Multi-scale model quantification method and related device for power side end equipment

Abstract

The invention provides a multi-scale model quantification method of power side end equipment and a related device. The method comprises the following steps: acquiring a multi-scale model of the power side end equipment; then determining a first data calculation node from the plurality of data calculation nodes, and determining a data operation scale and a data operation mode of the first data calculation node; sending the data operation scale and the data operation mode to a second server, so that the second server determines a first quantification strategy based on the data operation scale and the data operation mode; and finally, the first quantification strategy is obtained from the second server, multi-scale model quantification is carried out on the first data calculation node based on the first quantification strategy, and accurate quantification of the multi-scale model of the power side end equipment is realized through the data operation scale and the data operation mode of the first data calculation node.

Description

Technical Field

[0001] The present application relates to the technical field of multi-scale model quantization of power edge devices, and in particular to a multi-scale model quantization method and related devices for power edge devices. Background Art

[0002] Model quantization is a model compression technology that converts floating-point storage (operation) into integer storage (operation).

[0003] With the rapid application of deep learning technology in many fields, including computer vision (CV), natural language processing (NLP), speech, etc., a large number of network models based on deep learning have emerged. These models have a characteristic, that is, they are large and complex, suitable for reasoning on N cards, and not suitable for application in embedded devices. However, customers usually need to deploy these complex models in some low-cost embedded devices, which creates a contradiction. In order to solve this contradiction well, model quantization came into being. It can compress the model under the premise of losing a small amount of accuracy, making it possible to apply these complex models to embedded terminals.

[0004] Smart grid is built on the basis of integrated, high-speed two-way communication network. Through the application of advanced sensing and measurement technology, advanced equipment technology, advanced control methods and advanced decision support system technology, the goal of achieving reliable, safe, economical, efficient, environmentally friendly and safe use of the power grid is achieved. As the prediction of models applied to the power grid becomes more and more accurate and the network becomes deeper, the memory size consumed by the neural network becomes a core issue. In addition, the model size is not only a problem of memory capacity, but also a problem of memory bandwidth. The model uses the weight of the model every time it predicts. Image-related applications usually require real-time data processing, which means at least 30FPS. Therefore, even if a relatively small ResNet-50 network is deployed for classification, running the network model requires 3GB / s of memory bandwidth. When the network is running, memory, CPU and battery will be consumed rapidly. At the same time, if a new model design is added to the multi-model architecture or a part of the data calculation design is inserted into the multi-model architecture after the multi-model architecture is formed, the new overall architecture system needs to be re-quantified, which greatly increases the operation and maintenance cost. Therefore, there is an urgent need to design a multi-scale model quantification technology for power edge devices to achieve lightweight design of smart grids, reduce the operation and maintenance costs of smart grids, and improve the overall intelligent operation efficiency of smart grids. Summary of the invention

[0005] In view of this, the purpose of this application is to propose a multi-scale model quantization method and related devices for power edge equipment to solve or partially solve the above-mentioned technical problems.

[0006] Based on the above objectives, the present application provides a multi-scale model quantification method for power edge equipment, including:

[0007] Acquire a multi-scale model of a power edge device; the multi-scale model includes a plurality of data computing nodes;

[0008] Determine a first data computing node from a plurality of data computing nodes, and determine a data operation scale and a data operation mode of the first data computing node;

[0009] Sending the data operation scale and the data operation mode to a second server, so that the second server determines a first quantization strategy based on the data operation scale and the data operation mode;

[0010] The first quantization strategy is obtained from the second server, and a multi-scale model is quantized on the first data computing node based on the first quantization strategy.

