CN116385857B - Calculation power distribution method based on AI intelligent scheduling - Google Patents
Calculation power distribution method based on AI intelligent scheduling Download PDFInfo
- Publication number
- CN116385857B CN116385857B CN202310645541.9A CN202310645541A CN116385857B CN 116385857 B CN116385857 B CN 116385857B CN 202310645541 A CN202310645541 A CN 202310645541A CN 116385857 B CN116385857 B CN 116385857B
- Authority
- CN
- China
- Prior art keywords
- calculation
- algorithm
- network
- intelligent
- computing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000004364 calculation method Methods 0.000 title claims abstract description 91
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 66
- 238000011160 research Methods 0.000 claims abstract description 12
- 239000010410 layer Substances 0.000 claims description 30
- 238000012544 monitoring process Methods 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 13
- 238000004458 analytical method Methods 0.000 claims description 11
- 238000011156 evaluation Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 9
- 238000004891 communication Methods 0.000 claims description 5
- 230000003993 interaction Effects 0.000 claims description 3
- 239000011229 interlayer Substances 0.000 claims description 3
- 239000002356 single layer Substances 0.000 abstract description 6
- 238000007726 management method Methods 0.000 description 10
- 230000006870 function Effects 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 6
- 238000004590 computer program Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000007405 data analysis Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000008447 perception Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000004065 semiconductor Substances 0.000 description 2
- 241000282806 Rhinoceros Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/96—Management of image or video recognition tasks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5072—Grid computing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/94—Hardware or software architectures specially adapted for image or video understanding
- G06V10/95—Hardware or software architectures specially adapted for image or video understanding structured as a network, e.g. client-server architectures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2209/00—Indexing scheme relating to G06F9/00
- G06F2209/50—Indexing scheme relating to G06F9/50
- G06F2209/502—Proximity
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The application relates to the technical field of calculation force distribution software, and discloses a calculation force distribution method based on AI intelligent scheduling, which comprises a calculation force server and a cloud server, wherein one or more calculation force networks are composed of a plurality of calculation force units so as to run an encryption algorithm, and the calculation force of the calculation force server is adjusted by distributing the plurality of calculation force units, so that the calculation force server supports at least one encryption algorithm; the AI intelligent recognition module inputs at least one calculation force to carry out scene recognition calculation, and a scene research and judgment result aiming at least one encryption algorithm is obtained. The application solves the problem of smaller local area network computing power through a computing power network algorithm and a cross entropy algorithm, and the number of users is gradually increased along with the increase of the users, but the number of users selecting network nodes is almost unchanged. The method is characterized in that the computing power resources of the network points are close to the task set which can be continuously made, and the computing power network can effectively solve the problems of small computing power or large time delay caused by a single-layer network.
Description
Technical Field
The application relates to the technical field of calculation force distribution software, in particular to an AI intelligent scheduling-based calculation force distribution method.
Background
With the continuous development of technologies such as AI, cloud computing, edge computing, big data, internet of things and the like and the continuous increase of data, deployment demands based on cloud, edge and end collaborative architecture are also increasing. The intelligent analysis gateway/cloud platform of the TSINGSEE green rhinoceros video not only fuses an AI intelligent recognition technology, but also relies on the architectural advantages of cloud, edge and end coordination, the terminal is responsible for data perception, the edge is responsible for local data analysis, the cloud end gathers perception data, business data and internet data of all edges, and finally services such as situation perception, analysis result output, data distribution and the like under a scene are completed.
Although the edge computing technology can be regarded as an extension of the cloud computing technology, due to the number of edge computing power modules, limited edge site resources and the like, a certain difference exists between the architecture of the edge computing and the architecture of the traditional cloud computing. If flexible scheduling of computing resources at edges and cores is to be achieved, a new architecture needs to be designed for resource management. For example, the current warehouse management monitoring system is a series of cloud computing maintenance and management systems, and only when the cloud computing maintenance and management system is fused with the currently advancing NFV system, the construction cost of a cloud resource pool can be reduced better, and the edge computing value is improved.
The 5G age brings about sinking of computing nodes, computing power is distributed over the whole network, and the service demand for computing power gradually presents diversified and diversified characteristics. Most of the monitoring systems are currently used for computing by a local area network, and the problem that the local area network belongs to a single-layer network is easy to generate small computing power or large delay is solved, and how to combine the network to schedule and fully utilize computing power resources and how to integrally transform the whole network is a problem which needs to be studied currently.
