CN116720132A - Power service identification system, method, device, medium and product - Google Patents

Power service identification system, method, device, medium and product Download PDF

Info

Publication number
CN116720132A
CN116720132A CN202310596642.1A CN202310596642A CN116720132A CN 116720132 A CN116720132 A CN 116720132A CN 202310596642 A CN202310596642 A CN 202310596642A CN 116720132 A CN116720132 A CN 116720132A
Authority
CN
China
Prior art keywords
power
data
business
service
classification model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310596642.1A
Other languages
Chinese (zh)
Inventor
翁俊鸿
高强
高易年
吕为
吴谦
周建勇
徐琼
陈嘉
周盈延
曾旭
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Power Supply Bureau Co Ltd
Original Assignee
Shenzhen Power Supply Bureau Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Power Supply Bureau Co Ltd filed Critical Shenzhen Power Supply Bureau Co Ltd
Priority to CN202310596642.1A priority Critical patent/CN116720132A/en
Publication of CN116720132A publication Critical patent/CN116720132A/en
Pending legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Molecular Biology (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Business, Economics & Management (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Economics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Water Supply & Treatment (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Public Health (AREA)
  • Primary Health Care (AREA)
  • Probability & Statistics with Applications (AREA)
  • Medical Informatics (AREA)
  • Marketing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The application relates to a power service identification system, a method, equipment, a medium and a product. The system comprises an intelligent gateway, wherein a power service data acquisition module, a power service data preprocessing module and a power service identification module are arranged on the intelligent gateway, and the power service data acquisition module is used for acquiring power service flow data to be identified; the power business data preprocessing module is used for preprocessing the power business flow data to obtain target power business characteristic data; and the power business identification module is used for carrying out power business identification on the target power business characteristic data by utilizing a pre-trained target power business classification model. By adopting the system, the power service can be timely identified, the running states of node resources and equipment in each layer in the power grid system can be obtained, and all service flows in the network can be controlled and managed in a layered manner and allocated according to different service types, so that the utilization rate of network resources is improved to the greatest extent.