[0011] Based on the same concept, the present application also proposes a multi-scale model quantization device for power edge equipment, which is applied to a first server and includes:

[0012] An acquisition module is used to acquire a multi-scale model of a power edge device; the multi-scale model includes a plurality of data computing nodes;

[0013] A determination module is configured to determine a first data computing node from among a plurality of data computing nodes, and determine a data operation scale and a data operation mode of the first data computing node;

[0014] a sending module, which sends the data operation scale and the data operation mode to a second server, so that the second server determines a first quantization strategy based on the data operation scale and the data operation mode;

[0015] A quantization module obtains the first quantization strategy from the second server, and performs multi-scale model quantization on the first data computing node based on the first quantization strategy.

[0016] Based on the same concept, the present application also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and run on the processor. When the processor executes the program, it implements the multi-scale model quantization method of the power edge device as described above.

[0017] Based on the same concept, the present application also proposes a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to execute the multi-scale model quantization method of the power edge equipment as described above.

[0018] Based on the same concept, the present application also proposes a computer program product, which includes a computer program, and the computer program is executed by one or more processors so that the processor executes the multi-scale model quantization method of the power edge device as described above.

[0019] From the above, it can be seen that the multi-scale model quantization method and related devices for power edge devices provided in the present application first obtain the multi-scale model of the power edge device; the multi-scale model includes multiple data computing nodes; then determine the first data computing node from the multiple data computing nodes, and determine the data calculation scale and data operation mode of the first data computing node; send the data calculation scale and the data operation mode to the second server, so that the second server determines the first quantization strategy based on the data calculation scale and the data operation mode; finally, obtain the first quantization strategy from the second server, and quantize the multi-scale model of the first data calculation node based on the first quantization strategy. The multi-scale model of the power edge device is accurately quantized through the data calculation scale and data operation mode of the first data calculation node, reducing the operation and maintenance cost of the power system. At the same time, by completing the calculation of the quantization strategy in the second server, the risk of adjusting the multi-scale model is effectively reduced, the customer experience is improved, and technical support is provided for the development of intelligent technology for power edge devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the present application or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 A schematic diagram of an application scenario of an embodiment of the present application;

[0022] Figure 2 A schematic diagram of a flow chart of a multi-scale model quantization method for power edge equipment according to an embodiment of the present application;

[0023] Figure 3 A schematic diagram of the structure of a multi-scale model quantization device for power edge equipment according to an embodiment of the present application;

[0024] Figure 4 A schematic diagram of a specific electronic device hardware structure of an embodiment of the present application. DETAILED DESCRIPTION

[0025] It is understandable that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and relevant provisions.

[0026] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present application should be the usual meanings understood by people with general skills in the field to which the present application belongs. The "first", "second" and similar words used in the present application do not represent any order, quantity or importance, but are only used to distinguish different components. "Including" or "comprising" and similar words mean that the elements or objects appearing in front of the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connecting" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, which can be direct connections, indirect connections, wired connections, and wireless connections. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly. In addition, in the description of the present application, unless otherwise specified, the term "multiple" refers to two or more. The term "and / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects are in an "or" relationship.

[0027] The principles and spirit of the present application will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present application, and are not intended to limit the scope of the present application in any way. On the contrary, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

[0028] It is understandable that before using the technical solutions of each embodiment of the present application, the type, scope of use, usage scenarios, etc. of the personal information involved will be informed to the user in an appropriate manner, and the user's authorization will be obtained.

[0029] For example, in response to receiving an active request from a user, a prompt message is sent to the user to clearly remind the user that the operation requested to be performed will require obtaining and using the user's personal information. Thus, the user can independently choose whether to provide personal information to the electronic device, application, server, storage medium or other software or hardware that performs the operation of the technical solution of the present application according to the prompt message.

[0030] As an optional but non-limiting implementation, in response to receiving the user's active request, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. In addition, the pop-up window may also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0031] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation method of this application. Other methods that meet relevant laws and regulations may also be applied to the implementation method of this application.

[0032] It should be understood herein that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction rather than having any limiting meaning.

