CN116781147B

CN116781147B - Artificial intelligence method and device based on satellite communication distributed computing system

Info

Publication number: CN116781147B
Application number: CN202310993603.5A
Authority: CN
Inventors: 闫晓亮; 蔡勇; 郭怀亮; 屈泉酉
Original assignee: Galaxyspace Beijing Communication Technology Co ltd
Current assignee: Galaxyspace Beijing Communication Technology Co ltd
Priority date: 2023-08-09
Filing date: 2023-08-09
Publication date: 2023-10-27
Anticipated expiration: 2043-08-09
Also published as: CN116781147A

Abstract

The application discloses an artificial intelligence method and device based on a satellite communication distributed computing system, wherein the method comprises the following steps: the terminal equipment responds to the operation of constructing the artificial intelligent model by a user and generates model information corresponding to the artificial intelligent model; the terminal equipment responds to operation of operation configuration between all the ground stations by a user, the artificial intelligent model unit is divided into a plurality of artificial intelligent model units, corresponding operation configuration information is generated, the ground stations interact with the terminal equipment through satellites, and the operation configuration information indicates distribution of the artificial intelligent model units among all the ground stations; the terminal equipment sends the operation configuration information to the ground station through the satellite; and the ground station constructs an artificial intelligent model unit corresponding to the operation configuration information according to the received operation configuration information. Thus, by the above method, the artificial intelligence service can be provided for users which are located in remote areas and cannot establish connection with the Internet.

Description

Artificial intelligence method and device based on satellite communication distributed computing system

Technical Field

The application relates to the field of satellite data transmission, in particular to an artificial intelligence method and device based on a satellite communication distributed computing system.

Background

With the development of satellite technology, distributed operation services based on satellites will be greatly developed. By this service, even if a user located in a remote area cannot provide an operation service through the internet and the cloud platform, a desired operation service can be obtained by a distributed operation based on satellites.

Currently, more and more platforms provide artificial intelligence and deep learning services to users, so that users can build their own artificial intelligence architecture by logging in the platform and train by using samples. However, when a user is located in a remote area and cannot communicate with the internet and thus cannot communicate with a cloud platform providing an artificial intelligence service, the artificial intelligence service cannot be obtained.

Although satellite-based distributed computing systems have been proposed, satellite-based distributed computing systems are implemented based on parallel computing by multiple satellites. However, for artificial intelligence models such as deep learning, the model scale is larger and larger, and a single satellite is difficult to complete the training work of the artificial intelligence model. Thus making it impossible for users in remote areas to obtain artificial intelligence services from satellite-based distributed computing systems.

Publication number CN105282038A, entitled distributed star group optimization method for stability analysis in mobile satellite networks. In the initial stage, link delay and service flow are used as measurement indexes of link importance, a central node is selected, an access factor is reconstructed by combining a shortest path algorithm and a minimum spanning tree algorithm, the node link importance is evaluated, and an initial network is constructed; then, local optimization is carried out on the network based on the service flow, an overhead matrix and a demand matrix are used as optimization conditions, evaluation parameters of all links in the network are corrected, a balance factor is constructed to adjust access priority, and the priority links are accessed into the network; and finally, global optimization is carried out based on stability, network service is updated while links are gradually accessed, and the distributed network reaches a stability threshold value through a circulation control mechanism.

The publication number is CN103634842A, and the name is a distributed satellite network inter-group routing method. Comprising the following steps: when data of a source satellite group in the distributed satellite network needs to be transmitted, a routing request is transmitted to a main satellite of each satellite group in the distributed network; when a routing request reaches a satellite group main satellite, collecting satellite information and an pheromone value in an AODV routing protocol; calculating the probability that the routing request is to be sent to each satellite according to a next hop selection algorithm, and selecting the next hop satellite for sending the routing request from each satellite according to the calculated probability of each satellite; updating local pheromone according to the link delay and the available bandwidth of the link after passing through one satellite, and selecting a next hop satellite for sending a routing request according to the local pheromone; when the destination satellite receives the route request, a route reply is generated, thereby establishing a reverse route between the source satellite and the destination satellite.

Aiming at the technical problem that users in the prior art, which are located in remote areas and cannot be connected with the Internet, cannot conveniently obtain artificial intelligence services, no effective solution is proposed at present.

Disclosure of Invention

The embodiment of the disclosure provides an artificial intelligence method and an artificial intelligence device based on a satellite communication distributed computing system, which at least solve the technical problem that users in remote areas cannot be connected with the Internet and cannot conveniently obtain artificial intelligence services in the prior art.

According to one aspect of the disclosed embodiments, there is provided an artificial intelligence method based on a satellite communication distributed computing system, including: the terminal equipment responds to the operation of constructing the artificial intelligent model by a user and generates model information corresponding to the artificial intelligent model; the terminal equipment responds to operation of operation configuration between all the ground stations by a user, the artificial intelligent model is divided into a plurality of artificial intelligent model units, corresponding operation configuration information is generated, the ground stations interact with the terminal equipment through satellites, and the operation configuration information indicates distribution of the artificial intelligent model units among all the ground stations; the terminal equipment sends the operation configuration information to the ground station through the satellite; and the ground station constructs an artificial intelligent model unit corresponding to the operation configuration information according to the received operation configuration information.

According to another aspect of the embodiments of the present disclosure, there is also provided a storage medium including a stored program, wherein the method of any one of the above is performed by a processor when the program is run.