Disclosure of Invention
The application provides an AI intelligent scheduling-based computing power distribution method and system, which promote the improvement of the efficiency and quality of perceived data analysis output and data distribution by two intelligent distribution algorithms, namely a computing power network algorithm and a cross entropy algorithm, and are used for solving the problem of smaller computing power or larger delay of a local area network.
The application provides the following technical scheme: the utility model provides a power calculation distribution method based on AI intelligent scheduling, includes power calculation server and high in the clouds server, power calculation server includes:
at least one or more power networks, one or more control circuits, and an AI intelligent scheduling cluster;
the one or more power calculation networks are connected with the one or more control circuits, and each power calculation network consists of a plurality of power calculation units so as to run an encryption algorithm, and the power calculation units are distributed to adjust the power calculation of the power calculation server, so that the power calculation server supports at least one encryption algorithm;
the AI intelligent scheduling cluster comprises an automatic scheduling module and an AI intelligent identification module;
the AI intelligent recognition module and the one or more computing power modules are used for carrying out intelligent recognition and analysis on video images in the monitored scene, and supporting that at least one computing power is put into a scene recognition algorithm in the scene analysis process to obtain a scene research and judgment result aiming at the at least one encryption algorithm;
generating an early warning instruction and an auxiliary decision instruction according to the scene research and judgment result of the at least one encryption algorithm and the automatic scheduling module, wherein the early warning instruction and the auxiliary decision instruction are sent to the scene calculation result;
the calculation force distribution method comprises the following steps:
s1, collecting video images in a monitoring scene for identification and analysis;
s2, calculating by one or more calculation networks through a centralized calculation network algorithm and a cross entropy algorithm to obtain result instructions such as snapshot, comparison, alarm and the like;
s3, sending the result instruction to a cloud server through an automatic scheduling module to store an encryption algorithm, and generating a decision result which is required to be verified through at least one encryption algorithm;
and S4, outputting the decision result to an edge end in the cloud server, and enabling the edge end to quickly execute the decision result.
The method of the algorithm of the power network comprises the following steps:
one or more computing networks have M layers, a first layer comprising N users, each user having at least one task;
the user set algorithm is obtained by the following formula:seen as a set of tasks;
here, each task is not detachable, i.e. the user chooses to assign the task to a network node, obtaining the user's assignmentIs expressed as +.>；
wherein ,identifying task size, ++>Representing the processing density of the task, i.e. the number of revolutions of the central processor required to process the unit bit data;
the second layer to the M layer are network nodes, wherein the lower layer network nodes and the higher layer network nodes are all in a converging connection mode, namely, a certain network node of the M-1 layer is converged upwards and connected with a certain network node of the M layer, and the decision result is obtained by calculation after extracting the data of the certain network node;
the method of the cross entropy algorithm comprises the following steps:
the occurrence probability of rare events is corrected through continuous iteration, so that the corrected probability is continuously increased until the probability reaches convergence; the convergence probability is the optimal probability, and an optimal solution is obtained according to the optimal probability; the average cost function algorithm of the user is obtained through the following formula;
wherein ：for probability density coefficients, A is a weighted cost function(A) is the value space of A, < ->Representing computer processing tasks.
As an alternative scheme of the calculation power distribution method based on AI intelligent scheduling, the application comprises the following steps: the AI intelligent recognition module comprises an algorithm force recognition unit and an event unit;
the computing power recognition unit is used for obtaining algorithm categories, different computing power recognition units obtain different recognized categories, obtain a user computing task and analyze images of faces, human bodies and vehicles recognized in the computing task;
the event unit obtains the number of people invading and gathering in the identification area in the scene, independently sets a monitoring area for the preview picture of the entering acquisition camera, identifies the number of people entering the area, calculates the number of people in the monitoring area, and generates alarm information for events exceeding a threshold number of people.
As an alternative scheme of the intelligent AI-scheduling-based computing power distribution method, the user and the second-layer network node are in full connection, and the multi-layer computing power network does not consider interlayer direct communication.
As an alternative scheme of the computing power distribution method based on AI intelligent scheduling in the present application, the step S2 further includes:
receiving dynamic computing power information returned by each network node and a computing process;
and adjusting each network node according to the power calculation information and the calculation process.