Description

Power service identification system, method, device, medium and product
Technical Field
The application relates to the technical field of smart grids, in particular to a power service identification system, a method, equipment, a medium and a product.
Background
With popularization of internet application and advancement of digital transformation, an electric power system gradually merges into a large environment of information interconnection from a relatively closed and self-contained system, and nowadays, efficient operation of the electric power system is increasingly dependent on an information service system borne by an electric power communication network, wherein the proportion of electric power services occupying large bandwidth such as large data flow, P2P service and VoP service is increasingly improved, and the electric power services with high instantaneity, large flow and IP and communication index requirements easily bring huge impact to the network.
The data flow direction of the electric power system is different from the network data flow direction of the traditional network operators, the network data of the operators mainly acts as follows, the network data of the electric power system mainly acts as follows, all video data, state data, metering data, relay protection data and the like of network access equipment in the electric power system are uploaded to a server system, if events are concurrent, network congestion is possibly caused, key data cannot be uploaded in real time, system response is slow, and even the situation that the response cannot be timely occurs.
Therefore, in order to ensure reliable, timely and stable transmission of important power service data, different service quality grades are provided for different types of applications, fine management and dynamic control of service flow are realized, and the power service needs to be identified timely, so that the utilization rate of network resources is improved to the greatest extent.
Disclosure of Invention
Based on this, it is necessary to provide a power service identification system, method, device, medium and product capable of identifying a power service in time and improving the utilization rate of network resources.
In a first aspect, the present application provides a power service identification system. The system comprises an intelligent gateway, wherein a power service data acquisition module, a power service data preprocessing module and a power service identification module are arranged on the intelligent gateway, and are connected in sequence;
the power business data acquisition module is used for acquiring power business flow data to be identified;
the power business data preprocessing module is used for receiving the power business flow data from the power business data acquisition module and preprocessing the power business flow data to obtain target power business characteristic data;
And the power business identification module is used for receiving the target power business characteristic data from the power business data preprocessing module and carrying out power business identification on the target power business characteristic data by utilizing a pre-trained target power business classification model.
In one embodiment, the system further comprises an edge server, and the edge server is connected with the intelligent gateway;
the power business data acquisition module is also used for acquiring time sequence sample data of power business and a power business class label, wherein the time sequence data is power business flow sample data at the current moment or in a preset period;
the edge server is used for receiving the time sequence sample data and the power service class label from the power service data acquisition module, generating a real-time sample set according to the time sequence sample data and the power service class label, updating a pre-trained target power service classification model by utilizing the real-time sample set, and transmitting the updated target power service classification model to the power service identification module.
In one embodiment, the edge server, when executing updating the pre-trained target power traffic classification model with the real-time sample set, is further configured to:
Dividing the real-time sample set into a plurality of mutually exclusive sample subsets, wherein the sample subsets comprise training subsets and verification subsets;
training a pre-trained target power business classification model according to each training subset, and verifying the prediction accuracy of the model obtained after current training through a verification subset corresponding to each training subset;
and taking the model parameters when the prediction precision meets the precision condition as target parameters, and updating the target power business classification model according to the target parameters.
In one embodiment, the system further comprises a cloud data center, wherein the cloud data center is connected with the edge server;
the cloud data center is used for training the initial power business classification model by utilizing the public power business data set to obtain a basic power business classification model;
the edge server is also used for receiving the basic power business classification model from the cloud data center and obtaining a target power business classification model according to the basic power business classification model.
In one embodiment, the edge server, when executing the generation of the target power traffic classification model from the base power traffic classification model, is further configured to:
and training the basic power business classification model by using the private power business data set to obtain a target power business classification model.
In one embodiment, when the power service data preprocessing module performs preprocessing on the power service flow data to obtain the target power service feature data, the power service data preprocessing module is further configured to:
extracting the characteristics of the power service flow data to obtain initial power service characteristic data;
and carrying out normalization processing on the initial power service characteristic data to obtain target power service characteristic data.
In a second aspect, the application further provides a power service identification method. The method comprises the following steps:
the intelligent gateway collects the data of the electric power business flow to be identified;
the intelligent gateway preprocesses the power service flow data to obtain target power service characteristic data;
and the intelligent gateway performs power service identification on the target power service characteristic data by utilizing a pre-trained target power service classification model.
In a third aspect, the present application further provides a computer device cluster. The cluster of computer devices comprises at least one computer device, each computer device comprising a processor and a memory, the memory storing a computer program, the processor of the at least one computer device executing the computer program stored in the memory of the at least one computer device performing the steps of:
The intelligent gateway collects the data of the electric power business flow to be identified;
the intelligent gateway preprocesses the power service flow data to obtain target power service characteristic data;
and the intelligent gateway performs power service identification on the target power service characteristic data by utilizing a pre-trained target power service classification model.
In a fourth aspect, the present application also provides a computer-readable storage medium. The computer readable storage medium having stored thereon a computer program which when executed by a cluster of computer devices performs the steps of:
the intelligent gateway collects the data of the electric power business flow to be identified;
the intelligent gateway preprocesses the power service flow data to obtain target power service characteristic data;
and the intelligent gateway performs power service identification on the target power service characteristic data by utilizing a pre-trained target power service classification model.
In a fifth aspect, the present application also provides a computer program product. The computer program product comprises a computer program which, when executed by a cluster of computer devices, performs the steps of:
the intelligent gateway collects the data of the electric power business flow to be identified;
the intelligent gateway preprocesses the power service flow data to obtain target power service characteristic data;
And the intelligent gateway performs power service identification on the target power service characteristic data by utilizing a pre-trained target power service classification model.
The system comprises the intelligent gateway, wherein the intelligent gateway is provided with the power service data acquisition module, the power service data preprocessing module and the power service recognition module, and the power service data acquisition module, the power service data preprocessing module and the power service recognition module are sequentially connected. The power business data acquisition module is used for acquiring power business flow data to be identified; the power business data preprocessing module is used for receiving the power business flow data from the power business data acquisition module and preprocessing the power business flow data to obtain target power business characteristic data; the power business identification module is used for receiving the target power business characteristic data from the power business data preprocessing module and carrying out power business identification on the target power business characteristic data by utilizing a pre-trained target power business classification model. According to the embodiment of the application, through the intelligent gateway-based power service identification system, the intelligent gateway is utilized to accelerate the power service flow data acquisition and preprocessing, the intelligent gateway can be internally provided with the intelligent network card, the power service data acquisition module can complete the high-speed acquisition of mass power service flow by utilizing the high-performance network interface of the intelligent network card, and higher bandwidth, lower delay and larger throughput are provided; the power business data preprocessing module can utilize the intelligent network card to realize preprocessing of flow data, and load data processing tasks from the CPU to the intelligent network card, so that queuing delay of an I/O bus of a system is reduced, network performance is improved, and power business identification is accelerated. According to the embodiment of the application, the intelligent gateway is introduced to accelerate the acquisition and preprocessing of the power service data, the power service is timely identified, the running states of node resources and equipment in each layer in the power grid system are obtained, and all service flows in the network can be controlled and managed in a layered manner and allocated according to different service types, so that the utilization rate of network resources is improved to the greatest extent.
Drawings
FIG. 1 is a block diagram of a power service identification system in one embodiment;
FIG. 2 is a flow diagram of a classification function in one embodiment;
FIG. 3 is a block diagram of a power service identification system in another embodiment;
FIG. 4 is a flow chart of incremental learning in another embodiment;
FIG. 5 is a flowchart of a power service identification method according to another embodiment;
fig. 6 is an internal structural diagram of a computer device in one embodiment.
Detailed Description
