Data processing method and system, electronic equipment and readable storage medium

Abstract

The invention discloses a data processing method and system, electronic equipment and a readable storage medium, and belongs to the field of data processing. The data processing method comprises the steps that AI model training data in a target area is sent to a cloud platform through a target slice network corresponding to the target area, so that the cloud platform trains an AI model based on the AI model training data to obtain a trained target AI model, and the cloud platform is deployed outside the target area; receiving the target AI model issued by the cloud platform through the target slice network; and processing target data collected in the target area based on the target AI model.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of communications, and in particular, to a data processing method, system, electronic device and readable storage medium. Background Art

[0002] With the development of artificial intelligence (AI) technology, AI technology has been widely used in some target areas, such as industrial production parks, where product quality inspection can be performed based on image classification technology. The deep learning algorithms involved in AI technology, such as convolutional neural networks (CNN), recurrent neural networks (RNN), and self-attention, require large model training and consume a lot of memory.

[0003] When deploying AI technology in industrial production scenarios, data security needs to be ensured. However, related technologies cannot ensure accurate training of AI models while ensuring data security. Therefore, it is impossible to accurately perform AI operations on data in the target area. Summary of the invention

[0004] The embodiments of the present application provide a data processing method, system, electronic device and readable storage medium, which can solve the problem that it is impossible to ensure accurate training of AI models while ensuring data security, resulting in the inability to accurately perform AI operations on data in the target area.

[0005] In a first aspect, a data processing method is provided, which is applied to a first edge computing device, and the method includes: sending AI model training data in the target area to a cloud platform through a target slice network corresponding to the target area, so that the cloud platform trains an AI model based on the AI ​​model training data to obtain a trained target AI model, and the cloud platform is deployed outside the target area; receiving the target AI model issued by the cloud platform through the target slice network; and processing the target data collected in the target area based on the target AI model.

[0006] In a second aspect, a data processing system is provided, including: a first edge computing device, which sends the AI ​​model training data in the target area to the cloud platform through a target slice network corresponding to the target area; the cloud platform is deployed outside the target area and connected to the first edge computing device, for receiving the AI ​​model training data, and training the AI ​​model based on the AI ​​model training data to obtain the trained target AI model, and sending the target AI model to the first edge computing device; the first edge computing device is also used to receive the target AI model sent by the cloud platform through the target slice network, and process the target data collected in the target area based on the target AI model.

[0007] According to a third aspect, an electronic device is provided, the terminal comprising a processor and a memory, the memory storing programs or instructions executable on the processor, the programs or instructions implementing the steps of the method described in the first aspect when executed by the processor.

[0008] In a fourth aspect, a readable storage medium is provided, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0009] In a fifth aspect, a chip is provided, comprising a processor and a communication card slot, wherein the communication card slot is coupled to the processor, and the processor is used to run a program or instruction to implement the steps of the method described in the first aspect.

[0010] In a sixth aspect, a computer program / program product is provided, wherein the computer program / program product is stored in a storage medium and is executed by at least one processor to implement the steps of the method described in the first aspect.

[0011] In an embodiment of the present application, the AI ​​model training data within the target area is sent to the cloud platform through a target slice network corresponding to the target area, so that the cloud platform trains the AI ​​model based on the AI ​​model training data to obtain a trained target AI model, and the cloud platform is deployed outside the target area; the target AI model issued by the cloud platform is received through the target slice network; the target data collected in the target area is processed based on the target AI model, which can ensure accurate training of the AI ​​model while ensuring data security, and then accurately perform AI operations on the data in the target area. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is a schematic flow chart of a data processing method according to an embodiment of the present invention;

[0013] Figure 2 is a schematic flow chart of a data processing method according to another embodiment of the present invention;

[0014] Figure 3 is a schematic diagram of the structure of a data processing system according to an embodiment of the present invention;

[0015] Figure 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of this application.

[0017] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way are interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first" and "second" are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally represents that the objects associated with each other are in an "or" relationship.

[0018] like Figure 1 As shown, an embodiment of the present invention provides a data processing method, which can be executed by software or hardware of a first edge computing device, and the method includes the following steps:

[0019] S102: Send the AI ​​model training data in the target area to the cloud platform through the target slice network corresponding to the target area.

[0020] The cloud platform trains the AI ​​model based on the AI ​​model training data to obtain a trained target AI model, and the cloud platform is deployed outside the target area. The first edge computing device is deployed in the target area.