As an alternative scheme of the calculation power distribution method based on AI intelligent scheduling, the application comprises the following steps: the automatic scheduling module scheduling method also comprises historical storage data, and the data of the network node is compared and analyzed with the data in the historical storage unit to obtain calculation evaluation similarity;
and after the evaluation identity is obtained, carrying out data interaction according to the evaluation identity and a strategy module in a history storage unit.
As an alternative scheme of the calculation power distribution method based on AI intelligent scheduling, the application comprises the following steps: in the step S4, a new early warning instruction and a decision result are obtained through the cloud server.
As an alternative scheme of the calculation power distribution method based on AI intelligent scheduling, the application comprises the following steps: and the control circuit distributes the early warning instruction to the edge end in the cloud server according to the new early warning instruction, optimizes the rule or model of the output business manufacturer, and sends the instruction to the edge end so as to enable the edge end to execute quickly.
The application has the following beneficial effects:
1. the application relates to an algorithm power distribution method and system based on AI intelligent scheduling, which is suitable for a warehouse management AI service system, and is characterized in that video in a monitored scene is collected, a plurality of algorithm power networks are arranged, video images in a plurality of algorithm power modules are identified and calculated to obtain scene research and judgment results of at least one encryption algorithm for the scene, the scene research and judgment results of at least one encryption algorithm are associated by combining with a network automatic scheduling module to generate early warning instructions and auxiliary decision instructions, the instructions are sent to an edge end, so that the edge end can rapidly execute the instructions, the algorithm power resources are scheduled and fully utilized by combining with the network, the association of an AI intelligent identification module and an automatic module is facilitated, the analysis and the output of perceived data are improved, and the efficiency and the quality of data distribution are promoted.
2. According to the calculation power distribution method and system based on the AI intelligent scheduling, a calculation power identification unit identifies, identification data is obtained through an event unit to carry out calculation power calculation, the number of people in the monitoring area is calculated, alarm information is generated for events exceeding a threshold number of people, centralized calculation power network calculation is carried out through a calculation power network, and a result instruction is sent to a cloud server through an automatic scheduling module to carry out decision processing; and the distribution of AI intelligent scheduling is promoted, and the full utilization of the associated computing capacity of the computing power network is improved.
3. The computing power distribution method and the computing power distribution system based on the AI intelligent scheduling are used for solving the problem of smaller computing power of a local area network through two computing power calculation methods of a computing power network algorithm and a cross entropy algorithm, and the number of users is gradually increased along with the increase of the users, but the number of users selecting network nodes is almost unchanged. The method is characterized in that the computing power resources of the network points are close to the task set which can be continuously made, and the computing power network can effectively solve the problems of small computing power or large time delay caused by a single-layer network.
Drawings
FIG. 1 is a schematic flow chart of the method of the application.
FIG. 2 is a schematic flow chart of the system of the present application.
FIG. 3 is a flow chart of the algorithm of the power network of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Embodiment 1 although the edge computing technology can be regarded as an extension of the cloud computing technology, there is a certain difference between the architecture of the edge computing and the architecture of the conventional cloud computing due to the numerous edge computing power modules, limited edge site resources, and the like. If flexible scheduling of computing resources at edges and cores is to be achieved, a new architecture needs to be designed for resource management. For example, the current warehouse management monitoring system is a series of cloud computing maintenance and management systems, and only when the cloud computing maintenance and management system is fused with the currently advancing NFV system, the construction cost of a cloud resource pool can be reduced better, and the edge computing value is improved.
The 5G age brings about sinking of computing nodes, computing power is distributed over the whole network, and the service demand for computing power gradually presents diversified and diversified characteristics. Most of the monitoring systems are currently used for computing by a local area network, and the problem that the local area network belongs to a single-layer network is easy to generate small computing power or large delay is solved, and how to combine the network to schedule and fully utilize computing power resources and how to integrally transform the whole network is a problem which needs to be studied currently.