According to another aspect of the embodiments of the present disclosure, there is also provided an artificial intelligence device based on a satellite communication distributed computing system, including: the model information generation module is used for responding to the operation of constructing the artificial intelligent model by the user through the terminal equipment and generating model information corresponding to the artificial intelligent model; the system comprises an operation configuration information generation module, a control module and a control module, wherein the operation configuration information generation module is used for dividing an artificial intelligent model into a plurality of artificial intelligent model units through a terminal device in response to operation of operation configuration between all ground stations of a user, and generating corresponding operation configuration information, wherein the ground stations interact with the terminal device through satellites, and the operation configuration information indicates distribution of the artificial intelligent model units among all the ground stations; the configuration information sending module is used for sending the operation configuration information to the ground station through the satellite by the terminal equipment; and the model unit construction module is used for constructing an artificial intelligent model unit corresponding to the operation configuration information according to the received operation configuration information through the ground station.

Thus, in accordance with an embodiment of the present application, there is provided an artificial intelligence method for a satellite-based distributed computing system that utilizes ground stations in communication with terminal devices via satellites to implement distributed computing. Therefore, compared with the operation by using satellites, the ground station is arranged on the ground, so that the ground station can bear the electric energy loss caused by large operation amount, and the ground station can provide energy assurance for artificial intelligent service. Furthermore, according to embodiments of the present application, an artificial intelligence model constructed by a user may be divided into a plurality of different model units, and each model unit may be associated with a different ground station, respectively, such that the different model units of the artificial intelligence model are loaded at the different ground stations. The advantage of the number of ground stations can be utilized to provide the operational resources required by the artificial intelligence model. Thus, by the above method, the artificial intelligence service can be provided for users which are located in remote areas and cannot establish connection with the Internet. Therefore, the technical problem that users in remote areas cannot be connected with the Internet and cannot conveniently obtain artificial intelligence services in the prior art is solved.

Drawings

The accompanying drawings, which are included to provide a further understanding of the disclosure, illustrate and explain the present disclosure, and together with the description serve to explain the present disclosure. In the drawings:

FIG. 1 is a schematic diagram of a satellite communication distributed computing system according to a first aspect of embodiment 1 of the present application;

FIG. 2A further illustrates a schematic diagram of the hardware architecture of the satellite of FIG. 1;

fig. 2B further illustrates a schematic diagram of the hardware architecture of the ground station of fig. 1;

fig. 3A further shows a block diagram of a client of the terminal device of fig. 1;

FIG. 3B further illustrates a block diagram of the server side of the ground station of FIG. 1;

FIG. 4 is a flow chart of an artificial intelligence method based on a satellite communication distributed computing system according to the first aspect of embodiment 1 of the present application;

fig. 5A is a schematic diagram of an interface when a user starts a client displayed by the terminal device according to the first aspect of embodiment 1 of the present application;

FIG. 5B is a schematic diagram of a navigation interface for constructing an artificial intelligence model for a client of a terminal device according to the first aspect of embodiment 1 of the present application;

FIG. 5C is a schematic diagram of an interface for a client of a terminal device for building and editing an artificial intelligence model according to the first aspect of embodiment 1 of the present application;

FIG. 5D is a schematic diagram of an interface for constructing and editing an artificial intelligence model at the interface of a client of a terminal device for a user according to the first aspect of embodiment 1 of the present application;

FIG. 5E is a schematic illustration of a user-constructed artificial intelligence model according to the first aspect of embodiment 1 of the application;

FIG. 6 is a schematic diagram of model information of an artificial intelligence model according to the first aspect of embodiment 1 of the present application;

FIG. 7A is a schematic diagram of an operational configuration interface of a client according to the first aspect of embodiment 1 of the present application;

FIG. 7B is a schematic diagram of a user dividing an artificial intelligence model into a plurality of model elements at an operational configuration interface according to the first aspect of embodiment 1 of the present application;

FIG. 7C is a schematic diagram of an artificial intelligence model divided into a plurality of model elements according to the first aspect of embodiment 1 of the present application;

FIG. 7D is a schematic illustration of a user associating a ground station with a model element according to the first aspect of embodiment 1 of the present application;

FIG. 8 is a schematic diagram of a neural network model for determining a priority probability of a ground station according to the first aspect of embodiment 1 of the present application;

FIG. 9 is a flow chart of training an artificial intelligence model with various ground stations according to the first aspect of embodiment 1 of the present application;

fig. 10 is a flowchart of processing data to be processed using respective ground stations according to the first aspect of embodiment 1 of the present application; and

fig. 11 is a schematic diagram of an artificial intelligence device based on a satellite communication distributed computing system according to the first aspect of embodiment 2 of the present application.

Detailed Description

In order to better understand the technical solutions of the present disclosure, the following description will clearly and completely describe the technical solutions of the embodiments of the present disclosure with reference to the drawings in the embodiments of the present disclosure. It will be apparent that the described embodiments are merely embodiments of a portion, but not all, of the present disclosure. All other embodiments, which can be made by one of ordinary skill in the art without inventive effort, based on the embodiments in this disclosure, shall fall within the scope of the present disclosure.

It should be noted that the terms "first," "second," and the like in the description and claims of the present disclosure and in the foregoing figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the disclosure described herein may be capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

Example 1

According to the present embodiment, there is provided a method embodiment of a satellite-based distributed network processing data, it being noted that the steps illustrated in the flowcharts of the figures may be performed in a computer system such as a set of computer executable instructions, and although a logical order is illustrated in the flowcharts, in some cases the steps illustrated or described may be performed in an order other than that herein.

Fig. 1 shows a schematic diagram of a satellite-based artificial intelligence service system according to the present embodiment. Referring to fig. 1, the system includes a terminal device 10, a plurality of satellites 201 to 203, ground stations 301 to 303 corresponding to communication areas, and a database 400 communicatively connected to the ground stations 301 to 303. Wherein the ground stations 301-303 are also communicatively coupled to each other to form a computing system 300 that provides computing power. That is, unlike the prior art, the present invention uses the ground stations 301 to 303 in communication with the satellites 201 to 203 as computing systems for providing computing power, instead of using the satellites 201 to 203 as computing systems for providing computing power, and the satellites 201 to 203 are merely used for realizing communication between the terminal device 10 and the ground stations 301 to 303. In addition, database 400 stores data, such as training sample sets, for performing tasks related to artificial intelligence.

Thus, the terminal device 10 may communicate with each ground station 301 to 303 through the satellites 201 to 203, thereby deploying an operation task to the ground stations 301 to 303.

Fig. 2A further illustrates a schematic diagram of a hardware architecture of the satellite 20 (201-203) in fig. 1. Referring to fig. 2A, the satellite 20 includes an integrated electronic system including: processor, memory, bus management module and communication interface. Wherein the memory is coupled to the processor such that the processor can access the memory, read program instructions stored in the memory, read data from the memory, or write data to the memory. The bus management module is connected to the processor and also to a bus, such as a CAN bus. The processor can communicate with the satellite-borne peripheral connected with the bus through the bus managed by the bus management module. In addition, the processor is also in communication connection with the camera, the star sensor, the measurement and control transponder, the data transmission equipment and other equipment through the communication interface. It will be appreciated by those of ordinary skill in the art that the configuration shown in fig. 2A is merely illustrative and is not intended to limit the configuration of the electronic device described above. For example, the satellite may also include more or fewer components than shown in FIG. 2A, or have a different configuration than shown in FIG. 2A.

Fig. 2B further illustrates a schematic diagram of a hardware architecture of the ground station 30 (301-303) in fig. 1. Referring to fig. 2B, the ground station 30 may include one or more processors (which may include, but are not limited to, a microprocessor MCU, a processing device such as a programmable logic device FPGA), a memory for storing data, a transmission device for communication functions, and an input/output interface. Wherein the memory, the transmission device and the input/output interface are connected with the processor through a bus. In addition, the method may further include: a display connected to the input/output interface, a keyboard, and a cursor control device. It will be appreciated by those of ordinary skill in the art that the configuration shown in fig. 2B is merely illustrative and is not intended to limit the configuration of the electronic device described above. For example, the ground station may also include more or fewer components than shown in fig. 2B, or have a different configuration than shown in fig. 2B.

It should be noted that one or more of the processors and/or other data processing circuits shown in fig. 2A and 2B may be referred to herein generally as a "data processing circuit. The data processing circuit may be embodied in whole or in part in software, hardware, firmware, or any other combination. Furthermore, the data processing circuitry may be a single stand-alone processing module, or incorporated in whole or in part into any of the other elements in the computing device. As referred to in the embodiments of the present disclosure, the data processing circuit acts as a processor control (e.g., selection of the variable resistance termination path to interface with).

The memories shown in fig. 2A and 2B may be used to store software programs and modules of application software, such as program instructions/data storage devices corresponding to the methods of processing data in a satellite-based distributed network in the embodiments of the present disclosure, and the processor may execute various functional applications and data processing by executing the software programs and modules stored in the memories, that is, the methods of processing data in a satellite-based distributed network implementing the application programs described above. The memory may include high speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid state memory

It should be noted here that in some alternative embodiments, the apparatus shown in fig. 2A and 2B described above may include hardware elements (including circuits), software elements (including computer code stored on a computer readable medium), or a combination of both hardware elements and software elements. It should be noted that fig. 2A and 2B are only one example of a specific example, and are intended to illustrate the types of components that may be present in the above-described devices.

Wherein fig. 3A shows a schematic diagram of a client 100 deployed in a terminal device 10. Referring to fig. 3A, the client 100 includes: artificial intelligence module 110, operational configuration module 120, user interface module 130, and satellite communications module 140.

The user interface module 130 is configured to interact with a user of the terminal device 10 through an interface, and display a calculation result to the user, or transmit an instruction of the user to the artificial intelligence module 110 and the operation configuration module 120.

The artificial intelligence module 110 is used for displaying an interface related to the operation of the artificial intelligence on the display screen of the terminal device 10 through the user interface module 130, and receiving the operation of the user, including building a model, configuring parameters, inputting data information to be processed, and the like. The details of the artificial intelligence module 110 are described in detail later.

The operation configuration module 120 is configured to configure the calculation force according to the model constructed by the user through the artificial intelligence module 110. The details of the operation configuration module 120 will be described later.

The satellite communication module 140 is used for communicating with satellites 201 to 203. Thereby communicating with the ground stations 301-303 via satellites 201-203.

Fig. 3B also shows a schematic diagram of a server side 300 deployed by the respective ground stations. Referring to fig. 3B, the server side 300 includes a satellite communication module 310, an information parsing module 320, an artificial intelligence processing module 330, and a ground station communication module 340. In addition, the server 300 is further configured with a plurality of artificial intelligence model units 1 to n. The artificial intelligence model elements may be, for example, program segments to implement the functions of various artificial intelligence model elements such as a convolution layer component, an activation function, a pooling layer component, a full connection layer component, a softmax classifier component, a transformer layer component, an LSTM component, and a CRF component. And the ground station communication module 340 is used to interact with other ground stations.

In the above-described operating environment, according to a first aspect of the present embodiment, there is provided an artificial intelligence method based on a satellite communication distributed computing system. Fig. 4 shows a flow chart of the method, with reference to fig. 4, comprising:

s402: the terminal equipment responds to the operation of constructing the artificial intelligent model by a user and generates model information corresponding to the artificial intelligent model;

s404: the terminal equipment responds to operation of operation configuration between all the ground stations by a user, the artificial intelligent model unit is divided into a plurality of artificial intelligent model units, corresponding operation configuration information is generated, the ground stations interact with the terminal equipment through satellites, and the operation configuration information indicates distribution of the artificial intelligent model units among all the ground stations;

s406: the terminal equipment sends the operation configuration information to the ground station through the satellite; and

s408: and the ground station constructs an artificial intelligent model unit corresponding to the operation configuration information according to the received operation configuration information.

Specifically, fig. 5A shows a schematic diagram of an interface displayed by the terminal device 10 when the user starts the client 100. Referring to fig. 5A, when a user activates the client 100 at the terminal device 10, the user interface module 130 pops up a navigation interface to prompt the user to select "build model" or "import model".

In addition, referring to FIG. 5B, when the user selects "build model," the user interface module 130 pops up a navigation interface to further prompt the user to select the type of build model. For example, when a user clicks "CNN", an interface for constructing a CNN model is entered, when a user clicks "NLP", an interface for constructing an NLP model is entered, and when a user clicks "R-CNN", an interface for constructing an R-CNN model is entered. And, the user can click on 'custom', so that the artificial intelligence model can be built according to own ideas. And accordingly, the artificial intelligence module 110 will create a data structure corresponding to the artificial intelligence model.

FIG. 5C illustrates a schematic diagram of an interface 500 for a user to build and edit an artificial intelligence model. Referring to FIG. 5C, the interface includes a workspace 510 and a component bar 520. The component column 520 includes a plurality of components 521. Component 521 includes the components required for building an artificial intelligence model. For example, the assembly 521 includes: a convolution layer component, an activation function, a pooling layer component, a full connection layer component, a softmax classifier component, a transformer layer component, an LSTM component, and a CRF component. And the components 521 also include a connection line component or the like for constraining the association between the different artificial intelligence components.

In addition, referring to fig. 5A, when a user imports an artificial intelligence model that has been constructed by clicking on "import model", the interface 500 shown in fig. 5C is also entered, thereby editing the imported artificial intelligence model. This description is not repeated.

Specifically, the user may drag icons of the various components 521 to the workspace 510 by way of a drag, thereby constructing an artificial intelligence model.

For example, when a user drags the convolution layer component to the workspace 510, the user interface module 130 pops up an interface for the user to input the number of convolution kernels contained in the convolution layer and the dimensions of the convolution kernels. Thus, not only is the graph of the convolution layer displayed in the workspace 510, but the artificial intelligence module 110 also generates component information corresponding to the convolution kernel, e.g., for a file containing 50 3A convolution layer of 3 convolution kernels, which may be 3 +.>3/>And 50 denotes component information of the convolutional layer.

For example, when a user drags a component of an activation function to the workspace 510, the user interface module 130 pops up an interface for the user to select the corresponding activation function. The workspace 510 thus displays not only the graphic of the activation function, but also the component information corresponding to the activation function by the artificial intelligence module 110.

Similarly, when the user drags the component to the work area 510, a graphic corresponding to the component is displayed in the work area 510, and the artificial intelligence module 110 also generates component information corresponding to the component.

In addition, the user may also build a link in the workspace 510 by dragging, thereby determining the relevance between the various components. For example, a user may associate an input layer with a convolutional layer back and forth by a join line, or may associate a convolutional layer with a pooled layer by a join line. And accordingly, the artificial intelligence module 110 determines association information between the various components 521 based on the user's wiring.

Thus, when the user completes the construction of the artificial intelligence model, the artificial intelligence module 110 also determines model information of the corresponding artificial intelligence model, which includes component information of the respective components 521 and association information between the respective components 521. For example, FIG. 6 shows model information based on deep learning. Since model information expressing an artificial intelligence model is common knowledge in the art, the model information may be constructed in any form that can be understood by those skilled in the art. The present embodiment is not described in detail.

Thus, referring to FIG. 5D, the user has completed the construction of the artificial intelligence model at interface 500. Referring to FIG. 5E, the artificial intelligence model includes: two branch networks for extracting features, which are used for extracting features from two input images respectively; and a feature extraction and classification network for combining the feature graphs of the two branch networks to extract features and classify the features.

And accordingly, after the user builds the artificial intelligence model, the artificial intelligence module 110 synchronously completes the construction of the corresponding model information (S402).

Then, referring to FIG. 5D, after the artificial intelligence model is built, the user may click on the "operational configuration" button. Thus, the user interface module 130 pops up the operation configuration interface 600 for operation configuration, and displays it on the terminal device 10.

Specifically, fig. 7A shows a schematic diagram of an operational configuration interface 600. Referring to FIG. 7A, a schematic diagram of a user-constructed artificial intelligence model is shown in a workspace 610.

Referring then to FIG. 7B, the user may divide the artificial intelligence model into a plurality of different units by framing on the interface 600 to load the artificial intelligence model to the different ground stations 301-303, respectively.

Referring to FIG. 7C, the artificial intelligence model 530 is divided into 3 model elements: the first model unit 531, the second model unit 532, and the third model unit 533. Thus, the artificial intelligence module generates model element information corresponding to each model element 531 to 533 based on the model information of the artificial intelligence model 530 according to the division of the user. Specifically, the model unit information may be obtained by dividing the model information according to the division of the model units 531 to 533, for example, which will not be described herein.

Then, with continued reference to FIG. 7B, the operational configuration interface 600 includes a ground station identification field 620. Referring to fig. 7B, a ground station identifier column 620 displays a plurality of ground station identifiers 621 to 623, respectively representing the respective ground stations 301 to 303 interacting with the terminal device 10 through satellites 201 to 203. Referring to fig. 1, since the ground stations with which the terminal apparatus 10 interacts include the ground stations 301 to 303, the ground station identification 621 corresponds to the ground station 301, the ground station identification 622 corresponds to the ground station 302, and the ground station identification 623 corresponds to the ground station 303.

Thus, referring to FIG. 7D, a user may associate different ground stations with different model elements by dragging the ground station identification. For example, the user associates the first model cell 531 with the ground station 301 by dragging the ground station identification 621 to the first model cell 531; associating the second model unit 532 with the ground station 302 by dragging the ground station identification 622 to the second model unit 532; and associating the third model unit 533 with the ground station 303 by dragging the ground station identification 623 to the third model unit 533.

Thus, in this way, the user completes the operational configuration of the constructed artificial intelligent model between the ground stations 301-303 through the operational configuration interface.

And, the operation configuration module 120 generates corresponding operation configuration information according to the operation configuration while the user completes the operation configuration. The operation configuration information includes information of artificial intelligence model units distributed to different ground stations 301-303. Information indicating model elements loaded to different ground stations 301 to 303 (S404). Specifically, the operation configuration information may be, for example, as shown in the following table:

TABLE 1

Then, the terminal device 10 transmits the operation configuration information to the ground stations 301 to 303 through the satellites 201 to 203 (S406). Thus, each ground station 301 to 303 may load each artificial intelligence model unit according to the operation configuration information (S408).

Specifically, after the ground station 301 obtains the operation configuration information from the satellites 201 to 203 through the satellite communication module 310, the information analysis module 320 analyzes the information of the first model unit 531 from the operation configuration information, and determines the third model unit 533 of the ground station 303 under the first model unit 531. Then, the artificial intelligence processing module 330 obtains corresponding artificial intelligence model units from the pre-deployed artificial intelligence model units 1 to n according to the information of the first model unit 531, and constructs the first model unit 531.

In addition, after the ground station 302 obtains the operation configuration information from the satellites 201 to 203 through the satellite communication module 310, the information analysis module 320 analyzes the information of the second model unit 532 from the operation configuration information, and determines the third model unit 533 of the ground station 303 under the second model unit 532. Then, the artificial intelligence processing module 330 obtains corresponding artificial intelligence model units from the pre-deployed artificial intelligence model units 1 to n according to the information of the second model unit 532, and constructs the second model unit 532.

After the ground station 303 obtains the operation configuration information from the satellites 201 to 203 through the satellite communication module 310, the information analysis module 320 analyzes the information of the third model unit 533 from the operation configuration information, and determines the first model unit 301 of the ground station 301 and the second model unit 532 of the ground station 302 on the third model unit 533. Then, the artificial intelligence processing module 330 obtains corresponding artificial intelligence model units from the pre-deployed artificial intelligence model units 1 to n according to the information of the third model unit 533, and constructs the third model unit 533.

As described in the background art, when a user cannot communicate with the internet due to being in a remote area and thus cannot communicate with a cloud platform providing an artificial intelligence service, the artificial intelligence service cannot be obtained.

In view of this, according to an embodiment of the present application, there is provided a satellite-based distributed computing system that implements distributed computing using ground stations that communicate with terminal devices through satellites. Therefore, compared with the operation by using satellites, the ground station is arranged on the ground, so that the ground station can bear the electric energy loss caused by large operation amount, and the ground station can provide energy assurance for artificial intelligent service. Furthermore, according to embodiments of the present application, an artificial intelligence model constructed by a user may be divided into a plurality of different model units, and each model unit may be associated with a different ground station, respectively, such that the different model units of the artificial intelligence model are loaded at the different ground stations. The advantage of the number of ground stations can be utilized to provide the operational resources required by the artificial intelligence model. Thus, by the above method, the artificial intelligence service can be provided for users which are located in remote areas and cannot establish connection with the Internet. Therefore, the technical problem that users in remote areas cannot be connected with the Internet and cannot conveniently obtain artificial intelligence services in the prior art is solved.

Optionally, the operation of generating the model information corresponding to the artificial intelligence model by the terminal device in response to the operation of constructing the artificial intelligence model by the user includes: the terminal equipment displays a first graphical user interface for constructing and editing the artificial intelligent model; and the terminal equipment responds to the operation of constructing the artificial intelligent model on the first graphical user interface by a user and generates model information corresponding to the artificial intelligent model.

In particular, as shown in fig. 5A-5E and 7C, the client 100 deployed by the terminal device 10 may display an interface 500 (i.e., a first graphical user interface) for building and editing artificial intelligence models. The client 100 of the terminal apparatus 10 then generates model information corresponding to the artificial intelligence model 530 in response to an operation of the user constructing the artificial intelligence model 530 at the interface 500.

Thus, according to the embodiment, the user can construct and edit the artificial intelligence model in the graphical user interface, thereby facilitating the use of the user.

Optionally, the operation of the terminal device in response to the operation of performing operational configuration between the ground stations by the user, dividing the artificial intelligence model unit into a plurality of artificial intelligence model units, and generating corresponding operational configuration information includes: the terminal equipment displays a second graphical user interface for operation configuration; the terminal equipment responds to the operation of dividing the artificial intelligent model into a plurality of model units in a second graphical user interface by a user, and generates model unit information corresponding to the model units; and the terminal equipment responds to the configuration operation of associating the divided model units with the corresponding ground stations at the second graphical user interface by a user, and generates operation configuration information.

Specifically, referring to fig. 7A to 7D, the client 100 of the terminal device 10 may display the operation configuration interface 600 (i.e., the second graphical user interface), so that the user may divide the artificial intelligence model 530 into a plurality of different model units 531 to 533 in the operation configuration interface by selecting a frame, and the artificial intelligence module 110 of the client 100 generates corresponding model unit information. Then, when the user associates each model unit 531 to 533 with the corresponding ground station in a dragging manner, the configuration of the computing resource is completed for the artificial intelligence model 530. Accordingly, the artificial intelligence module 110 generates corresponding operational configuration information.

Thus, according to this embodiment, a user may divide a constructed or edited artificial intelligence model into a plurality of different model elements and associate each model element with a corresponding ground station so that the functionality of the artificial intelligence model may be implemented together by the plurality of ground stations. Therefore, the user can fully utilize the operation resources communicated with the terminal equipment 10 through the satellites 201-203 to realize the operation related to the constructed artificial intelligent model.

Optionally, the method further comprises: the terminal equipment acquires operation parameters corresponding to each ground station from the satellite; the terminal equipment utilizes a preset neural network model, and determines a target ground station capable of bearing an operation task related to the artificial intelligent model according to the operation parameters; and displaying an identification associated with the target ground station on a second graphical user interface.

Specifically, the terminal device 10 may acquire, in real time, operation parameters of each ground station 201 to 203 corresponding to the communication area through the satellites 201 to 203. Wherein the operating parameters comprise, for example, the individual ground stations 301 to 303 arithmetic processing capabilityR ₁ ~R ₃ Data transmission rate between each ground station 301-303 and terminal device 10V ₁ ~V ₃ And communication noise between each of the ground stations 301 to 303 and the terminal device 10S ₁ ~S ₃ 。

Then, the terminal device 10 inputs the operation parameters corresponding to the ground stations 301 to 303 to the neural network model, and obtains the priority probabilities corresponding to the ground stations 301 to 303. Fig. 8 is a schematic diagram of a neural network for determining priority probabilities corresponding to ground stations 301-303 according to an embodiment of the present application. Referring to fig. 8, the neural network is provided with an input layer, a hidden layer, an output layer, and a softmax classification layer.

Thus, the terminal device 10 determines the operation parameters corresponding to the ground stations 301 to 303. For example, the arithmetic processing capability of the ground station 301R ₁ Data transfer rateV ₁ Communication noiseS ₁ The method comprises the steps of carrying out a first treatment on the surface of the Operational processing capabilities of ground station 302R ₂ Data transfer rateV ₂ Communication noiseS ₂ The method comprises the steps of carrying out a first treatment on the surface of the Arithmetic processing capability of ground station 303R ₃ Data transfer rateV ₃ Communication noise S ₃ . Wherein the operational parameters corresponding to ground station 301, the operational parameters corresponding to ground station 302, and the operational parameters corresponding to ground station 303 may form a vector matrixA ₁ (i.e., vector matrix in FIG. 8)A ₁ ）。

Then, the terminal device 10 matrices the vectorA ₁ Input to the neural network model, thereby outputting a priority probability corresponding to the ground station 301 asP ₁ The priority probability corresponding to the ground station 302 isP ₂ And the priority probability corresponding to the ground station 303 isP ₃ . For example, the priority probability corresponding to the ground station 301 is 35% and the priority probability corresponding to the ground station 302 is 30%The priority probability corresponding to the ground station 303 is 35%.

Further, the terminal device 10 determines whether the priority probability of each ground station is greater than a preset priority probability thresholdP _k . The preset priority probability threshold may be 25%, for example.

Determining the ground station as a target ground station in the case that the priority probability corresponding to the ground station is greater than a preset priority probability threshold; and eliminating the ground station under the condition that the priority probability threshold value corresponding to the ground station is smaller than or equal to the preset priority probability threshold value. Thus, the terminal device 10 displays the ground station identifications 621 to 623 corresponding to the respective ground stations 301 to 303 in the ground station column 620.

Therefore, according to the technical scheme of the embodiment, the ground station with the priority probability larger than the preset threshold value is used as the target ground station according to the operation parameters of each ground station. Therefore, the ground station with poor running state is prevented from being associated with the artificial intelligence model by the user, and the quality of artificial intelligence service provided for the user is ensured.

Optionally, the method further comprises: the terminal equipment responds to triggering operation of training the artificial intelligent model by a user, and a training instruction for training the artificial intelligent model is sent to the ground station through a satellite; the ground station acquires a training sample for training the artificial intelligent model according to the training instruction; the ground station trains the artificial intelligence model with training samples.

Specifically, referring to FIG. 7D, after loading the artificial intelligence model 530 into the ground stations 301-303, the user may click "model training" to train the artificial intelligence model 530.

Specifically, the client 100 of the terminal device 10 sends the training instruction to the ground stations 301 to 303 through the satellite communication module 140.

The ground stations 301-303 thus obtain a sample set from the database 400 for training the artificial intelligence model 530. For example, the sample set is shown in table 2 below:

TABLE 2

Sample number	Input image 1	Input image 2	Output vector
				1	Imga1	Imgb1	V1
2	Imga2	Imgb2	V2
				3	Imga3	Imgb3	V3
...	...	...	...
				4	Imgan	Imgbn	Vn

Wherein input image 1 is used for the input model unit 531 and input image 2 is used for the input model unit 532. The output vector is used to compare with the vector output by the model unit 533 and calculate the loss according to a preset loss function.

Thus, during training of artificial intelligence model 530, ground station 301 obtains a sample set from database 400 and will sample Ima corresponding to input image 1 ₁ ~Ima _n Is input to the first model unit 531. Thereby generating a sample Ima ₁ ~Ima _n Corresponding feature maps and transmits the generated feature maps to the ground station 303.

The ground station 302 obtains the sample set from the database 400 and will correspond to the sample Imb of the input image 2 ₁ ~Imb _n Is input to the second model unit 532. Thereby generating an and sample Imb ₁ ~Imb _n Corresponding feature maps and transmits the generated feature maps to the ground station 303.

The ground station 303 inputs the feature maps received from the ground stations 301 and 302 to the third model unit 533. So that the third model unit 533 outputs the corresponding class vector. And the ground station 303 further compares the outputted class vector with the vector of the corresponding sample and calculates the loss using a preset loss function.

The ground station 303 then adjusts the parameters of the model unit 533 by back propagation and then transmits the adjusted parameters to the ground station 301 and the ground station 302. Thus, ground station 301 and ground station 302 continue back propagation based on the received parameters, thereby completing training of the artificial intelligence model.

Wherein figure 9 shows a flow chart for training an artificial intelligence model.

Optionally, the method further comprises: the terminal equipment responds to the triggering operation of determining the data object to be processed by the user, and determines the data object to be processed; the terminal equipment sends the data object to be processed to the ground station through the satellite; and the ground station processes the data object to be processed by using the artificial intelligent model.

Specifically, for example, the user may select the data object to be processed through the client 100 of the terminal device 10, for example, in this embodiment, the user selects the first image and the second image at the terminal device 10, so that the client 100 transmits the first image and the second image to the ground stations 301 to 303 through the satellites 201 to 203 through the satellite communication module 140.

Then, the ground station 301 inputs the first image to the first model unit 531 to generate a corresponding feature map, and transmits the generated feature map to the ground station 303; the ground station 302 inputs the second image to the second model unit 532 to generate a corresponding feature map, and transmits the generated feature map to the ground station 303.

The ground station 303 inputs the feature maps received from the ground station 301 and the ground station 302 into the third model unit 533 to obtain corresponding class vectors, and transmits the class vectors to the terminal device 10 through the satellites 201 to 203 via the satellite communication module 310, so that the class vectors can be displayed on the client 100 of the terminal device 10.

Wherein figure 10 shows a flow chart for processing a data object using an artificial intelligence model.

Further, according to a second aspect of the present embodiment, there is provided a storage medium. The storage medium includes a stored program, wherein the method of any one of the above is performed by a processor when the program is run.

According to an embodiment of the present application, there is provided a satellite-based distributed computing system that implements distributed computing using ground stations that communicate with terminal devices through satellites. Therefore, compared with the operation by using satellites, the ground station is arranged on the ground, so that the ground station can bear the electric energy loss caused by large operation amount, and the ground station can provide energy assurance for artificial intelligent service. Furthermore, according to embodiments of the present application, an artificial intelligence model constructed by a user may be divided into a plurality of different model units, and each model unit may be associated with a different ground station, respectively, such that the different model units of the artificial intelligence model are loaded at the different ground stations. The advantage of the number of ground stations can be utilized to provide the operational resources required by the artificial intelligence model. Thus, by the above method, the artificial intelligence service can be provided for users which are located in remote areas and cannot establish connection with the Internet. Therefore, the technical problem that users in remote areas cannot be connected with the Internet and cannot conveniently obtain artificial intelligence services in the prior art is solved.

It should be noted that, for simplicity of description, the foregoing method embodiments are all described as a series of acts, but it should be understood by those skilled in the art that the present invention is not limited by the order of acts described, as some steps may be performed in other orders or concurrently in accordance with the present invention. Further, those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily required for the present invention.

From the description of the above embodiments, it will be clear to a person skilled in the art that the method according to the above embodiments may be implemented by means of software plus the necessary general hardware platform, but of course also by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to perform the method according to the embodiments of the present invention.

Example 2

Fig. 11 shows an artificial intelligence device 1100 based on a satellite communication distributed computing system according to the present embodiment, the device 1100 corresponding to the method according to the first aspect of embodiment 1. As shown in reference 11, the apparatus 1100 includes: a model information generating module 1110, configured to generate model information corresponding to an artificial intelligence model in response to an operation of constructing the artificial intelligence model by a user through a terminal device; the operation configuration information generating module 1120 is configured to respond to an operation of performing operation configuration between each ground station by a user through the terminal device, divide the artificial intelligent model unit into a plurality of artificial intelligent model units, and generate corresponding operation configuration information, where the ground station interacts with the terminal device through a satellite, and the operation configuration information indicates allocation of the artificial intelligent model unit between each ground station; the configuration information sending module 1130 is configured to send the operation configuration information to the ground station through the satellite by using the terminal device; and a model element construction module 1140, configured to construct, by the ground station, an artificial intelligence model element corresponding to the operational configuration information according to the received operational configuration information.

Optionally, the model information generating module 1110 includes: a first graphical user interface sub-module for displaying a first graphical user interface for constructing and editing an artificial intelligence model through the terminal device; and the model information generation sub-module is used for responding to the operation of constructing the artificial intelligent model in the first graphical user interface by the user through the terminal equipment and generating model information corresponding to the artificial intelligent model.

Optionally, the operation configuration information generating module 1120 includes: the second graphic user interface sub-module is used for displaying a second graphic user interface for operation configuration through the terminal equipment; the model unit information generation sub-module is used for responding to the operation of dividing the artificial intelligent model into a plurality of model units in the second graphical user interface by the user through the terminal equipment and generating model unit information corresponding to the model units; and the operation configuration information generation sub-module is used for responding to the configuration operation of the user for associating the divided model units with the corresponding ground stations in the second graphical user interface through the terminal equipment to generate operation configuration information.

Optionally, the apparatus 1100 further comprises: the operation parameter acquisition module is used for acquiring operation parameters corresponding to each ground station from the satellite through the terminal equipment; the target ground station determining module is used for determining a target ground station capable of bearing an operation task related to the artificial intelligent model according to the operation parameters by utilizing a preset neural network model through the terminal equipment; and a second graphical user interface generation module for displaying, by the terminal device, an identification associated with the target ground station on the second graphical user interface.

Optionally, the apparatus 1100 further comprises: the training instruction transmission module is used for responding to the triggering operation of training the artificial intelligent model by the user through the terminal equipment and transmitting the training instruction for training the artificial intelligent model to the ground station through the satellite; the sample acquisition module is used for acquiring training samples for training the artificial intelligent model according to the training instructions through the ground station; and the training module is used for training the artificial intelligent model by using the training sample through the ground station.

Optionally, the apparatus 1100 further comprises: the to-be-processed data object determining module is used for determining the to-be-processed data object through the terminal equipment in response to the triggering operation of determining the to-be-processed data object by the user; the data object sending module is used for sending the data object to be processed to the ground station through the satellite by the terminal equipment; and the data processing module is used for processing the data object to be processed by using the artificial intelligent model through the ground station.