The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. The terminology used in the following examples is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used in the specification of the application and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include, for example, "one or more" such forms of expression, unless the context clearly indicates to the contrary. It should also be understood that in embodiments of the present application, "one or more" means one, two, or more than two; "and/or", describes an association relationship of the association object, indicating that three relationships may exist; for example, a and/or B may represent: a alone, a and B together, and B alone, wherein A, B may be singular or plural. The character "/" generally indicates that the context-dependent object is an "or" relationship.
Reference in the specification to "one embodiment" or "some embodiments" or the like means that a particular feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the application. Thus, appearances of the phrases "in one embodiment," "in some embodiments," "in other embodiments," and the like in the specification are not necessarily all referring to the same embodiment, but mean "one or more but not all embodiments" unless expressly specified otherwise. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless expressly specified otherwise.
The plurality of the embodiments of the present application is greater than or equal to two. It should be noted that, in the description of the embodiments of the present application, the terms "first," "second," and the like are used for distinguishing between the descriptions and not necessarily for indicating or implying a relative importance, or alternatively, for indicating or implying a sequential order.
The current power communication network has more and more power service data flows, the timeliness is strong, the flow is large, the IP is high, the power service with high communication index requirements easily brings huge impact to the network, the congestion of network nodes is caused, the network balance is disturbed, and the requirements of the power system service on the information transmission performance are difficult to meet.
The power service identification system provided by the embodiment of the application comprises an intelligent gateway, wherein the intelligent gateway is used for acquiring service flow data of the power intelligent terminal and carrying out power service identification according to the service flow data.
The power service identification refers to a technology for determining a power service type and a state thereof according to characteristics of the power service during network transmission. By means of the identification and classification of the service flow, the running states of node resources and equipment in each layer in the power grid system are obtained, a reference basis is provided for network management personnel to deploy a service quality control mechanism, and the network management personnel can conduct hierarchical control management and resource allocation on all the service flow in the network according to different service types.
Fig. 1 schematically illustrates a power service identification system provided by an embodiment of the present application, where, as shown in fig. 1, the system includes an intelligent gateway 100, and the system may further include a power intelligent terminal, also referred to as a power intelligent device 200, where the intelligent gateway 100 is communicatively connected to the power intelligent device 200. The smart gateway 100 is a core node in a power communication network or an internet of things system, and may connect the power smart devices 200 to the internet or a power center server and is responsible for communication and data transmission between the power smart devices 200. The power intelligent device 200 refers to various intelligent devices applied to a power system for ensuring stable operation of a power grid, and can be, but not limited to, various intelligent electric meters, intelligent switch cabinets, intelligent alarms, power grid monitoring devices, power grid control devices, intelligent sensing devices and the like.
It should be understood that the number of intelligent gateways and the number of intelligent power devices in the power service identification system are not limited to the above examples.
In an embodiment of the present application, the intelligent gateway 100 and the electric intelligent device 200 may communicate through a communication protocol or interface, such as Zigbee, wiFi, bluetooth, and the like. The intelligent gateway 100 can realize the functions of controlling, monitoring state, collecting data and the like of the power intelligent device 200 through the communication protocol.
Specifically, the intelligent gateway 100 of the embodiment of the present application is configured to collect a power service flow packet of the power intelligent device 200, where the power service flow packet includes flow data of at least one power service, process and classify and identify the flow data of the at least one power service, and determine a power service type.
Referring to fig. 1, the intelligent gateway 100 is provided with a power service data acquisition module 102, a power service data preprocessing module 104 and a power service identification module 106, where the power service data acquisition module 102, the power service data preprocessing module 104 and the power service identification module 106 are sequentially connected.
Referring to fig. 1, the power service data acquisition module 102 is configured to acquire power service flow data to be identified.
The power service flow data refers to service data flow generated among all nodes in the power system, and comprises power utilization data, power generation data, power transmission and distribution data, metering data, relay protection data and the like.
The intelligent gateway 100 of the embodiment of the present application accesses the power intelligent devices 200 in the power system, and obtains the status data and real-time data of these devices and systems through a communication manner.
For example, the power intelligent device 200 connected to the intelligent gateway 100 includes a transformer substation, a sensor in a power distribution station, or an automation device, and the power service data acquisition module 102 on the intelligent gateway 100 is used for acquiring status data or monitoring data of the device.
For another example, an OpenFlow protocol is deployed in the intelligent gateway 100, and the power service data acquisition module 102 on the intelligent gateway 100 realizes online service data acquisition of a large number of heterogeneous power intelligent devices through the OpenFlow protocol.
In the embodiment of the application, the to-be-identified power traffic data may include one type of to-be-identified power traffic data, or the to-be-identified power traffic data may further include multiple types of to-be-identified power traffic data.
The intelligent gateway 100 of the embodiment of the application can also be configured with an intelligent network card, wherein the intelligent network card is a network card with a programmable chip, can customize the processing process of a network packet, and can flexibly unload the task of a general CPU.
For example, depending on the type of core programmable Processor, an intelligent network card designed based on an FPGA (Field-Programmable Gate Array), MP (Multi-Processor), or ASIC (Application-Specific Integrated Circuit) may be configured within the intelligent gateway 100.
For another example, depending on the type of data path, an intelligent network card designed based on an on-path or off-path may be configured within the intelligent gateway 100.
With the unbalanced development of network technology, storage technology and chip design and manufacturing technology, the traditional power system server CPU is deficient in speed and computing power. For example, a CPU is suitable for handling serial complex instruction operations, is not suitable for computation in a large number of parallel fixed patterns, and the overhead of data movement by accessing memory by the CPU is a significant proportion of many applications, such as 40% overhead in fast fourier transform (fast Fourier transform, FFT) computation.
The intelligent network card configured in the intelligent gateway 100 of the embodiment of the application belongs to the architecture design of application drive, and is oriented to the application of a lower layer, and the intelligent network card comprises an industry standard, high-performance and software programmable multi-core computing unit which can work together with a system on chip SoC (System ona Chip) component; the system also comprises a high-performance network interface which can analyze and process data at linear speed and efficiently finish the task of data forwarding; also included are various flexible and programmable acceleration engines that can offload applications such as artificial intelligence AI (ArtificialIntelligence), machine learning, security, telecommunications, and storage to achieve higher performance; and the method further comprises virtualized high-speed I/O channels and fine-granularity time division multiplexing, so that I/O queuing delay is reduced, and the network traffic acquisition and analysis work is completed at a high speed.
Therefore, the power service data acquisition module 102 of the embodiment of the present application can complete high-speed acquisition of power service traffic through the OpenFlow protocol deployed in the intelligent gateway and by using the high-performance network interface of the intelligent network card, and provides higher bandwidth, lower delay and larger throughput.
Referring to fig. 1, the power service data preprocessing module 104 is configured to receive power service flow data from the power service data acquisition module 102, and preprocess the power service flow data to obtain target power service feature data.
The target power business feature data refers to feature data input into a target power business classification model and used for power business identification.
The target power traffic characteristic data of embodiments of the present application may include one or more power traffic characteristics.
For example, the target power traffic profile data may include latency requirement profile data.
As another example, the target power traffic profile data may also include TTL (Time to Live) profile data, latency requirement profile data, duration profile data, resource requirement profile data, and latency jitter requirement profile data.
The embodiment of the application is not limited to the preprocessing mode of the power business flow data, and the target power business feature data can be obtained by directly extracting the features of the power business flow data, or can be obtained by arranging the power business flow data or arranging the data after the feature extraction.
For example, the power service data preprocessing module 104 is configured to perform preliminary cleaning, filtering, interpolation on the power service flow data, and then perform feature extraction to obtain target power service feature data.
For another example, the power service data preprocessing module 104 is configured to perform feature extraction on the power service traffic data, then perform data reconstruction, determine formatting data applicable to the target power service classification model, and obtain target power service feature data.
It can be appreciated that when the power traffic data to be identified includes a plurality of types, the power traffic data is preprocessed for each power traffic data to be identified, so as to obtain target power traffic characteristic data corresponding to the power traffic data.