[0033] As described in the background technology section, in the current related technologies, as the predictions of models applied to power grids become more and more accurate and the networks become deeper, the memory size consumed by neural networks has become a core issue. In addition, model size is not only a problem of memory capacity, but also a problem of memory bandwidth. The model uses the weights of the model every time it makes a prediction, and image-related applications usually require real-time data processing, which means at least 30FPS. Therefore, even if a relatively small ResNet-50 network is deployed for classification, running the network model requires 3GB / s of memory bandwidth. When the network is running, memory, CPU, and battery will be consumed rapidly. At the same time, if a new model design is added to the multi-model architecture or a partial data calculation design is inserted into the multi-model architecture after the multi-model architecture is formed, the new overall architecture system needs to be re-quantified, which greatly increases the operation and maintenance costs.

[0034] In summary, in order to solve the above problems, the embodiment of the present application proposes a multi-scale model quantization method for power edge equipment, including:

[0035] First, a multi-scale model of the power edge device is obtained; the multi-scale model includes multiple data computing nodes; then, a first data computing node is determined from the multiple data computing nodes, and the data computing scale and data operation mode of the first data computing node are determined; the data computing scale and the data operation mode are sent to the second server, so that the second server determines a first quantization strategy based on the data computing scale and the data operation mode; finally, the first quantization strategy is obtained from the second server, and the multi-scale model of the first data computing node is quantified based on the first quantization strategy. The multi-scale model of the power edge device is accurately quantified through the data computing scale and data operation mode of the first data computing node, thereby reducing the operation and maintenance cost of the power system. At the same time, by completing the calculation of the quantization strategy in the second server, the risk of adjusting the multi-scale model is effectively reduced, the customer experience is improved, and technical support is provided for the development of intelligent technology for power edge devices.

[0036] In some specific application scenarios, the multi-scale model quantization method of the power edge device of the present application can be applied to various platforms or systems involving multi-scale model quantization of power edge devices. As an example, the application scenario includes at least one server and at least one terminal. Figure 1 , which is a schematic diagram of an application scenario of a multi-scale model quantification method for a power edge device provided by an exemplary embodiment of the present disclosure. The application scenario includes a terminal device 101, a server 102, and a data storage system 103. Among them, the terminal device 101, the server 102 and the data storage system 103 can be connected through a wired or wireless communication network. The terminal device includes but is not limited to a desktop computer, a mobile phone, a mobile computer, a tablet computer, a media player, a smart wearable device, a personal digital assistant (PDA) or other electronic devices capable of realizing the above functions. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server and the terminal can communicate through the network to realize data transmission. Among them, the network can be a wired network or a wireless network, which is not specifically limited in this application.

[0037] The server may be a server that provides various services. Specifically, the server may be used to provide background services for applications running on the terminal. Optionally, in some implementations, the multi-scale model quantization method for the power edge device provided in the embodiment of the present application may be executed by the server. Optionally, in some implementations, the multi-scale model quantization method for the power edge device provided in the embodiment of the present application may be directly executed by the terminal device.

[0038] Optionally, the server can be hardware or software. When the server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or it can be implemented as a single server. When the server is software, it can be implemented as multiple software or software modules (for example, software or software modules for providing distributed services), or it can be implemented as a single software or software module. This embodiment of the application does not specifically limit this.

[0039] Optionally, the wireless network or wired network uses standard communication technology and / or protocol. The network is usually the Internet, but it can also be any network, including but not limited to a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a mobile, wired or wireless network, a dedicated network or any combination of a virtual private network). In some embodiments, the data exchanged through the network is represented by technologies and / or formats including hypertext mark-up language (HTML), extensible markup language (XML), etc. In addition, conventional encryption technologies such as secure socket layer (SSL), transport layer security (TLS), virtual private network (VPN), and Internet protocol security (IPsec) can be used to encrypt all or some links. In other embodiments, customized and / or dedicated data communication technologies can also be used to replace or supplement the above data communication technologies.

[0040] The data storage system stores a large amount of data. Optionally, the data storage system can partition the data according to different uses and data types. The source of the data includes but is not limited to an existing database, data crawled from the Internet, or data uploaded by the user when using the client.

[0041] The following describes the multi-scale model quantization method of the power edge device according to the exemplary embodiment of the present application in combination with a specific application scenario. It should be noted that the above application scenario is only shown to facilitate understanding of the spirit and principle of the present application, and the implementation of the present application is not limited in this regard. On the contrary, the implementation of the present application can be applied to any applicable scenario.

[0042] refer to Figure 2 , is a flow chart of a multi-scale model quantization method for power edge equipment in an embodiment of the present application, the method comprising the following steps:

[0043] S101, obtaining a multi-scale model of a power edge device; the multi-scale model includes multiple data computing nodes.

[0044] In specific implementation, the multi-scale model of the power edge device includes multiple data computing nodes, each of which can be responsible for a part of the computing tasks of the power edge device. The specific correspondence between each computing node and the computing tasks it is responsible for can be set as needed, and there is no limitation on this.

[0045] S102, determining a first data computing node from a plurality of data computing nodes, and determining a data operation scale and a data operation mode of the first data computing node.

[0046] In specific implementation, the first data computing node is the node that currently needs to be quantized. Optionally, in theory, the first data computing node can be any one of multiple data computing nodes. In some embodiments, the data computing form is used to represent the processing method of the data computing node for data computing, and the data operation method includes independent computing and distributed computing. The data computing scale is used to represent the data size of the participating data computing nodes.

[0047] S103: Send the data operation scale and the data operation mode to the second server, so that the second server determines a first quantization strategy based on the data operation scale and the data operation mode.

[0048] In specific implementation, in order to ensure the stability of each data computing node in the first server during model quantization, this embodiment uses the second server to calculate the quantization strategy, so that when the current quantization strategy has an adverse effect on the second server, the various data computing nodes of the first server can still operate normally. It should be noted that in order to enable the second server to simulate the operation process of the first server well, this embodiment can map the multi-scale model architecture in the first server to the second server. For convenience, the multi-scale model mapped to the second server can be called a simulated multi-scale model. When the first service obtains the data operation scale and data operation mode of the first data computing node, the data operation scale and the data operation mode can be sent to the second server, so that the second server determines the first quantization strategy based on the data operation scale and the data operation mode.

[0049] In some embodiments, before obtaining the first quantitative strategy from the second server, the method further includes:

[0050] In response to determining that a preset fault occurs in the simulated multi-scale model in the second server, stop obtaining the first quantization strategy from the second server, and re-quantize the multi-scale model of the first data computing node based on the data operation scale and the data operation mode; wherein the simulated multi-scale model is obtained by mapping the multi-scale model in the first server.

[0051] In specific implementation, after the second server generates the first quantization strategy, it can test run the first quantization strategy in a simulated multi-scale model to determine whether a fault will occur during the operation of the first quantization strategy. When it is determined that the simulated multi-scale model in the second server has a preset fault, the first server can promptly stop obtaining the first quantization strategy from the second server, and re-quantize the multi-scale model of the first data calculation node based on the data calculation scale and the data operation mode. Optionally, when the simulated multi-scale model in the second server has a preset fault, a fault prompt can be sent to the first server so that the first server determines that the simulated multi-scale model in the second server has a preset fault. Optionally, when the first server does not receive a fault prompt sent by the second server within a preset time after sending the data calculation scale and the data operation mode to the second server, it can be assumed that the first quantization strategy will not cause a fault in the operation of the multi-scale model.

[0052] In some implementations, sending the data operation scale and the data operation mode to the second service specifically includes:

[0053] In response to determining that the data operation mode is distributed computing, acquiring an association relationship between the first data computing node and each of the multiple data computing nodes;

[0054] The association relationship, the data operation scale and the data operation mode are sent to the second server, so that the second server generates a first quantization strategy based on the association relationship, the data operation scale and the data operation mode to determine the first quantization strategy.

[0055] In specific implementation, optionally, in some embodiments, the second server may determine the first quantization strategy according to the data operation scale and the data operation mode after receiving the data operation scale and the data operation mode of the first data computing node. Optionally, when determining the first quantization strategy, the data operation mode of the first data computing node may be first determined. When the data operation mode is independent operation, the first quantization strategy may be directly determined by the data operation scale. When the data operation mode is distributed operation, the first quantization strategy may be further determined by the association relationship between the first data computing node and each of the multiple data computing nodes and the data operation scale. Optionally, when determining the first quantization strategy, multiple model quantization strategies may be first obtained according to relevant technologies, for example, lit16 quantization, lit8 quantization, lit4 quantization, etc., and then a target model quantization strategy may be determined from the multiple model quantization strategies according to the data operation scale and the data operation mode as the first quantization strategy. Optionally, in general, when the scale of data operations is large and the association relationship is complex (that is, the number of data computing nodes that have an association relationship with the current data computing node is large), priority is given to model quantization strategies with faster processing efficiency; when the scale of data operations is small and the association relationship is simple, priority is given to model quantization strategies with higher processing accuracy.

[0056] It should be noted that in this embodiment, the degree of replication of the current association relationship can be determined by the number of association relationships between the current data computing node and other data computing nodes, and the association relationship can be determined to be replicated or simple according to the first preset threshold. Optionally, the data operation scale can be determined to be large or small according to the second preset threshold.

[0057] Optionally, in some embodiments, when generating the first quantization strategy, the importance level of the data currently processed by the first data computing node may be further obtained, and according to the importance level, it may be determined whether to select a model quantization strategy with better calculation accuracy.

[0058] S104: Acquire the first quantization strategy from the second server, and perform multi-scale model quantization on the first data computing node based on the first quantization strategy.

[0059] In a specific implementation, after obtaining the first quantization strategy, the multi-scale model may be quantized on the first data computing node according to the first quantization strategy.

[0060] In some embodiments, the method further comprises:

[0061] In response to a second data computing node being newly added to the multi-scale model, determining a target data computing node among the plurality of data computing nodes that has an interactive relationship with the second data computing node;

[0062] Sending the target data computing node to the second server, so that the second server generates a second quantization strategy based on the target data computing node;

[0063] The second quantization strategy is obtained from the second server, and the multi-scale model is quantized on the second data computing node based on the second quantization strategy.

[0064] In some embodiments, after obtaining the target data computing nodes, the second server regards all the target data computing nodes as a target entity, and generates the second quantization strategy based on the quantization strategy of the target entity.

[0065] In specific implementation, when a new data computing node is added to the multi-scale model, that is, a second data computing node is added, in order not to affect the quantization of the multi-scale model as a whole and avoid the need to re-quantize each data computing node in the related art, in this embodiment, a target data computing node having an interactive relationship with the second data computing node is first determined among the multiple data computing nodes, and then the target data computing node is sent to the second server, so that the second server generates a second quantization strategy based on the target data computing node. After receiving the target data computing node, the second server regards all the target data computing nodes as a target as a whole and determines the quantization strategy of the target as a whole. When determining the quantization strategy of the target as a whole, the target as a whole can be regarded as a data computing node, and the data operation scale and data operation mode of the target as a whole are determined, and then the quantization strategy of the target as a whole is determined according to the above method for determining the first quantization strategy of the first data computing node. After determining the quantization strategy of the target as a whole, it is equivalent to determining the total quantization accuracy and total quantization efficiency of the target as a whole, and fixing the total quantization accuracy and total quantization efficiency so that the output data volume per unit time of the target as a whole is a certain value. Then, the second data calculation node is added to the target as a whole and, under the condition that the output data volume per unit time of the target as a whole remains unchanged, the quantization of each data calculation node is performed within the target as a whole, thereby avoiding the influence of the newly added second data calculation node on other parts of the multi-scale model, that is, the other parts of the multi-scale model continue to maintain the previous quantization strategy.

[0066] The multi-scale model quantization method of the power edge device provided in the present application first obtains the multi-scale model of the power edge device; the multi-scale model includes multiple data computing nodes; then determines the first data computing node from the multiple data computing nodes, and determines the data computing scale and data operation mode of the first data computing node; sends the data computing scale and the data operation mode to the second server, so that the second server determines a first quantization strategy based on the data computing scale and the data operation mode; finally, obtains the first quantization strategy from the second server, and quantizes the multi-scale model of the first data computing node based on the first quantization strategy. The multi-scale model of the power edge device is accurately quantized through the data computing scale and data operation mode of the first data computing node, reducing the operation and maintenance cost of the power system. At the same time, by completing the calculation of the quantization strategy in the second server, the risk of adjusting the multi-scale model is effectively reduced, the customer experience is improved, and technical support is provided for the development of intelligent technology for power edge devices.

[0067] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present application also provides a multi-scale model quantization device for power edge equipment, which is applied to the first server.

[0068] refer to Figure 3 , the multi-scale model quantization device of the power edge equipment comprises:

[0069] An acquisition module 201 is used to acquire a multi-scale model of a power edge device; the multi-scale model includes a plurality of data computing nodes;

[0070] A determination module 202 determines a first data computing node from a plurality of data computing nodes, and determines a data operation scale and a data operation mode of the first data computing node;

[0071] A sending module 203 sends the data operation scale and the data operation mode to a second server, so that the second server determines a first quantization strategy based on the data operation scale and the data operation mode;

[0072] The quantization module 204 obtains the first quantization strategy from the second server, and performs multi-scale model quantization on the first data computing node based on the first quantization strategy.

[0073] For the convenience of description, the above device is described in terms of functions divided into various modules. Of course, when implementing the present application, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0074] The multi-scale model quantization device of the power edge device in the above-mentioned embodiment is used to implement the multi-scale model quantization method of the corresponding power edge device in any of the above-mentioned embodiments, and has the beneficial effects of the corresponding multi-scale model quantization method embodiment of the power edge device, which will not be repeated here.

[0075] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the multi-scale model quantization method for the power edge device described in any of the above embodiments is implemented.

[0076] It should be noted that the method of the embodiment of the present application can be performed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present application, and the multiple devices will interact with each other to complete the described method.

[0077] It should be noted that the above describes some embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0078] Figure 4 A more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment is shown. The device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040 and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030 and the communication interface 1040 are connected to each other in communication within the device through the bus 1050.

[0079] The processor 1010 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0080] The memory 1020 may be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 may store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 1020 and are called and executed by the processor 1010.

[0081] The input / output interface 1030 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.

[0082] The communication interface 1040 is used to connect a communication module (not shown) to realize communication interaction between the device and the power equipment. The communication module can realize communication through a wired method (such as USB, network cable, etc.) or a wireless method (such as mobile network, WIFI, Bluetooth, etc.).

[0083] The bus 1050 includes a path that transmits information between the various components of the device (eg, the processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040).

[0084] It should be noted that, although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040 and the bus 1050, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.

[0085] The electronic device of the above-mentioned embodiment is used to implement the multi-scale model quantization method of the corresponding power edge device in any of the above-mentioned embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0086] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, the present disclosure also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the multi-scale model quantization method of the power edge equipment as described in any of the above embodiments.

[0087] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0088] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the multi-scale model quantization method of the power edge equipment as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0089] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments, the present application also provides a computer program product, which includes a computer program. In some embodiments, the computer program instructions can be executed by one or more processors of a computer so that the computer and / or the processor execute the multi-scale model quantization method for the power edge device described in the above-mentioned embodiment. Corresponding to the execution subject corresponding to each step in each embodiment of the multi-scale model quantization method for the power edge device, the processor that executes the corresponding step may belong to the corresponding execution subject.

[0090] To simplify the description and discussion, and in order not to make the embodiments of the present application difficult to understand, the known power supply / ground connection with the integrated circuit (IC) chip and other components may or may not be shown in the provided drawings. In addition, the device can be shown in the form of a block diagram to avoid making the embodiments of the present application difficult to understand, and this also takes into account the following fact, that is, the details of the implementation of these block diagram devices are highly dependent on the platform to be implemented in the embodiments of the present application (that is, these details should be fully within the scope of understanding of those skilled in the art). In the case of elaborating specific details (e.g., circuit) to describe exemplary embodiments of the present application, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details or when these specific details are changed. Therefore, these descriptions should be considered to be illustrative rather than restrictive.

[0091] Although the present application has been described in conjunction with specific embodiments of the present application, many replacements, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0092] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present application (including the claims) is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the present application as described above, which are not provided in detail for the sake of simplicity.

[0093] The embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application should be included in the scope of protection of the present application.

Claims

1. A multi-scale model quantification method for power edge equipment, characterized in that: The method is applied in a first server, and the method includes: Acquire a multi-scale model of a power edge device; the multi-scale model includes a plurality of data computing nodes; Determine a first data computing node from a plurality of data computing nodes, and determine a data operation scale and a data operation mode of the first data computing node; Sending the data operation scale and the data operation mode to a second server, so that the second server determines a first quantization strategy based on the data operation scale and the data operation mode; The first quantization strategy is obtained from the second server, and the first data computing node is processed based on the first quantization strategy.

2. The method according to claim 1, characterized in that The data operation modes include independent operation and distributed operation.

3. The method according to claim 1, characterized in that The data operation scale is used to represent the data size of the nodes involved in the data calculation.

4. The method according to claim 2, characterized in that: Sending the data operation scale and the data operation mode to the second server specifically includes: In response to determining that the data operation mode is distributed computing, acquiring an association relationship between the first data computing node and each of the multiple data computing nodes; The association relationship, the data operation scale and the data operation mode are sent to the second server, so that the second server determines a first quantization strategy based on the association relationship, the data operation scale and the data operation mode.

5. The method according to claim 1, characterized in that Before acquiring the first quantization strategy from the second server, the method further includes: In response to determining that a preset fault occurs in the simulated multi-scale model in the second server, stop obtaining the first quantization strategy from the second server, and re-quantize the multi-scale model of the first data computing node based on the data operation scale and the data operation mode; wherein the simulated multi-scale model is obtained by mapping the multi-scale model in the first server.

6. The method according to claim 1, characterized in that The method further comprises: In response to a second data computing node being newly added to the multi-scale model, determining a target data computing node among the plurality of data computing nodes that has an interactive relationship with the second data computing node; Sending the target data computing node to the second server, so that the second server generates a second quantization strategy based on the target data computing node; The second quantization strategy is obtained from the second server, and the multi-scale model is quantized on the second data computing node based on the second quantization strategy.

7. The method according to claim 6, characterized in that After obtaining the target data computing nodes, the second server regards all the target data computing nodes as a target entity, and generates the second quantization strategy based on the quantization strategy of the target entity.

8. A multi-scale model quantification device for power edge equipment, characterized in that: The device is applied to a first server, and includes: An acquisition module is used to acquire a multi-scale model of a power edge device; the multi-scale model includes a plurality of data computing nodes; A determination module is configured to determine a first data computing node from among a plurality of data computing nodes, and determine a data operation scale and a data operation mode of the first data computing node; a sending module, which sends the data operation scale and the data operation mode to a second server, so that the second server determines a first quantization strategy based on the data operation scale and the data operation mode; A quantization module obtains the first quantization strategy from the second server, and performs multi-scale model quantization on the first data computing node based on the first quantization strategy.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 7.

Images