The foregoing embodiment numbers of the present application are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.

In the foregoing embodiments of the present application, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.

In the several embodiments provided in the present application, it should be understood that the disclosed technology may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and the division of the units, such as the division of the units, is merely a logical function division, and may be implemented in another manner, for example, multiple units or components may be combined or may be 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 through some interfaces, units or modules, or may be in electrical 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 on 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 invention 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 integrated units may be implemented in hardware or in software functional units.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.

The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which are intended to be comprehended within the scope of the present invention.

Claims

1. An artificial intelligence method based on a satellite communication distributed computing system, which is characterized by comprising the following steps:

the method comprises the steps that a terminal device responds to the operation of constructing an artificial intelligent model by a user, and model information corresponding to the artificial intelligent model is generated;

the terminal equipment responds to operation of operation configuration between all ground stations by a user, the artificial intelligent model is divided into a plurality of artificial intelligent model units, corresponding operation configuration information is generated, the ground stations interact with the terminal equipment through satellites, and the operation configuration information indicates distribution of the artificial intelligent model units among all the ground stations;

the terminal equipment sends the operation configuration information to the ground station through a satellite;

the ground station constructs an artificial intelligent model unit corresponding to the operation configuration information according to the received operation configuration information, wherein

Further comprises:

the terminal equipment acquires operation parameters corresponding to each ground station from the satellite;

the terminal equipment utilizes a preset neural network model, and determines a target ground station capable of bearing an operation task related to the artificial intelligent model according to the operation parameters; and

an identification associated with the target ground station is displayed on a second graphical user interface for operational configuration.

2. The method of claim 1, wherein the operation of the terminal device generating model information corresponding to the artificial intelligence model in response to a user operation of constructing the artificial intelligence model comprises:

the terminal equipment displays a first graphical user interface for constructing and editing the artificial intelligent model; and

and the terminal equipment responds to the operation of constructing the artificial intelligent model on the first graphical user interface by a user, and generates model information corresponding to the artificial intelligent model.

3. The method of claim 2, wherein the operation of the terminal device dividing the artificial intelligence model into a plurality of artificial intelligence model elements and generating corresponding operational configuration information in response to a user's operation of operational configuration between the respective ground stations comprises:

The terminal equipment displays the second graphical user interface;

the terminal equipment responds to the operation of dividing the artificial intelligent model into a plurality of artificial intelligent model units in the second graphical user interface by a user, and generates model unit information corresponding to the artificial intelligent model units; and

the terminal equipment responds to the configuration operation of associating the divided model units with the corresponding ground stations at the second graphical user interface by a user, and generates operation configuration information.

4. The method as recited in claim 1, further comprising:

the terminal equipment responds to triggering operation of training the artificial intelligent model by a user, and a training instruction for training the artificial intelligent model is sent to the ground station through the satellite;

the ground station acquires a training sample for training the artificial intelligent model according to the training instruction; and

the ground station trains the artificial intelligence model by using the training samples.

5. The method as recited in claim 1, further comprising:

the method comprises the steps that terminal equipment responds to trigger operation of determining a data object to be processed by a user, and the data object to be processed is determined;

The terminal equipment sends the data object to be processed to the ground station through the satellite; and

and the ground station processes the data object to be processed by using the artificial intelligence model.

6. A storage medium comprising a stored program, wherein the method of any one of claims 1 to 5 is performed by a processor when the program is run.

7. An artificial intelligence device based on a satellite communication distributed computing system, comprising:

the system comprises a model information generation module, a model information generation module and a control module, wherein the model information generation module is used for responding to the operation of constructing an artificial intelligent model by a user through terminal equipment and generating model information corresponding to the artificial intelligent model;

the system comprises an operation configuration information generation module, a control module and a control module, wherein the operation configuration information generation module is used for responding to operation of operation configuration between each ground station by a user through terminal equipment, dividing the artificial intelligent model into a plurality of artificial intelligent model units and generating corresponding operation configuration information, wherein the ground station interacts with the terminal equipment through satellites, and the operation configuration information indicates distribution of the artificial intelligent model units among the ground stations;

The configuration information sending module is used for sending the operation configuration information to the ground station through a satellite by the terminal equipment;

a model unit construction module for constructing an artificial intelligent model unit corresponding to the operation configuration information according to the received operation configuration information through a ground station, wherein

The apparatus further comprises:

the operation parameter acquisition module is used for acquiring operation parameters corresponding to each ground station from the satellite through the terminal equipment;

the target ground station determining module is used for determining a target ground station capable of bearing an operation task related to the artificial intelligent model according to the operation parameters by utilizing a preset neural network model through the terminal equipment; and

and the second graphical user interface generation module is used for displaying the identification associated with the target ground station on a second graphical user interface for operation configuration through the terminal equipment.

8. The apparatus of claim 7, wherein the model information generation module comprises:

a first graphical user interface sub-module for displaying a first graphical user interface for constructing and editing the artificial intelligence model through a terminal device; and

and the model information generation sub-module is used for responding to the operation of constructing the artificial intelligent model on the first graphical user interface by a user through the terminal equipment and generating model information corresponding to the artificial intelligent model.

9. The apparatus of claim 8, wherein the operational configuration information generation module comprises:

the second graphic user interface sub-module is used for displaying a second graphic user interface for operation configuration through the terminal equipment;

a model unit information generating sub-module, configured to generate model unit information corresponding to the artificial intelligent model units by using a terminal device in response to an operation of dividing the artificial intelligent model into the plurality of artificial intelligent model units at the second graphical user interface by a user; and

and the operation configuration information generation sub-module is used for responding to the configuration operation of associating the divided model units with the corresponding ground stations at the second graphical user interface by a terminal device to generate the operation configuration information.