The power service data preprocessing module 104 of the embodiment of the application can also realize preprocessing of flow data through an intelligent network card built in the intelligent gateway 100, and comprises intelligent functions of flow data analysis and filtration, flow control, load balancing and the like, and load data processing tasks from a CPU to the intelligent network card, so that queuing waiting time delay of an I/O bus of a system is reduced, and network performance is improved.
Referring to fig. 1, the power service identification module 106 is configured to receive the target power service characteristic data from the power service data preprocessing module 104, and perform power service identification on the target power service characteristic data by using a pre-trained target power service classification model.
The pre-trained target power business classification model may be a classification model using a convolutional neural network (Convolutional Neural Network, CNN) algorithm, for example, a CNN model, or may be another model based on a model.
After receiving the target power service feature data, the power service identification module 106 of the embodiment of the application inputs the target power service feature data into a pre-trained target power service classification model, and outputs a power service type corresponding to the target power service feature data.
The above power service identification system comprises an intelligent gateway 100, wherein a power service data acquisition module 102, a power service data preprocessing module 104 and a power service identification module 106 are disposed on the intelligent gateway, and the power service data acquisition module 102, the power service data preprocessing module 104 and the power service identification module 106 are sequentially connected. The power business data acquisition module 102 is used for acquiring power business flow data to be identified; the power business data preprocessing module 104 is configured to receive power business flow data from the power business data acquisition module 102, and preprocess the power business flow data to obtain target power business feature data; the power service identification module 106 is configured to receive the target power service feature data from the power service data preprocessing module 104, and perform power service identification on the target power service feature data by using a pre-trained target power service classification model. According to the embodiment of the application, through the power service identification system based on the intelligent gateway 100, the intelligent gateway 100 is utilized to accelerate the power service flow data acquisition and preprocessing, the intelligent gateway 100 can be internally provided with the intelligent network card, the power service data acquisition module 102 can complete the high-speed acquisition of mass power service flow by utilizing the high-performance network interface of the intelligent network card, and higher bandwidth, lower delay and larger throughput are provided; the power business data preprocessing module 104 can utilize the intelligent network card to preprocess the flow data, and offload the data processing task from the CPU to the intelligent network card, thereby reducing the queuing delay of the system I/O bus, improving the network performance and accelerating the power business recognition.
According to the embodiment of the application, the intelligent gateway 100 is introduced to accelerate the collection and preprocessing of the power service data, and the operation states of node resources and equipment of each layer in a power grid system are acquired by means of the identification and classification of the service flow, so that all the service flow in the network can be controlled and managed in a layering manner according to different service types, the resource scheduling and controlling capacity of the existing power data communication network is improved, the flexibility, the safety and the real-time performance of the power data communication network can be adapted to the change brought by the construction and development of the power grid, the power communication network resources are optimally distributed according to a given strategy, different service quality grades are provided for different types of application, and the refined management and the dynamic control of the service flow are realized, so that the network congestion is avoided or reduced, the operation stability of the power system under high communication load is improved, the smooth operation of the network is maintained, the multi-service high-efficiency bearing capacity of the network is comprehensively optimized and expanded, and the utilization rate of the network resources is furthest improved.
In one embodiment, the power service data preprocessing module 104 is further configured to, when performing preprocessing on the power service flow data to obtain the target power service feature data:
And step A1, extracting the characteristics of the power service flow data to obtain initial power service characteristic data.
And step A2, carrying out normalization processing on the initial power business characteristics to obtain target power business characteristic data.
In the embodiment of the application, each power service may include a plurality of feature data such as TTL (Time to Live) feature data, time delay requirement feature data, duration feature data, resource requirement feature data, time delay jitter requirement feature data, and the like, where the plurality of feature data may not be in the same order of magnitude, and are difficult to be compared with each other.
In order to better realize feature modeling and service identification, the embodiment of the application performs normalization processing on the initial power service feature data after the feature extraction, and the feature data is in the same order of magnitude through normalization operation.
Illustratively, n power services are provided, which may be represented as apps i I=1, 2, …, n, assuming that there are m power traffic characteristics per power traffic, the j power traffic characteristics of the i power traffic can be expressed as app i,j J=1, 2, …, m. Let the matrix Fe represent a set of n power traffic and m power traffic characteristics.
In the embodiment of the application, the power business characteristics are normalized by using a min-max method, and the normalization is performed as follows:
wherein, let the matrix Fe 'represent a normalized set of n×m power service eigenvalues, the eigenvalues being between 0 and 1, i.e. af' i，j ∈(0,1)。
The power service data preprocessing module 104 in this embodiment is configured to perform feature extraction on power service flow data to obtain initial power service feature data, perform normalization processing on the initial power service feature to obtain target power service feature data, and enable the power service feature to be in the same order of magnitude through normalization operation, so as to achieve better power service identification.
In one embodiment, the power service identification system further includes an edge server, which is connected to the intelligent gateway 100.
In an embodiment of the present application, the intelligent gateway 100 may obtain a pre-trained target power service classification model from an edge server. The target power service classification model may be trained by the edge server and then transmitted to the power service identification module 106 of the intelligent gateway 100, or the target power service classification model may be trained by other data platforms and then transmitted to the edge server and then transmitted to the power service identification module 106 of the intelligent gateway 100 by the edge server.
The power service data acquisition module 102 of the embodiment of the present application is further configured to acquire time-series sample data of a power service and a power service class label.
The time series sample data may refer to real-time power traffic sample data at or at the current time in the time series, or the time series sample data may also refer to real-time power traffic sample data within a predetermined period, where the predetermined period may include a plurality of time domains.
For example, the time series may include a plurality of power traffic sample data collected from the power smart device 200 at a fixed frequency, and the time series sample data may refer to data at a current time in the time series, or the time series sample data may refer to data of a certain period or a plurality of periods in the time series.
For another example, the time series may also include a plurality of power traffic sample data collected from the power intelligent device 200 at a non-fixed frequency, where the time series sample data may refer to data at a current time in the time series, or the time series sample data may refer to data of a certain period or a plurality of periods in the time series.
It will be appreciated that the time series sample data may include power traffic sample data corresponding to one or more types of power traffic, and the time series sample data may also include power traffic sample data corresponding to one type of power traffic at different times or different time domains.
The power service class label in the embodiment of the application can refer to a label corresponding to each type of power service flow sample data.
In addition, the method for acquiring the power service class label is not limited in the embodiment of the present application, for example, the power service class label may be determined by the power intelligent device 200, the power service data acquisition module 102 of the intelligent gateway 100 directly acquires the power service class label, or the power service class label may be determined by other modules configured by the intelligent gateway 100.
The edge server of the embodiment of the application is used for receiving the time series sample data and the power business class label from the power business data acquisition module 102, generating a real-time sample set according to the time series sample data and the power business class label, updating a pre-trained target power business classification model by utilizing the real-time sample set, and transmitting the updated target power business classification model to the power business identification module.
When the edge server generates the real-time sample set according to the time-series sample data and the power service class label, the edge server includes preprocessing the time-series sample data, such as data cleaning, preprocessing or formatting, wherein the formatting includes normalizing the sample features extracted from the time-series sample data, and the normalizing may refer to the process of the power service data preprocessing module 104 in step A2 to normalize the initial power service features to obtain the target power service feature data.
It should be noted that, when the time series sample data includes the power service flow sample data corresponding to one type of power service at different time instants or different time domains, the corresponding samples are sequentially used to train the target power service classification model according to the time ascending order, so as to update the target power service classification model.