[0021] The embodiment of the present application can provide edge computing for AI source data in the target area. The target area can be set according to the needs, for example, it can include a factory that needs to perform edge computing as described in the embodiment of the present application. Figure 3As shown, an AI source data acquisition device 1 can also be deployed in the target area. The AI ​​source data acquisition device 1 can include, for example, a camera for acquiring images, a scanner for acquiring QR codes, and a recorder for acquiring sounds.

[0022] The first edge computing device is used to provide Hyper-Edge Computing (HEC) services. It can be an AI computing power edge gateway device with access to the target slice network, and can perform image recognition, AI reasoning, AI model deployment, reasoning acceleration and other related application functions. The first edge computing device can also provide open capabilities, such as reporting network resources, computing power resources, AI reasoning results and other information, and can also provide production data on demand for AI model training. The first edge computing device can reuse the packet transmission equipment deployed in the target area.

[0023] The cloud platform trains the AI ​​model based on the AI ​​model training data to obtain a trained target AI model, and saves the target AI model in the cloud platform storage. Optionally, the method of training the model can be determined based on the amount of training data and other factors. The method of training the AI ​​model on the cloud platform may include one of the following:

[0024] Based on the AI ​​model training data, incremental training is performed on the existing model. Optionally, if the requirements of the target area are close to the functions of the existing relevant AI model, incremental training can be performed based on the existing relevant AI model, thereby speeding up the AI ​​model training speed and reducing the production line sampling time.

[0025] Based on the AI ​​model training data, the parameters of the pre-trained model are adjusted. Optionally, if the requirements of the target area are not close to the existing relevant AI model and only a small number of samples can be provided, the parameters of the pre-trained model can be adjusted to meet the rapid learning requirements of the small sample scenario and reduce the sampling time of the production line.

[0026] Based on the AI ​​model training data, full training is performed. Optionally, if the needs of the target area are not close to the existing relevant AI models, but sufficient samples can be provided, a new model can be created for full training to obtain an independent model for the specific needs of the target area.

[0027] S104: Receive the target AI model sent by the cloud platform through the target slicing network.

[0028] Combination Figure 3As shown, in a possible implementation, in this step S102, the first edge computing device 31 can send the AI ​​model training data collected by the application data collection terminal 1 in the target area to the management device 33 deployed outside the target area through the target slice network, so that the management device 33 can send the AI ​​model training data to the cloud platform 34.

[0029] The management device 33 reads the target AI model from the cloud platform storage based on the model tag, and sends the target AI model to the first edge computing device 31 through the target slice network. The first edge computing device 31 receives the target AI model sent by the management device through the target slice network.

[0030] In one implementation, when the target slice network is connected to the cloud platform in the north-south direction, the Metro Transport Network (MTN) hard isolation technology or the Fine-Grained Unit (FGU) end-to-end hard isolation channel can be used. When the target slice network is connected to the cloud platform in the east-west direction, the above can be shared with the north-south connection, and the core layer node can be used for detour. For the east-west computing power connection with a certain latency requirement, the MTN / FGU Channel can be used according to the lowest latency path to achieve direct connection between the target slice network access nodes.

[0031] S106: Processing the target data collected in the target area based on the target AI model.

[0032] In the embodiment of the present application, the network operator providing the network for the target area can set up a cloud platform, which can provide cloud storage and AI training functions, and can also provide labeling services. For example, if the target area is a factory, the industrial user of the factory can perform production control through the HEC and the workbench.

[0033] In an embodiment of the present application, based on the properties of the slice network, data isolation can be achieved between a target slice network corresponding to a target area and a target slice network corresponding to another target area, thereby ensuring data security, and sending the AI ​​model training data in the target area to the cloud platform through the target slice network corresponding to the target area, so that the cloud platform can train the AI ​​model based on the AI ​​model training data to obtain a trained target AI model, and the cloud platform is deployed outside the target area; receiving the target AI model issued by the cloud platform through the target slice network; processing the target data collected in the target area based on the target AI model, and sending the AI ​​model training data in the target area to the cloud platform, realizing accurate training of the target AI model through the cloud platform, thereby ensuring that the target data collected in the target area can be processed based on the accurate target AI model. Therefore, the embodiment of the present application can ensure accurate training of the AI ​​model while ensuring data security, and then accurately perform accurate AI operations on the data in the target area.

[0034] In one implementation, S106 may include at least one of the following steps:

[0035] Based on the AI ​​model, the target data is AI operated to obtain an AI operation result, and the AI ​​operation result is sent to the AI ​​operation result response device. The response device is deployed in the target area. The AI ​​operation result response device includes: the AI ​​source data acquisition device or a third device running an AI application, and the AI ​​operation result is processed by the result processing end of the AI ​​application software. The AI ​​operation result response device can be the same device as the AI ​​source data acquisition device, or it can be a different device. For example, the AI ​​source data acquisition device can be a terminal, which collects user data and sends it to the first edge computing device 31 for AI operation. The first edge computing device 31 can send the AI ​​operation result to the terminal, and the terminal performs subsequent data processing. For another example, the AI ​​source data acquisition device can be a camera, which collects the user's facial image and sends it to the edge computing device for AI image recognition. The edge computing device can send the AI ​​operation result to a terminal, and the terminal performs subsequent data processing on the image recognition result.

[0036] The target AI model is sent to each second edge computing device through the target slicing network, so that each second edge computing device can perform AI operations on the target data based on the AI ​​model; each second edge computing device is deployed in the target area. Therefore, in the embodiment of the present application, when there are multiple edge computing devices in the target area, the target AI model can be received by the first edge computing device among the multiple edge computing devices. The AI ​​operation result is obtained by performing AI operations on the target data based on the AI ​​model through the second edge computing device. Thus, accurate AI operations are performed on the data in the target area and computing power sharing of multiple edge computing devices in the target area is achieved.

[0037] The target data is sent to each of the second edge computing devices through the target slicing network, so that each of the second edge computing devices can perform AI operations based on the target data. Therefore, in the embodiment of the present application, when there are multiple edge computing devices in the target area, the target data can be received by the first edge computing device among the multiple edge computing devices. The AI ​​operation result is obtained by performing AI operations on the target data based on the AI ​​model through the second edge computing device. Thus, accurate AI operations are performed on the data in the target area and computing power sharing of multiple edge computing devices in the target area is achieved.

[0038] The AI ​​operation result is sent to the target edge computing device through the target slice network, wherein the target edge computing device includes: an edge computing device connected to the AI ​​operation result response device through a port, and the AI ​​operation result is the result obtained by performing AI operation on the target data based on the target AI model. That is, the AI ​​operation result is sent to the target edge computing device directly connected to the AI ​​operation result response device through the operator slice network, so as to send the AI ​​operation result to the AI ​​operation result response device through the port of the target edge computing device. For example, the AI ​​operation result response device is Figure 3 The AI ​​source data acquisition device 1, the first edge computing device is Figure 3 In the edge computing device B, in this step, the edge computing device B sends the AI ​​operation result to the edge computing device C through the operator slice network, and sends the AI ​​operation result to the AI ​​source data collection device 1 through the edge computing device C. Similarly, this step also includes the case where the AI ​​operation result response device is the third device mentioned above, which is not exhaustive.

[0039] Several comparative examples will be introduced below, and the scheme and beneficial effects of the embodiments of the present application will be further illustrated by comparing the embodiments of the present application with the following comparative examples.

[0040] Comparative Example 1: Deployment of AI algorithms based on cloud computing. That is, AI model training and AI operations are performed through the operator's private network or public network. Although this method is low-cost and easy to deploy, and can also train accurate AI models, it has the following disadvantages: a. Industrial production data has already left the factory, and there is a risk of data leakage; b. The data is first sent to the core network through the operator network, and then transmitted to the public cloud, and then the results are transmitted back to the AI ​​operation result response device through the operator network. The data transmission is time-consuming, and the real-time performance required for industrial production cannot be guaranteed.

[0041] In comparison, in the embodiment of the present application, the AI ​​source data in the target area has not left the factory, and there is no risk of data leakage; b. The data is processed near the target area, the data transmission time is short, the real-time required for data processing is guaranteed, and the training data can be sent to the cloud platform, and an accurate AI model can be trained. That is, the data processing provided by the embodiment of the present application can simultaneously meet the requirements of low cost, easy deployment, data security, centralized management, low latency and accurate training model, thereby ensuring the accuracy of data processing.

[0042] Comparative Example 2: Deployment of AI algorithms based on private cloud. This method builds a network and AI server in the factory. The industrial production line terminal transmits the raw data required for AI operation to the AI ​​server through the self-built network, and then completes the operation of the industrial AI application software on the server to obtain the operation result, and transmits the operation result back to the AI ​​operation result response device through the self-built network. The data in Comparative Example 2 does not leave the factory, and the real-time performance is guaranteed. However, the construction of the entire AI acceleration system is large, the deployment is difficult, and the cost is high. Completing AI model training on a self-built network requires a larger amount of work and cost, and the subsequent operation and maintenance requires a lot of manpower, and the technical requirements for the operation and maintenance personnel are high.

[0043] In comparison, the hardware equipment required for the edge computing operation and maintenance method in the embodiment of the present application can reuse the hardware equipment provided by existing operators, with a small construction workload, simple deployment, easy implementation, and low cost. In the later stage, the cloud platform provides AI model training and network operators operate and maintain it, without the need for additional manpower. That is, the edge computing operation and maintenance method provided by the embodiment of the present application can simultaneously meet the requirements of low cost, easy deployment, data security, centralized management, low latency, and accurate training models to ensure data processing accuracy.

[0044] Optionally, after S106, the AI ​​operation result can also be saved locally in the first edge computing device. The AI ​​operation result includes, for example, real-time workpiece images, production rates, yield rates, etc., and the data in the target area can be controlled in real time based on the AI ​​operation result. Thus, it is ensured that the AI ​​operation result obtained by performing AI operations on the target data based on the AI ​​model will not flow out of the target area, thereby ensuring data security. Combined with steps S102-S106, the embodiment of the present application can ensure that the training data of the AI ​​model can be sent to the cloud to train an accurate AI model, and can also ensure that the real-time data in the target area will not flow out, thereby ensuring data security.

[0045] Optionally, the target slice network of the embodiment of the present application can adopt the metropolitan area transport network MTN (Metro Transport Network) hard isolation technology supporting N×5G particles and the fine-grained unit FGU (Fine-Grained Unit) hard isolation technology based on N×10M particles.

[0046] like Figure 2 As shown, an embodiment of the present invention provides a data processing method, which can be executed by software or hardware of a first edge computing device, and the method includes the following steps:

[0047] S200: Acquire multiple target areas in the physical network.

[0048] In one implementation, the target area can be divided according to the actual data security requirements, and the target data in the target area is ultimately guaranteed not to flow out of the target area. For example, if the production data security of a certain factory A needs to be guaranteed, that is, the production data of factory A needs not to flow out of the physical area of ​​factory A, then the physical network area of ​​factory A can be determined as the target area. Optionally, factory A may include multiple factory buildings, each of which is located in a different city. For example, the first factory building of factory A is located in city 1, and the second factory building is located in city 2. Then, factory A including multiple factory buildings as a whole corresponds to a target area. Edge computing data can be sent between each factory building through the target slice network based on edge computing devices. The edge computing data may include edge computing data including all edge computing operation and maintenance system related data, such as the AI ​​source data collected by the AI ​​source data acquisition device, the running results of the AI ​​application software, and the processing results after processing the running results. The embodiment of the present application does not limit the transmission method of the operator network data. Therefore, multiple HECs can be integrated through the target slice network, and multiple HECs can be connected using the target slice network to perform model sharing or computing power integration scheduling, so as to achieve faster production line opening time and better control capabilities. The target slice network can ensure the data isolation and confidentiality of the target area. At the same time, the same target area can also utilize the low-latency and lossless transmission characteristics of the slice network to uniformly deploy and manage HECs across multiple regions.

[0049] S201: Divide the target slice network according to the data transmission parameters of each target area.

[0050] One target region corresponds to one target slice network, and target slice networks corresponding to different target regions are hard-isolated.

[0051] After this step, the network resource allocation of each target slice network may be determined according to the type of the target slice network, where the type of the target slice network is determined according to the data transmission parameter in the target area, wherein the network resource allocation includes one of the following:

[0052] Assigned based on business priorities;

[0053] Allocate according to business priority within dedicated bandwidth;

[0054] Allocate service-specific slice networks.

[0055] The data transmission parameters in the target area may include: rate, delay and data security parameters, such as bit error rate, etc. The data transmission parameters may include historical transmission parameters, current transmission parameters and target expected values ​​for future transmission parameters. Therefore, the embodiment of the present application can divide the target slice network and allocate network resources according to the actual data transmission situation.

[0056] Specifically, for situations where latency and data security are extremely high, a dedicated service slice network can be allocated to the target slice network. For situations where latency requirements are high, network resources can be allocated based on service priority within a dedicated bandwidth. For situations where latency requirements are relatively low, network resources can be allocated based on priority, that is, network resources can be allocated based on priority within a public bandwidth.

[0057] S202: Send the AI ​​model training data in the target area to the cloud platform through the target slice network corresponding to the target area.

[0058] The target area of ​​this step may be one of the multiple target areas obtained in step S200. After the target slice network is divided, the target area corresponds to a target slice network. In this step, the AI ​​model training data in the target area can be sent to the cloud platform through the target slice network corresponding to the target area.

[0059] The cloud platform trains the AI ​​model based on the AI ​​model training data to obtain a trained target AI model, and the cloud platform is deployed outside the target area.

[0060] S204: Receive the target AI model sent by the cloud platform through the target slicing network.

[0061] S206: Processing the target data collected in the target area based on the target AI model.

[0062] Steps S202-S206 can be performed by Figure 1 The description of steps S102 to S106 in the embodiment is not repeated here to avoid repetition.

[0063] The embodiment of the present application can flexibly and accurately divide the target slice network and flexibly and accurately determine the network resource allocation of each target slice network based on the data transmission parameters of each target area and according to the actual data transmission situation or demand of the target area.

[0064] like Figure 3 As shown, one embodiment of the present invention provides a data processing system 30 , which includes: a first edge computing device 31 and a cloud platform 32 .

[0065] The first edge computing device 31 sends the AI ​​model training data in the target area to the cloud platform through the target slice network corresponding to the target area. The cloud platform 32 is deployed outside the target area and connected to the first edge computing device 31, and is used to receive the AI ​​model training data, and train the AI ​​model based on the AI ​​model training data to obtain the trained target AI model, and send the target AI model to the first edge computing device 31. The first edge computing device 31 is also used to receive the target AI model sent by the cloud platform through the target slice network, and process the target data collected in the target area based on the target AI model.

[0066] In one implementation, the system 30 also includes: a management device 33, the management device 33 is arranged outside the target area, and the management device 33 is used to receive the AI ​​model training data sent by the first edge computing device 31 through the target slice network, and send the AI ​​model training data to the cloud platform. Optionally, the management device 33 can send the AI ​​model training data to the cloud platform through another slice network, and the other slice network is hard-isolated from the above-mentioned target slice network. Optionally, the management device 33 can also be set in the cloud, and the AI ​​model training data is sent to the cloud platform through the transmission method within the cloud platform. Figure 3 The line between the first edge computing device C and the management device 33 shown in the figure is used to indicate that the management device 33 can send the AI ​​model training data to the cloud platform, but does not indicate that the first edge computing device C is connected to the management device 33. The management device 33 is also used to read the target AI model from the cloud platform storage based on the model label and send it to the first edge computing device.

[0067] In one implementation, the management device 33 is also used to: perform computing power orchestration on each edge computing device in the target area according to the resource occupancy rate of each edge computing device in the target area, so as to instruct each edge computing device in the target area to perform processing corresponding to the computing power orchestration on the target data collected in the target area.

[0068] In one implementation, the management device 33 includes: an edge gateway control platform (Hyper-Management & Orchestration, HMO) or a unified management expert system (Unified Management Expert, UME) including HMO.

[0069] The first edge computing device 31 and the management device 33 may be used to perform Figure 1-2 The steps performed by the first edge computing device and the management device in the embodiment will not be repeated here for the repeated parts.

[0070] In one implementation, the management device 33 may also control the edge computing devices in the target area, which may specifically include at least one of the following:

[0071] Control the installation of the AI ​​application software in the edge computing device. This method can be executed by the installation module of the AI ​​application software, which supports uploading the AI ​​application software to the edge computing device through the network cable and installing or updating it; supports configuring the corresponding data transmission channel for the AI ​​application software, that is, specifying the network port / serial port, and configuring the corresponding edge computing device, AI chip, and AI computing board. Support the completion of uninstallation of the AI ​​application software.

[0072] Control the uninstallation of the AI ​​application software in the edge computing device. This method can be executed by the uninstallation module of the AI ​​application software, and supports the completion of the uninstallation of the AI ​​application software.

[0073] Controlling the configuration of the AI ​​application software in the edge computing device;

[0074] Controlling the update of AI application software in the edge computing device;

[0075] Control the startup of the AI ​​application software in the edge computing device; this method can be executed by the control module of the AI ​​application software, which supports the start / stop function of controlling the AI ​​application software.

[0076] Controlling the stopping of the AI ​​application software in the edge computing device. This method can be executed by a control module of the AI ​​application software, which supports the start / stop function of controlling the AI ​​application software.

[0077] Control the data management of the AI ​​application software in the edge computing device. This method can be executed by the operation data management module of the AI ​​application software, which supports the acquisition / save / statistics of AI operation process data transmitted from the edge computing device through the network port, including but not limited to the original image data collected from the production line, operation results, processing time, etc. The module also supports displaying this data to factory staff through the web page / application software UI interface.

[0078] In one implementation, the system 30 also includes: a control device 34, the control device 34 is set in the target area, and the control device 34 may include a hardware environment that can support the AI ​​application management software and operator network management software for controlling the packet transmission device and their operation, for example, it may include devices such as computers or servers. Specifically, the network operator provides a network connection for the target area, and the network management software will run on a certain hardware management device to run. The control device 34 provided in the embodiment of the present application can also be reused on the hardware management device provided by the network operator for the target area, thereby providing management edge computing functions without increasing additional hardware costs. The control device 34 can control the edge computing devices in the target area. Compared with the control of the edge computing devices by the management device 33 set outside the target area, the control device 34 can control the edge computing devices in the target area, which can better ensure that the data in the target area will not flow out of the target area, thereby ensuring data security. The control of the edge computing device can specifically include at least one of the following:

[0079] Control the installation of the AI ​​application software in the edge computing device. This method can be executed by the installation module of the AI ​​application software, which supports uploading the AI ​​application software to the edge computing device through the network cable and installing or updating it; supports configuring the corresponding data transmission channel for the AI ​​application software, that is, specifying the network port / serial port, and configuring the corresponding edge computing device, AI chip, and AI computing board. Support the completion of uninstallation of the AI ​​application software.

[0080] Control the uninstallation of the AI ​​application software in the edge computing device. This method can be executed by the uninstallation module of the AI ​​application software, and supports the completion of the uninstallation of the AI ​​application software.

[0081] Controlling the configuration of the AI ​​application software in the edge computing device;

[0082] Controlling the update of AI application software in the edge computing device;

[0083] Control the startup of the AI ​​application software in the edge computing device; this method can be executed by the control module of the AI ​​application software, which supports the start / stop function of controlling the AI ​​application software.

[0084] Controlling the stopping of the AI ​​application software in the edge computing device. This method can be executed by a control module of the AI ​​application software, which supports the start / stop function of controlling the AI ​​application software.

[0085] Control the data management of the AI ​​application software in the edge computing device. This method can be executed by the operation data management module of the AI ​​application software, which supports the acquisition / save / statistics of AI operation process data transmitted from the edge computing device through the network port, including but not limited to the original image data collected from the production line, operation results, processing time, etc. The module also supports displaying this data to factory staff through the web page / application software UI interface.

[0086] Therefore, the data processing system 30 provided in the embodiment of the present application can perform AI operations based on the target data and AI model in the target area through the first edge computing device 31, can accurately train the AI ​​model through the cloud platform to ensure the accuracy of AI operations, and can ensure the data security in the target area through the properties of the slice network. The training data can be transmitted to the cloud platform through the management device 34, and the first edge computing device 31 can be controlled by the management device 34. The first edge computing device can be controlled by the control device 32.

[0087] Optional, such as Figure 4 As shown, the embodiment of the present application further provides an electronic device 400, including a processor 401, a memory 402, and a program or instruction stored in the memory 402 and executable on the processor 401, wherein the program or instruction, when executed by the processor 401, implements each process of the above data processing method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described here.

[0088] An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the various processes of the above-mentioned data processing method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0089] The processor is the processor in the terminal described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk.

[0090] The embodiment of the present application further provides a chip, the chip includes a processor and a communication card slot, the communication card slot is coupled to the processor, the processor is used to run programs or instructions, implement the various processes of the above data processing method embodiment, and can achieve the same technical effect, to avoid repetition, no further description is given here. It should be understood that the chip mentioned in the embodiment of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0091] The embodiments of the present application further provide a computer program / program product, which is stored in a storage medium and is executed by at least one processor to implement the various processes of the above-mentioned data processing method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described here.