The application provides the following technical scheme: referring to fig. 1-3, the computing power distribution method based on AI intelligent scheduling includes a computing power server and a cloud server, wherein the computing power server includes:
at least one or more power networks, one or more control circuits, and an AI intelligent scheduling cluster;
the one or more power calculation networks are connected with the one or more control circuits, and each power calculation network consists of a plurality of power calculation units so as to run an encryption algorithm, and the power calculation units are distributed to adjust the power calculation of the power calculation server, so that the power calculation server supports at least one encryption algorithm;
the AI intelligent scheduling cluster comprises an automatic scheduling module and an AI intelligent identification module;
the AI intelligent recognition module and the one or more computing power modules are used for carrying out intelligent recognition and analysis on video images in the monitored scene, and supporting that at least one computing power is put into a scene recognition algorithm in the scene analysis process to obtain a scene research and judgment result aiming at the at least one encryption algorithm;
and generating an early warning instruction and an auxiliary decision instruction according to the scene research and judgment result of the at least one encryption algorithm and the automatic scheduling module, wherein the early warning instruction and the auxiliary decision instruction are sent to the scene calculation result.
The application is suitable for a warehouse management AI service system, monitors video images in a scene, carries out recognition calculation on the video images in a plurality of calculation modules by setting a plurality of calculation networks, obtains scene research and judgment results of at least one encryption algorithm for the scene, associates the scene research and judgment results of at least one encryption algorithm with a network automatic scheduling module, generates early warning instructions and auxiliary decision instructions, sends the instructions to an edge end, enables the edge end to quickly execute, schedules and fully utilizes calculation resources with the combination of the network, is convenient for the association calculation of an AI intelligent recognition module and an automatic module, and improves the efficiency and quality of perceived data analysis output and data distribution.
Example 2 this example is illustrated in example 1, referring to figures 1-3,
wherein: the AI intelligent recognition module comprises an algorithm force recognition unit and an event unit;
the computing power recognition unit is used for obtaining algorithm categories, different computing power recognition units obtain different recognized categories, obtain a user computing task and analyze images of faces, human bodies and vehicles recognized in the computing task;
the event unit obtains the number of people invading and gathering in the identification area in the scene, independently sets a monitoring area for the preview picture of the entering acquisition camera, identifies the number of people entering the area, calculates the number of people in the monitoring area, and generates alarm information for events exceeding a threshold number of people.
Wherein: the calculation force distribution method comprises the following steps:
s1, collecting video images in a monitoring scene for identification and analysis;
s2, calculating by one or more calculation networks through a centralized calculation network algorithm and a cross entropy algorithm to obtain result instructions such as snapshot, comparison, alarm and the like;
s3, sending the result instruction to a cloud server through an automatic scheduling module to store an encryption algorithm, and generating a decision result which is required to be verified through at least one encryption algorithm;
and S4, outputting the decision result to an edge end in the cloud server, and enabling the edge end to quickly execute the decision result.
According to the embodiment, the power calculation identification unit is used for identification, the event unit is used for obtaining identification data to perform power calculation, the number of people in the monitoring area is calculated, alarm information is generated for events exceeding a threshold number of people, the power calculation network is used for performing centralized power calculation, and a result instruction is sent to the cloud server through the automatic scheduling module to perform decision processing; the distribution of AI intelligent scheduling is promoted and the associated computing capacity of the computing power network is improved; the method is used for solving the problem of smaller computational power of the local area network, and as the number of users increases, the number of users increases gradually, but the number of users selecting network nodes is almost unchanged. The method is characterized in that the computing power resources of the network points are close to the task set which can be continuously made, and the computing power network can effectively solve the problems of small computing power or large time delay caused by a single-layer network.
Example 3 this example is illustrated in example 2, referring to figures 1-3,
wherein: the method of the algorithm of the power network comprises the following steps: please refer to fig. 2:
one or more computing networks have M layers, a first layer comprising N users, each user having at least one task;
the user set algorithm is obtained by the following formula:and can also be considered as a set of tasks;
here we assume that each task is not detachable, i.e. the user chooses to assign the task to a network node, obtaining the user's assignmentIs expressed as +.>；
wherein ,identify task size (bit), a +.>Representing the processing density (CPU specific/bit) of the task, i.e. the number of revolutions of the CPU required to process the unit bit data;
the second layer to the M layer are network nodes, wherein the lower layer network nodes and the higher layer network nodes are all in a convergent connection mode, namely, a certain network node of the M-1 layer is converged upwards and connected with a certain network node of the M layer, and the decision result is obtained by calculation after extracting the data of the certain network node.
The user and the second layer network node are fully connected, and the multi-layer calculation network does not consider interlayer direct communication.
The method of the cross entropy algorithm comprises the following steps:
the occurrence probability of rare events is corrected through continuous iteration, so that the corrected probability is continuously increased until the probability reaches convergence;
the convergence probability is the optimal probability, and an optimal solution is obtained according to the optimal probability;
the average cost function algorithm of the user is obtained through the following formula;
wherein ：for probability density coefficients, A is a weighted cost function (A) is the value space of A, +.>Representing computer processing tasks.
In this embodiment, the computing power computing method is used to solve the problem of smaller computing power of the local area network through two computing power network algorithms and cross entropy algorithm, and as the number of users increases, the number of users increases gradually, but the number of users selecting network nodes is almost unchanged. The method is characterized in that the computing power resources of the network points are close to the task set which can be continuously made, and the computing power network can effectively solve the problems of small computing power or large time delay caused by a single-layer network.
Example 4 this example is illustrated in example 2, referring to figures 1-3,
wherein: after sending the computing power control instruction to each network node in the step S2, the method further includes:
receiving dynamic computing power information returned by each network node and a computing process;
and adjusting each network node according to the power calculation information and the calculation process.
Wherein: the automatic scheduling module scheduling method also comprises historical storage data, and the data of the network node is compared and analyzed with the data in the historical storage unit to obtain calculation evaluation similarity;
and after the evaluation identity is obtained, carrying out data interaction according to the evaluation identity and a strategy module in a history storage unit.
Wherein: in the step S4, a new early warning instruction and a decision result are obtained through the cloud server.
Wherein: and the control circuit distributes the early warning instruction to the edge end in the cloud server according to the new early warning instruction, optimizes the rule or model of the output business manufacturer, and sends the instruction to the edge end so as to enable the edge end to execute quickly.
The above embodiments may be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented in software, the above-described embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the processes or functions described in accordance with embodiments of the present application are produced in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from one website site, computer, server, or data center to another website site, computer, server, or data center by wired (e.g., infrared, wireless, microwave, etc.). The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains one or more sets of available media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium may be a solid state disk.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, and are not repeated herein.
In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other forms.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a random access memory (random access memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Finally: the foregoing description of the preferred embodiments of the application is not intended to limit the application, but to enable any modification, equivalent or improvement to be made without departing from the spirit and principles of the application.
The foregoing is merely a preferred embodiment of the present application, and it should be noted that it will be apparent to those skilled in the art that several modifications and variations can be made without departing from the technical principle of the present application, and these modifications and variations should also be regarded as the scope of the application.
Claims (7)
1. The utility model provides a power distribution method based on AI intelligent scheduling, includes power calculation server and high in the clouds server, its characterized in that: the computing force server includes:
one or more power networks, one or more control circuits and an AI intelligent scheduling cluster;
the one or more power calculation networks are connected with the one or more control circuits, and each power calculation network consists of a plurality of power calculation units so as to run an encryption algorithm, and the power calculation units are distributed to adjust the power calculation of the power calculation server, so that the power calculation server supports at least one encryption algorithm;
the AI intelligent scheduling cluster comprises an automatic scheduling module and an AI intelligent identification module;
the AI intelligent recognition module and the one or more calculation power modules are used for intelligently recognizing and analyzing the video images in the monitored scene, and at least one calculation power is input for scene recognition calculation in the scene analysis process to obtain a scene research and judgment result aiming at the at least one encryption algorithm;
according to the scene research and judgment result and the automatic scheduling module, generating an early warning instruction and an auxiliary decision instruction from the scene research and judgment result of the at least one encryption algorithm, and transmitting the early warning instruction and the auxiliary decision instruction into a scene calculation result;
the calculation force distribution method comprises the following steps:
s1, collecting video images in a monitoring scene for identification and analysis;
s2, calculating by one or more calculation networks through a centralized calculation network algorithm and a cross entropy algorithm to obtain result instructions such as snapshot, comparison, alarm and the like;
s3, sending the result instruction to a cloud server through an automatic scheduling module to store an encryption algorithm, and generating a decision result which is required to be verified through at least one encryption algorithm;
s4, outputting the decision result to an edge end in the cloud server, and enabling the edge end to quickly execute the decision result;
the method of the algorithm of the power network comprises the following steps:
one or more computing networks have M layers, a first layer comprising N users, each user having at least one task;
the user set algorithm is obtained by the following formula:seen as a set of tasks;
here, each task is not detachable, i.e. the user chooses to assign the task to a network node, obtaining the user's assignmentIs expressed as +.>；
wherein ,identifying task size, ++>Representing the processing density of the task, i.e. the number of revolutions of the central processor required to process the unit bit data;
the second layer to the M layer are network nodes, wherein the lower layer network nodes and the higher layer network nodes are all in a converging connection mode, namely, a certain network node of the M-1 layer is converged upwards and connected with a certain network node of the M layer, and the decision result is obtained by calculation after extracting the data of the certain network node;
the method of the cross entropy algorithm comprises the following steps:
the occurrence probability of rare events is corrected through continuous iteration, so that the corrected probability is continuously increased until the probability reaches convergence; the convergence probability is the optimal probability, and an optimal solution is obtained according to the optimal probability; the average cost function algorithm of the user is obtained through the following formula;
wherein ：/>For probability density coefficients, A is a weighted cost function, (A) is the value space of A, +.>Representing computer processing tasks.
2. The computing power distribution method based on intelligent AI scheduling according to claim 1, wherein: the AI intelligent recognition module comprises an algorithm force recognition unit and an event unit;
the computing power recognition unit is used for obtaining algorithm categories, different computing power recognition units obtain different recognized categories, obtain a user computing task and analyze images of faces, human bodies and vehicles recognized in the computing task;
the event unit obtains the number of people invading the identification area and gathering the people in the scene, sets an independent monitoring area for previewing the picture when entering the field of view of the acquisition camera, identifies the number of people entering the area, calculates the number of people in the monitoring area, and generates alarm information for events exceeding a threshold number of people.
3. The computing power distribution method based on intelligent AI scheduling according to claim 1, wherein: the user and the second layer network node are fully connected, and the multi-layer calculation network does not consider interlayer direct communication.
4. The computing power distribution method based on intelligent AI scheduling according to claim 2, wherein: the step S2 further includes:
receiving dynamic computing power information returned by each network node and a computing process;
and adjusting each network node according to the dynamic calculation information and the calculation process.
5. The AI-intelligent-scheduling-based computing power distribution method according to claim 4, wherein: the automatic scheduling module scheduling method also comprises historical storage data, and the data of the network node is compared and analyzed with the data in the historical storage unit to obtain calculation evaluation similarity;
and after the evaluation identity is obtained, carrying out data interaction according to the evaluation identity and a strategy module in a history storage unit.
6. The AI-intelligent-scheduling-based computing power distribution method according to claim 5, wherein: in the step S4, a new early warning instruction and a decision result are obtained through the cloud server.
7. The AI-intelligent-scheduling-based computing power distribution method according to claim 6, wherein: and the control circuit sends the new early warning instruction to the edge end in the cloud server according to the rule or model of the service manufacturer for optimizing output, and sends the instruction to the edge end so as to enable the edge end to execute quickly.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310645541.9A CN116385857B (en) | 2023-06-02 | 2023-06-02 | Calculation power distribution method based on AI intelligent scheduling |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310645541.9A CN116385857B (en) | 2023-06-02 | 2023-06-02 | Calculation power distribution method based on AI intelligent scheduling |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116385857A CN116385857A (en) | 2023-07-04 |
| CN116385857B true CN116385857B (en) | 2023-08-18 |
Family
ID=86971469
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310645541.9A Active CN116385857B (en) | 2023-06-02 | 2023-06-02 | Calculation power distribution method based on AI intelligent scheduling |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116385857B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117640410B (en) * | 2024-01-26 | 2024-04-12 | 深圳市迈腾电子有限公司 | Functional unit deconstruction method and device based on functional network family calculation force self-adaption |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111970154A (en) * | 2020-08-24 | 2020-11-20 | 浙江工商大学 | Unloading decision and resource allocation method based on deep reinforcement learning and convex optimization |
| CN115134371A (en) * | 2022-06-15 | 2022-09-30 | 中国电力科学研究院有限公司 | Scheduling method, system, equipment and medium containing edge network computing resources |
| CN115168024A (en) * | 2022-05-30 | 2022-10-11 | 李鹏 | Early warning method, device and equipment based on edge calculation |
| CN115562824A (en) * | 2022-09-22 | 2023-01-03 | 国网冀北电力有限公司张家口供电公司 | Computing resource cooperative scheduling system, method, device and storage medium |