In this embodiment, the power service data collection module 102 is further configured to obtain time-series sample data and a power service class label of the power service, and the edge server is configured to receive the time-series sample data and the power service class label from the power service data collection module 102, generate a real-time sample set according to the time-series sample data and the power service class label, update a pre-trained target power service classification model with the real-time sample set, and transmit the updated target power service classification model to the power service identification module 106. Because of the continuous development and the intellectualization of the power system technology, the continuous change of the power business demands, the continuous updating and iteration of the power business data and business types, the embodiment of the application considers the online learning demands of the target power business classification model, utilizes the high-performance network interface of the intelligent gateway to complete the high-speed acquisition of the power business flow sample data and the lossless forwarding to the edge server, and based on the incremental power business data, the edge side model is frequently updated by an incremental learning method at the edge side so as to finally realize the accuracy iterative promotion and real-time guarantee of the model and prevent the leakage of business information; in addition, through the lossless high-speed data transmission capability of the intelligent network card, the problems of data missing and hysteresis existing in online learning of the traditional artificial intelligent model are solved, further incremental learning of the target power business classification model is assisted to be accelerated, circulation of data between business perception nodes and model training nodes is accelerated, model incremental training speed is improved, and accuracy of power business identification is further improved through real-time iterative updating of the target power business classification model.
In one embodiment, the edge server, when executing updating the pre-trained target power traffic classification model with the real-time sample set, is further configured to:
and B1, dividing the real-time sample set into a plurality of mutually exclusive sample subsets, wherein the sample subsets comprise training subsets and verification subsets.
And step B2, training a pre-trained target power business classification model according to each training subset, and verifying the prediction accuracy of the model obtained after the current training through a verification subset corresponding to each training subset.
And B3, taking the model parameters when the prediction precision meets the precision condition as target parameters, and updating the target power business classification model according to the target parameters.
The real-time sample set is divided into S groups of mutually exclusive sample subsets, each group of sample subsets is used as a verification subset, the other S-1 groups of sample subsets are used as training subsets, the S-1 groups of sample subsets are used for training a pre-trained target power service classification model, the prediction precision of the model obtained after current training is verified through the verification subsets corresponding to the S-1 groups of sample subsets, model parameters when the prediction precision meets precision conditions are used as target parameters, and the target power service classification model is updated according to the target parameters.
It should be noted that, when the time sequence sample data includes the power traffic flow sample data corresponding to one type of power traffic at different time instants or different time domains, that is, the corresponding real-time sample set also includes samples at different time instants or different time domains in time sequence, for each group of sample subsets, the sample subset is taken as a verification subset, the other S-1 groups of sample subsets are taken as training subsets, and samples in the training subsets are sequentially utilized to train the target power traffic classification model according to the time ascending order, so as to update the target power traffic classification model.
In one embodiment, the power service identification system further comprises a cloud data center, and the cloud data center is connected with the edge server.
The cloud data center is used for training the initial power business classification model by utilizing the public power business data set to obtain a basic power business classification model.
The edge server is also used for receiving the basic power business classification model from the cloud data center and obtaining a target power business classification model according to the basic power business classification model.
According to the embodiment of the application, firstly, an electric power business classification model is initialized in a cloud data center to obtain an initial electric power business classification model, then the public electric power business data set of the cloud data center is utilized to train the initial electric power business classification model to obtain a basic electric power business classification model, and the basic electric power business classification model is sent to an edge server.
The edge server can transmit the basic power service classification model to the intelligent gateway as a pre-trained target power service classification model, or the edge server can also utilize a migration learning mechanism to re-train the basic power service classification model through a private power service data set to obtain the pre-trained target power service classification model.
The cloud data center of the embodiment trains the initial power business classification model by using the public power business data set to obtain a basic power business classification model, and the edge server is further used for receiving the basic power business classification model from the cloud data center, obtaining a target power business classification model according to the basic power business classification model, obtaining a pre-trained target power business classification model, and transmitting the pre-trained target power business classification model to the power business identification module 106 of the intelligent gateway, so that the power business identification module 106 can perform power business identification based on the pre-trained target power business classification model.
In one embodiment, the initial power service classification model, the basic power service classification model, and the target power service classification model are convolutional neural network models based on CNN, and the basic architecture of the power service classification model adopted in the embodiments of the present application will be described in detail below.
The CNN model of the embodiment of the application comprises a plurality of hidden layers, can be locally connected and has shared parameters, and mainly comprises an activation function, a pooling function, a classification function and a loss function.
Regarding the activation function: in order to improve the calculation efficiency, reLU is used as an activation function, so that the gradient disappearance problem is effectively solved. Let ln i,j Representing the jth neuron belonging to the ith layer, the activation function is represented by:
wherein s (ln i,j ) Is ln i,j In (2) when the stimulus intensity reaches a certain level, ln i,j Is activated.
Regarding the pooling function: the power business data extracted by the embodiment of the application is characterized by high-dimension data, and in order to avoid the complexity of calculation, the embodiment of the application adopts t-SNE instead of Maxout as a pooling function for reducing dimension.
Is provided with af' i，j (1) And af' i ，j (2) Representing any two points in the high-dimensional space, the point mapped to the low-dimensional space is denoted as af i,j (1) And af i,j (2)，af″ i,j (1) And af i,j (2) The conditional probability in the high dimensional space is as follows:
hd 1,2 ＝||af′ i,j (1)-af′ i,j (2)||；
at the same time, a high-dimensional space af' i,j (1) And af' i,j (2) The joint probability distribution function of (2) is shown below.
By using the t distribution rule, af' in the low-dimensional space can be obtained i,j (1) And af i,j (2) The joint probability distribution function of (2) is as follows:
wherein ld is 1,2 Is af i,j (1) And af i,j (2) The Euclidean distance between them, let Cost represent hjpd 1,2 And ljpd 1,2 Differences between, e.gThe following is shown:
and fitting the low-dimensional distribution to the high-dimensional distribution by using a gradient training algorithm, as follows:
let RFe '(I) represent the dimension reduction result of Fe', the iterative equation is as follows:
wherein I, eta and alpha respectively represent iteration times, learning rate and momentum factors, and the situation that local optimum is trapped due to too high convergence speed can be avoided by adjusting the three parameters.
Regarding the classification function, in the embodiment of the present application, softmax is used as the classification function of the output layer, which can effectively process multiple classification problems, and can ensure that the probability of correct classification is higher, the probability of incorrect classification is lower, and the process of the Softmax function is as shown in fig. 2, w i,j Representation af i,j Weights, p i Representing pairs of apps i Wherein the prediction probability of the (c) is determined,
regarding the loss function, the loss function is used to calculate the deviation of the output result of the CNN model from the label result, and then used in the back propagation process to update the gradient. The embodiment of the application adopts a gradient descent algorithm to continuously train and optimize parameters in the CNN model, and aims to minimize a loss function, finally learns to obtain the optimal CNN model, wherein the loss function is as follows:
Wherein K is the total number of samples, lambda k For marking the status of the kth sample.
In one embodiment, as shown in fig. 3, there is provided a power service identification system including: the system comprises a power intelligent information infrastructure layer, a power service identification layer and a service identification training layer, wherein the power intelligent information infrastructure layer comprises at least one power intelligent device; the power service identification layer comprises an intelligent gateway, and the intelligent gateway is provided with a power service data acquisition module, a power service data preprocessing module and a power service identification module; the service identification training layer comprises a cloud data center and an edge server.
The cloud data center is used for training the initial power business classification model by utilizing the public power business data set to obtain a basic power business classification model.
The edge server is used for receiving the basic power business classification model from the cloud data center, and performing migration learning training on the basic power business classification model by utilizing the private power business data set to obtain a pre-trained target power business classification model.
The intelligent gateway is internally provided with an intelligent network card, a power service data acquisition module of the intelligent gateway completes high-speed acquisition of power service flow sample data by using a high-performance network interface of the intelligent network card through an OpenFlow protocol deployed in the intelligent gateway and forwards the power service flow sample data to an edge server in a lossless manner, the edge server receives the power service flow sample data and then generates a real-time sample set, a pre-trained target power service classification model is frequently updated by an incremental learning method on the edge side according to the real-time sample set, and the updated target power service classification model is sent to a power service identification module of the intelligent gateway, so that accuracy iterative promotion and real-time guarantee of the model are realized, and service information leakage is prevented.
The power business data acquisition module is used for acquiring power business flow data to be identified of the power intelligent equipment in real time through a high-performance interface of the intelligent network card; the power business data preprocessing module is used for receiving the power business flow data from the power business data acquisition module, and preprocessing the power business flow data through the intelligent network card to obtain target power business characteristic data; the power business identification module is used for receiving the target power business characteristic data from the power business data preprocessing module and carrying out power business identification on the target power business characteristic data by utilizing the updated target power business classification model.
In combination with fig. 4, the cloud data center may be a cloud, where the cloud may include a cloud storage and a cloud server, where the cloud storage stores a public data set, and the public data set includes a public power service data set. The cloud server pre-trains the initial power business classification model by using the public power business data set to obtain a basic model obtained by cloud training, namely the basic power business classification model.
The edge servers can be one or more, the edge servers are uniformly positioned in the edge layer, the edge layer further comprises edge storage for storing a privacy data set and an incremental data set, wherein the incremental data set refers to time sequence sample data and power service class labels which are acquired and transmitted by the power service data acquisition module, and the time sequence data is power service flow sample data at the current moment or in a preset period.
The first edge server reads the basic power business classification model from the cloud, and performs migration learning on the basic power business classification model by using the privacy data set to obtain a migration-learned edge side model, namely a target power business classification model. The second edge server reads the target power business classification model, performs incremental learning training on the target power business classification model through an incremental data set to obtain an edge side model after incremental learning, namely an updated target power business classification model, and then sends the updated target power business classification model to the intelligent gateway.
According to the embodiment, the electric power business identification system based on the intelligent gateway is utilized, firstly, electric power business data to be identified are collected and preprocessed, the data processing unit of the intelligent network card is utilized, the system I/O preemption time delay is reduced, the extraction and normalization of different electric power business characteristics are accelerated, the electric power business identification business is unloaded from the CPU to the intelligent gateway, the I/O operation is reduced, the performance deficiency of the CPU in the aspect of processing data packets is made up, and the electric power business identification rate is improved; in the second aspect, the model is directly led into the intelligent gateway, so that the power service identification accuracy is improved, the reasoning speed of the power service identification model is increased, and the defects of poor identification accuracy, low identification speed, weak safety and the like caused by the fact that the traditional power service identification relies on port binding and depth data packet detection are overcome; in the third aspect, the cloud and the server are utilized to cooperate with the online incremental learning of the power service, the high-performance network interface of the intelligent gateway is utilized to forward the power service data to the edge side server in real time, the incremental learning of the power service identification model is completed, and further accuracy iterative promotion and real-time guarantee are achieved.
The above-described individual modules in the power service identification system may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
Based on the same inventive concept, the embodiment of the application also provides a power service identification method which can be applied to the power service identification system. The implementation of the solution to the problem provided by the method is similar to that described in the above system, so the specific limitation in one or more embodiments of the power service identification method provided below may be referred to the limitation of the power service identification system hereinabove, and will not be described herein.
The power service identification method provided by the embodiment of the application can be applied to a power service identification system shown in fig. 1, wherein the power service identification system comprises an intelligent gateway 100, and the power service identification system also comprises a power intelligent terminal, which is also called as a power intelligent device 200. The smart gateway 100 is communicatively connected to the power smart device 200. The smart gateway 100 is a core node in a power communication network or an internet of things system, and may connect the power smart devices 200 to the internet or a power center server and is responsible for communication and data transmission between the power smart devices 200. The power intelligent device 200 refers to various intelligent devices applied to a power system for ensuring stable operation of a power grid, and can be, but not limited to, various intelligent electric meters, intelligent switch cabinets, intelligent alarms, power grid monitoring devices, power grid control devices, intelligent sensing devices and the like.
Based on the system architecture shown in fig. 1, an embodiment of the present application provides a power service identification method, as shown in fig. 5, where the flow of the method may be executed by the intelligent gateway in fig. 1, and the method includes the following steps:
step 502, the intelligent gateway collects the data of the electric power business flow to be identified.
And step 504, the intelligent gateway preprocesses the power service flow data to obtain target power service characteristic data.
And step 506, the intelligent gateway performs power service identification on the target power service characteristic data by using a pre-trained target power service classification model.
In one embodiment, the method further comprises: the power business data acquisition module acquires time sequence sample data of power business and a power business class label, wherein the time sequence data is power business flow sample data at the current moment or in a preset period. The edge server receives the time sequence sample data and the power service class label from the power service data acquisition module, generates a real-time sample set according to the time sequence sample data and the power service class label, updates a pre-trained target power service classification model by using the real-time sample set, and transmits the updated target power service classification model to the power service identification module.
In one embodiment, updating a pre-trained target power business classification model with a real-time sample set includes: dividing the real-time sample set into a plurality of mutually exclusive sample subsets, wherein the sample subsets comprise training subsets and verification subsets; training a pre-trained target power business classification model according to each training subset, and verifying the prediction accuracy of the model obtained after current training through a verification subset corresponding to each training subset; and taking the model parameters when the prediction precision meets the precision condition as target parameters, and updating the target power business classification model according to the target parameters.
In one embodiment, the method further comprises: the cloud data center is used for training the initial power business classification model by utilizing the public power business data set to obtain a basic power business classification model; the edge server is also used for receiving the basic power business classification model from the cloud data center and obtaining a target power business classification model according to the basic power business classification model.
In one embodiment, generating the target power traffic classification model from the base power traffic classification model comprises: and training the basic power business classification model by using the private power business data set to obtain a target power business classification model.
In one embodiment, preprocessing the power service flow data to obtain target power service feature data includes: extracting the characteristics of the power service flow data to obtain initial power service characteristic data; and carrying out normalization processing on the initial power service characteristic data to obtain target power service characteristic data.
It should be understood that, although the steps in the flowcharts related to the embodiments described above are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.
In one embodiment, a cluster of computer devices is provided, the cluster of computer devices including at least one computer device, which may be a server, the internal structure of which may be as shown in FIG. 6. The computer device includes a processor, a memory, an Input/Output interface (I/O) and a communication interface. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface is connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is used to store weather data and thermal load data. The input/output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used for communicating with an external terminal through a network connection. The computer program, when executed by the processor, implements a method for predicting a thermal load of a heating system or a model training method.
It will be appreciated by those skilled in the art that the structure shown in FIG. 6 is merely a block diagram of some of the structures associated with the present inventive arrangements and is not limiting of the computer device to which the present inventive arrangements may be applied, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
In one embodiment, a cluster of computer devices is provided, the cluster of computer devices comprising at least one computer device, each computer device comprising a processor and a memory, the memory storing a computer program; the steps of the method embodiments described above are implemented by a processor of at least one computer device executing a computer program stored in a memory of the at least one computer device.
In one embodiment, a computer readable storage medium is provided, on which a computer program is stored which, when executed by a cluster of computer devices, implements the steps of the method embodiments described above.
In an embodiment, a computer program product is provided, comprising a computer program which, when executed by a cluster of computer devices, implements the steps of the method embodiments described above.
It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or sufficiently authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards of the related country and region.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the embodiments provided herein may include at least one of a relational database and a non-relational database. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processor referred to in the embodiments provided in the present application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, or the like, but is not limited thereto.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples illustrate only a few embodiments of the application and are described in detail herein without thereby limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of the application should be assessed as that of the appended claims.

Claims (10)

1. The power service identification system is characterized by comprising an intelligent gateway, wherein a power service data acquisition module, a power service data preprocessing module and a power service identification module are arranged on the intelligent gateway, and the power service data acquisition module, the power service data preprocessing module and the power service identification module are connected in sequence;
The power business data acquisition module is used for acquiring power business flow data to be identified;
the power business data preprocessing module is used for receiving the power business flow data from the power business data acquisition module and preprocessing the power business flow data to obtain target power business characteristic data;
the power business identification module is used for receiving the target power business characteristic data from the power business data preprocessing module and carrying out power business identification on the target power business characteristic data by utilizing a pre-trained target power business classification model.
2. The system of claim 1, further comprising an edge server, the edge server being connected to the intelligent gateway;
the power business data acquisition module is further used for acquiring time sequence sample data of power business and a power business class label, wherein the time sequence data is power business flow sample data at the current moment or in a preset period;
the edge server is configured to receive the time-series sample data and the power service class label from the power service data acquisition module, generate a real-time sample set according to the time-series sample data and the power service class label, update the pre-trained target power service classification model by using the real-time sample set, and transmit the updated target power service classification model to the power service identification module.
3. The system of claim 2, wherein the edge server, when executing updating the pre-trained target power traffic classification model with the real-time sample set, is further configured to:
dividing the real-time sample set into a plurality of mutually exclusive sample subsets, wherein the sample subsets comprise training subsets and verification subsets;
training the pre-trained target power business classification model according to each training subset, and verifying the prediction accuracy of the model obtained after current training through the verification subset corresponding to each training subset;
and taking the model parameters when the prediction precision meets the precision condition as target parameters, and updating the target power business classification model according to the target parameters.
4. The system of claim 2, further comprising a cloud data center, the cloud data center being connected to the edge server;
the cloud data center is used for training the initial power business classification model by utilizing the public power business data set to obtain a basic power business classification model;
the edge server is further used for receiving the basic power business classification model from the cloud data center and obtaining the target power business classification model according to the basic power business classification model.
5. The system of claim 4, wherein the edge server, when executing generating the target power traffic classification model from the base power traffic classification model, is further configured to:
and training the basic power business classification model by using a private power business data set to obtain the target power business classification model.
6. The system of claim 1, wherein the power traffic data preprocessing module, when performing preprocessing on the power traffic flow data to obtain target power traffic feature data, is further configured to:
extracting the characteristics of the power service flow data to obtain initial power service characteristic data;
and normalizing the initial power business characteristic data to obtain the target power business characteristic data.
7. A power service identification method, characterized by being applied to the power service identification system according to any one of claims 1 to 6, the method comprising:
the intelligent gateway collects the data of the electric power business flow to be identified;
the intelligent gateway preprocesses the power service flow data to obtain target power service characteristic data;
And the intelligent gateway performs power service identification on the target power service characteristic data by utilizing a pre-trained target power service classification model.
8. A cluster of computer devices comprising at least one computer device, each computer device comprising a processor and a memory, the memory storing a computer program, characterized in that the processor of the at least one computer device implements the steps of the method of claim 7 when executing the computer program stored in the memory of the at least one computer device.
9. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a cluster of computer devices, implements the steps of the method of claim 7.
10. A computer program product comprising a computer program which, when executed by a cluster of computer devices, implements the steps of the method of claim 7.
CN202310596642.1A 2023-05-24 2023-05-24 Power service identification system, method, device, medium and product Pending CN116720132A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310596642.1A CN116720132A (en) 2023-05-24 2023-05-24 Power service identification system, method, device, medium and product

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310596642.1A CN116720132A (en) 2023-05-24 2023-05-24 Power service identification system, method, device, medium and product

Publications (1)

Publication Number Publication Date
CN116720132A true CN116720132A (en) 2023-09-08

Family

ID=87874314

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310596642.1A Pending CN116720132A (en) 2023-05-24 2023-05-24 Power service identification system, method, device, medium and product

Country Status (1)

Country Link
CN (1) CN116720132A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117557108A (en) * 2024-01-10 2024-02-13 中国南方电网有限责任公司超高压输电公司电力科研院 Training method and device for intelligent identification model of power operation risk
CN118037282A (en) * 2024-04-15 2024-05-14 华中科技大学 Intelligent power business data processing system and method based on AI and cloud technology

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117557108A (en) * 2024-01-10 2024-02-13 中国南方电网有限责任公司超高压输电公司电力科研院 Training method and device for intelligent identification model of power operation risk
CN118037282A (en) * 2024-04-15 2024-05-14 华中科技大学 Intelligent power business data processing system and method based on AI and cloud technology

Similar Documents

Publication Publication Date Title
Lv et al. AI empowered communication systems for intelligent transportation systems
Yu et al. Intelligent edge: Leveraging deep imitation learning for mobile edge computation offloading
Ali et al. A data aggregation based approach to exploit dynamic spatio-temporal correlations for citywide crowd flows prediction in fog computing
CN113282368B (en) Edge computing resource scheduling method for substation inspection
Dong et al. NOMA-based energy-efficient task scheduling in vehicular edge computing networks: A self-imitation learning-based approach
Habibi et al. Accelerating virtual network embedding with graph neural networks
Shadroo et al. The two-phase scheduling based on deep learning in the Internet of Things
Liu et al. Fine-grained flow classification using deep learning for software defined data center networks
Rabie et al. A new outlier rejection methodology for supporting load forecasting in smart grids based on big data
CN116720132A (en) Power service identification system, method, device, medium and product
CN110162390A (en) A kind of method for allocating tasks and system of mist computing system
Ko et al. Network prediction with traffic gradient classification using convolutional neural networks
Rahbari et al. Fast and fair computation offloading management in a swarm of drones using a rating-based federated learning approach
CN115603446A (en) Power distribution station area operation monitoring system based on convolution neural network and cloud edge synergistic effect
Yan et al. TL-CNN-IDS: transfer learning-based intrusion detection system using convolutional neural network
CN112925964A (en) Big data acquisition method based on cloud computing service and big data acquisition service system
Khoa et al. SplitDyn: Federated split neural network for distributed edge AI applications
WO2023029944A1 (en) Federated learning method and device
CN115208892B (en) Vehicle-road collaborative online task scheduling method and system based on dynamic resource demand
Alam Recurrent neural networks in electricity load forecasting
Lai et al. Short‐term passenger flow prediction for rail transit based on improved particle swarm optimization algorithm
CN116109058A (en) Substation inspection management method and device based on deep reinforcement learning
Liu et al. Efficient multi-user for task offloading and server allocation in mobile edge computing systems
Du et al. OctopusKing: A TCT-aware task scheduling on spark platform
Altieri et al. Scalable forecasting in sensor networks with graph convolutional lstm models

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination