , recorded as time bar graphs, where the width of each time bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the time bar graph is the optimal response time corresponding to each carrier device ZT; When the coordinate of the Y axis of the comprehensive screening coordinate system is the number of responses, the coordinate points are from the carrier device ZT1 to the carrier device ZT based on the optimal number of responses NL corresponding to each carrier device ZT. n Draw n bar graphs at , recorded as quantity bar graphs, where the width of each quantity bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the quantity bar graph is the number of best responses corresponding to each carrier device ZT; When the coordinate of the Y axis of the comprehensive screening coordinate system is the number of data, based on the optimal response depth NL corresponding to each carrier device ZT, the coordinate points are carrier device ZT1 to carrier device ZT n Draw n bar graphs at , recorded as depth bar graphs, where the width of each depth bar graph is a unit length of the X axis of the comprehensive screening coordinate system, and the length of the depth bar graph is the optimal response depth corresponding to each carrier device ZT; The sum of the areas of the time bar graph, quantity bar graph, and depth bar graph corresponding to the carrier device ZT is recorded as the comprehensive screening value, and the average value of the comprehensive screening value is recorded as the screening mean value; the carrier device ZT whose comprehensive screening value among the response preferred device, the capability preferred device, and the depth preferred device is greater than the screening mean value is recorded as the screening preferred device, where the maximum value of the screening preferred device is 3; the carrier devices ZT other than the screening preferred device are sorted from large to small based on the comprehensive screening value, and are recorded as screening spare device BY1 to screening spare device BY2 respectively. f , where f is a positive integer less than or equal to n and greater than or equal to n-3.

7. The AI ​​data processing method based on edge computing according to claim 6, characterized in that: The screening optimization method includes: after each screening of the carrier device ZT, when any carrier device ZT n1 When the carrier device ZT is recorded as the preferred device, the capability preferred device, and the depth preferred device at the same time, n1 It is recorded as the screening-free device. When the carrier screening method is used again to screen the carrier device ZT, the screening-free device is not screened.

8. The AI ​​data processing method based on edge computing according to claim 7, characterized in that: The carrier device ZT obtained through AI screening performs data processing based on edge computing, and the carrier device ZT for edge computing is replaced in real time based on the data processing results, including: When using AI devices for data processing, all carrier devices ZT in the environment of the AI ​​devices are screened based on the carrier screening method, and the data processing of the AI ​​devices is optimized using the edge computing processing method. The edge computing processing method includes: When there are screening-free devices among all carrier devices ZT, the screening is stopped and the screening-free devices are used as terminals for data processing based on edge computing; When there is no screening-free device among all carrier devices ZT, the preferred screening device is preferentially used as the terminal for data processing based on edge computing. When the screening result does not contain the preferred screening device, the screening backup device BY is selected as the terminal based on the sequence number of the screening backup device BY from small to large, and data processing is performed based on edge time; When the AI ​​device uses any carrier device ZT as a terminal for data processing, the AI ​​device sends an ICMP response request packet to the carrier device ZT as the terminal every capability test time. When the time taken for the response packet is greater than the capability test time, the carrier device ZT as the terminal is recorded as a high-latency terminal. The edge computing processing method is used to analyze the carrier devices ZT except the high-latency terminals and reselect the carrier device ZT as the terminal.

9. An AI data processing system based on edge computing, used to implement the AI ​​data processing method based on edge computing according to any one of claims 1 to 8, characterized in that: It includes a vector acquisition module, a vector screening and optimization module, and a vector processing module; The carrier acquisition module is used to acquire multiple carrier devices ZT capable of edge computing based on the location environment of the AI; The carrier screening optimization module is used to screen multiple carrier devices ZT using a carrier screening method, and to optimize the screening process after each screening using a screening optimization method; The carrier processing module is used to perform data processing based on edge computing using the carrier device ZT screened through AI, and to replace the carrier device ZT performing edge computing in real time based on the data processing results.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are executed.

Citation Information

Patent Citations

  • Multiple access edge computing system for unmanned aerial vehicle network

    CN114760605A

  • Edge computing performance optimization system for data-intensive AI computing

    CN117453411A

  • Recursive server state monitoring method based on edge computing

    CN117478681A

  • Routing control system for real-time edge computing

    CN118075190A

  • Calculation task scheduling method and system based on delay perception and load balancing

    CN118233469A