[0092] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0093] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, a disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0094] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

Claims

1. A data processing method, applied to a first edge computing device, the method comprising: Sending AI model training data in the target area to the cloud platform through a target slice network corresponding to the target area, so that the cloud platform can train the AI ​​model based on the AI ​​model training data to obtain a trained target AI model, and the cloud platform is deployed outside the target area; Receiving the target AI model issued by the cloud platform through the target slicing network; The target data collected in the target area is processed based on the target AI model.

2. The method according to claim 1, characterized in that The processing of the target data collected in the target area based on the target AI model includes at least one of the following: Performing AI operation on the target data based on the AI ​​model to obtain an AI operation result, and sending the AI ​​operation result to an AI operation result response device; The target AI model is sent to each second edge computing device through the target slice network, so that each second edge computing device performs AI operation on the target data based on the AI ​​model; the second edge computing device is deployed in the target area; Sending the target data to each of the second edge computing devices through the target slice network, so that each of the second edge computing devices performs AI computing based on the target data; The AI ​​operation result is sent to a target edge computing device through the target slicing network, wherein the target edge computing device includes: an edge computing device connected to an AI operation result response device through a port, and the AI ​​operation result is a result obtained by performing an AI operation on the target data based on the target AI model.

3. The method according to claim 1 or 2, characterized in that Before sending the AI ​​model training data in the target area to the cloud platform through the target slice network corresponding to the target area, the method further includes: Acquire a plurality of target areas in the physical network; The target slice network is divided according to the data transmission parameters of each target area, wherein one target slice network corresponds to one target area, and target slice networks corresponding to different target areas are hard isolated.

4. The method according to claim 3, characterized in that After dividing the target slice network, the method further includes: Determine network resource allocation for each of the target slice networks according to the type of the target slice network, wherein the type of the target slice network is determined according to a data transmission parameter in the target area, wherein the network resource allocation includes one of the following: Assigned based on business priorities; Allocate according to business priority within dedicated bandwidth; Allocate service-specific slice networks.

5. The method according to claim 2, characterized in that The method further comprises: The AI ​​operation results are saved locally.

6. The method according to claim 1, characterized in that The sending of the AI ​​model training data in the target area to the cloud platform through the target slice network corresponding to the target area includes: Through the target slice network, the AI ​​model training data collected by the application data collection terminal in the target area is sent to the management device deployed outside the target area, so that the management device can send the AI ​​model training data to the cloud platform.

7. The method according to claim 6, characterized in that The receiving, through the target slicing network, the target AI model issued by the cloud platform includes: The target AI model sent by the management device is received through the target slice network; wherein the target AI model is read by the management device from the cloud platform storage based on the model tag.

8. A data processing system, characterized in that: include: A first edge computing device, wherein the first edge computing device sends the AI ​​model training data in the target area to the cloud platform through a target slice network corresponding to the target area; The cloud platform is deployed outside the target area and connected to the first edge computing device, and is used to receive the AI ​​model training data, train the AI ​​model based on the AI ​​model training data, obtain the trained target AI model, and send the target AI model to the first edge computing device; The first edge computing device is also used to receive the target AI model issued by the cloud platform through the target slicing network, and process the target data collected in the target area based on the target AI model.

9. The system according to claim 8, characterized in that Also includes: A management device, the management device being used to receive the AI ​​model training data sent by the first edge computing device, and send the AI ​​model training data to the cloud platform; as well as The management device is also used to read the target AI model from the cloud platform storage based on the model tag and send it to the first edge computing device.

10. The system according to claim 9, characterized in that The management device is also used to: perform computing power orchestration on each edge computing device in the target area according to the resource occupancy rate of each edge computing device in the target area, so as to instruct each edge computing device in the target area to process the target data collected in the target area corresponding to the computing power orchestration.

11. The system according to claim 9, characterized in that The management device includes: an edge gateway control platform HMO or a unified management expert system UME including HMO.

12. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores programs or instructions that can be run on the processor, and when the programs or instructions are executed by the processor, the steps of the data processing method according to any one of claims 1 to 7 are implemented.

13. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by a processor, the steps of the data processing method according to any one of claims 1 to 7 are implemented.

Images