| CN116032767A (en) * | 2022-12-30 | 2023-04-28 | 北京交通大学 | Intelligent fusion identification network-oriented computing power service chain management and control system architecture |
| CN116132403A (en) * | 2022-11-29 | 2023-05-16 | 浪潮通信技术有限公司 | Route distribution method and device of computing power network, electronic equipment and storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11574478B2 (en) * | 2020-06-30 | 2023-02-07 | Microsoft Technology Licensing, Llc | Machine perception using video/image sensors in an edge/service computing system architecture |
| US20230077708A1 (en) * | 2021-09-15 | 2023-03-16 | International Business Machines Corporation | Microservice measurement and merging |
-
2023
- 2023-06-02 CN CN202310645541.9A patent/CN116385857B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111970154A (en) * | 2020-08-24 | 2020-11-20 | 浙江工商大学 | Unloading decision and resource allocation method based on deep reinforcement learning and convex optimization |
| CN115168024A (en) * | 2022-05-30 | 2022-10-11 | 李鹏 | Early warning method, device and equipment based on edge calculation |
| CN115134371A (en) * | 2022-06-15 | 2022-09-30 | 中国电力科学研究院有限公司 | Scheduling method, system, equipment and medium containing edge network computing resources |
| CN115562824A (en) * | 2022-09-22 | 2023-01-03 | 国网冀北电力有限公司张家口供电公司 | Computing resource cooperative scheduling system, method, device and storage medium |
| CN116132403A (en) * | 2022-11-29 | 2023-05-16 | 浪潮通信技术有限公司 | Route distribution method and device of computing power network, electronic equipment and storage medium |
| CN116032767A (en) * | 2022-12-30 | 2023-04-28 | 北京交通大学 | Intelligent fusion identification network-oriented computing power service chain management and control system architecture |
Non-Patent Citations (1)
| Title |
|---|
| 算力时代中心云与边缘云的协同部署;武振宇等;《电信工程技术与标准化》(第5期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116385857A (en) | 2023-07-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11362961B2 (en) | Service processing method and system and device | |
| CN110087257B (en) | Task unloading device and method supporting mobile edge calculation | |
| CN109194984B (en) | Video frame scheduling method based on edge calculation | |
| CN110941667A (en) | Method and system for calculating and unloading in mobile edge calculation network | |
| CN116385857B (en) | Calculation power distribution method based on AI intelligent scheduling | |
| Gudkova et al. | Service failure and interruption probability analysis for licensed shared access regulatory framework | |
| CN110650487B (en) | Internet of things edge computing configuration method based on data privacy protection | |
| Li et al. | Method of resource estimation based on QoS in edge computing | |
| CN108768695B (en) | KQI problem positioning method and device | |
| CN112365366B (en) | Micro-grid management method and system based on intelligent 5G slice | |
| CN111542073A (en) | Heterogeneous network selection method and system for power service and network adaptation | |
| CN114466194A (en) | Video coding adjusting method and device, storage medium and electronic equipment | |
| CN110611831B (en) | Video transmission method and device | |
| CN115879723A (en) | Computing resource scheduling system and scheduling method based on cloud edge-end integrated platform | |
| CN113537400B (en) | Distribution and exit method of edge computing nodes based on branch neural network | |
| Zhang et al. | Mfvp: Mobile-friendly viewport prediction for live 360-degree video streaming | |
| CN117201410B (en) | Flow management method and system for Internet of things | |
| CN109495595B (en) | IP address allocation method, device, communication system and storage medium | |
| CN110708370B (en) | Data processing method and terminal | |
| CN111385128B (en) | Method and device for predicting burst load, storage medium, and electronic device | |
| CN115242717A (en) | Centralized storage system based on end-side calculation and control method | |
| CN109739513B (en) | Dynamic scheduling method and device for service requests under multi-edge cloud | |
| CN112686374A (en) | Deep neural network model collaborative reasoning method based on adaptive load distribution | |
| CN115082911A (en) | Video analysis method and device and video processing equipment | |
| CN113992687B (en) | Intelligent service cluster scheduling method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |