Method and device for providing multi-subject personalized power data service

Abstract

The invention discloses a multi-subject personalized power data service providing method and device, and relates to the technical field of power grid data value addition. According to the main technical scheme, different types of algorithm libraries are constructed according to different artificial intelligence learning modes; taking each model algorithm as an algorithm template in the algorithm library, and correspondingly associating at least one function scene, so that the model algorithms are used as the algorithm templates to train corresponding models for providing scene services on the corresponding function scenes; configuring at least one power data service mode for different subjects; searching a matched target function scene from at least one function scene associated with a model algorithm of an algorithm library according to a service demand corresponding to the power data service mode; and performing auxiliary execution of providing a scene service corresponding to the target function scene by using the model algorithm associated with the target function scene in the algorithm library, and applying to establishing an application service meeting a service demand corresponding to the power data service mode.

Description

Technical Field

[0001] The present application relates to the technical field of power grid data value-added, and in particular to a method and device for providing multi-agent personalized power data services. Background Art

[0002] The electricity market connects all aspects of energy production, transmission, and consumption, and covers multiple areas such as market transactions, power grid operation, electricity production, user energy consumption, water conditions and meteorology, and financial services. Not only are the data sources extensive, but the data scale is also extremely large.

[0003] The current data information of the power market includes both business data of various professions within the market and information from the external network society; from the perspective of time scale, it not only covers the massive historical data of the past ten years, but also includes the market operation data generated in real time; from the perspective of space, it includes both micro-market data such as distributed energy and transaction data of intra-provincial and inter-provincial markets. The overall data architecture has evolved from determinism to time-varying, and the abstract relationship between entities has also increased the dynamic complexity in time and space.

[0004] Therefore, based on the wide coverage of electricity data, it will be used by more market players. How to provide personalized demand services to different market players is an urgent problem to be solved. Summary of the invention

[0005] In view of this, the present application provides a method and device for providing personalized power data services for multiple subjects. The main purpose is to use the constructed algorithm library and the model algorithm contained therein as a model template to assist in realizing the scenario services of functional scenarios, so as to meet the service needs of the power data service mode of different subjects. It can quickly and accurately recommend available model templates, and based on these model templates, assist in the execution of the scenario services corresponding to the target functional scenarios, so that a larger service set can be formed based on multiple scenario services, which is used to meet the application services corresponding to the service needs of the power data service mode. In the above, in response to the personalized needs of different subjects, the present application provides an efficient and service-diverse solution.

[0006] In order to achieve the above objectives, this application mainly provides the following technical solutions:

[0007] A first aspect of the present application provides a method for providing a multi-agent personalized power data service, the method comprising:

[0008] According to different artificial intelligence learning methods, different types of algorithm libraries are constructed, including at least: supervised learning library, unsupervised learning library, reinforcement learning library, and deep learning library; in the algorithm library, each model algorithm is used as an algorithm template, and is correspondingly associated with at least one functional scenario, so that the model algorithm is used as an algorithm template to train the corresponding model, which is used to provide scenario services in the corresponding functional scenario;

[0009] At least one power data service mode is configured for different subjects, and at least one service demand is configured for the power data service mode, and the subjects include subjects in the power market and subjects outside the power market;

[0010] According to the service requirement corresponding to the power data service mode, searching for a matching target functional scenario from at least one functional scenario associated with the model algorithm in the algorithm library;

[0011] The model algorithm associated with the target functional scenario in the algorithm library is used to assist in the execution of providing the scenario service corresponding to the target functional scenario, which is applied to forming an application service that meets the service requirements corresponding to the power data service model.

[0012] A second aspect of the present application provides a device for providing a multi-agent personalized power data service, the device comprising:

[0013] A construction unit is used to construct different types of algorithm libraries according to different artificial intelligence learning methods, including at least: supervised learning library, unsupervised learning library, reinforcement learning library, and deep learning library; in the algorithm library, each model algorithm is used as an algorithm template, and is correspondingly associated with at least one functional scenario, so that the model algorithm is used as an algorithm template to train the corresponding model, so as to provide scenario services on the corresponding functional scenario;

[0014] A configuration unit, configured to configure at least one power data service mode for different subjects, wherein the power data service mode corresponds to configuring at least one service demand, and the subjects include subjects in the power market and subjects outside the power market;

[0015] A searching unit, configured to search for a matching target functional scenario from at least one functional scenario associated with the model algorithm in the algorithm library according to the service requirement corresponding to the power data service mode;

[0016] An execution unit is used to utilize the model algorithm associated with the target function scenario in the algorithm library to assist in the execution of the scenario service corresponding to the target function scenario, and is used to form an application service that meets the service requirements corresponding to the power data service model.

[0017] A third aspect of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for providing a multi-agent personalized power data service as described above is implemented.

[0018] In a fourth aspect, the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the method for providing multi-subject personalized power data services as described above is implemented.

[0019] By means of the above technical solution, the technical solution provided by this application has at least the following advantages:

[0020] The present application provides a method and device for providing personalized power data services for multiple subjects. The present application pre-constructs different types of algorithm libraries according to different artificial intelligence learning methods, and each model algorithm in each type of algorithm library will be used as an algorithm template. The algorithm template can be applied to train a model to provide a scene service on a specified functional scene, so that a model algorithm corresponding to at least one functional scene is obtained based on the construction of the algorithm library, and a functional scene can provide the corresponding scene service. Therefore, the present application adds a fine-grained division of the achievable functional scenes when constructing the algorithm library. The higher the fine-grained division, the finer and more functional scenes can be realized, so as to obtain more rich and diverse scene services. Based on this, in response to the service requirements of the power data service model of different subjects of the present application, the model algorithm corresponding to each target functional scene can be quickly found from the algorithm library according to the multiple target functional scenes involved in the service requirements. This is equivalent to quickly and accurately recommending the available model templates to assist in the execution of the scene service corresponding to the target functional scene, so that a larger service set can be formed based on multiple scene services to meet the application service corresponding to the service requirements of the power data service model. As mentioned above, in response to the personalized needs of different entities, this application provides an efficient and service-diverse solution.

[0021] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0023] Figure 1 A flow chart of a method for providing a multi-agent personalized power data service provided in an embodiment of the present application;

[0024] Figure 2 The supervised learning library exemplified in the embodiment of this application;

[0025] Figure 3 The unsupervised learning library exemplified in the embodiment of the present application;

[0026] Figure 4 The reinforcement learning library exemplified in the embodiments of this application;

[0027] Figure 5 A deep learning library exemplified in the embodiments of this application;

[0028] Figure 6 A schematic diagram of the power market user portrait system architecture provided in an embodiment of the present application;

[0029] Figure 7 A block diagram of a device for providing a multi-agent personalized power data service provided in an embodiment of the present application;

[0030] Figure 8 A block diagram of another device for providing multi-agent personalized power data services provided in an embodiment of the present application. DETAILED DESCRIPTION

[0031] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0032] The present application embodiment provides a method for planning a flight path of an unmanned aerial vehicle. Figure 1 As shown, the present application embodiment provides the following specific steps:

[0033] 101. According to different artificial intelligence learning methods, different types of algorithm libraries are constructed, including at least: supervised learning library, unsupervised learning library, reinforcement learning library, and deep learning library; in the algorithm library, each model algorithm is used as an algorithm template and is associated with at least one functional scenario, so that the model algorithm is used as an algorithm template to train the corresponding model, which is used to provide scenario services in the corresponding functional scenario.

[0034] In an embodiment of the present application, different types of algorithm libraries are pre-built according to different artificial intelligence learning methods, and multiple model algorithms will be built in each algorithm library. Each model algorithm can be used as an algorithm template. In order to provide a specified specific scenario service, it can be used for model training to obtain the required model. It can be seen that a model algorithm can be used as an algorithm template, corresponding to a functional scenario, to provide a specified scenario service. In short, the mapping relationship between model algorithms, functional scenarios and scenario services can be obtained. Then, during construction, based on the fine-grained division of the model algorithm, the corresponding scenario services for the functional scenarios that can be provided are also of a relatively high degree of fine-grainedness. Based on this, the higher the degree of fine-grainedness, the more subdivided and more numerous scenario services can be provided for functional scenarios.

[0035] For example, different types of algorithm libraries constructed in the embodiments of the present application may include but are not limited to the following, and exemplarily, some model algorithms and supported functional scenarios are constructed in the algorithm library accordingly.

[0036] (1) Figure 2 The supervised learning library shown has prior knowledge. The model algorithm is learned through data cases and uses prior knowledge for application and extrapolation prediction. It includes LASSO regression, Ridge regression, Logistic regression, Softmax regression, LDA, KNN, decision tree, SVM, random forest and other model algorithms.

[0037] The application scenarios of the supervised learning library include: load forecasting, price forecasting, behavior forecasting, risk management, credit scoring, association analysis, email filtering, recommendation systems, etc.

[0038] (2) Figure 3 The unsupervised learning library shown does not have prior knowledge. The model algorithm learns by "observing" the data and mines unknown information and knowledge from the data. It includes model algorithms such as system clustering, hierarchical clustering, k-means clustering, Gaussian hybrid clustering, graph clustering, DBSCAN clustering, PCA, KPCA, etc.

[0039] The functional scenarios of the unsupervised learning library include: customer segmentation, customer profiling, load classification, pattern recognition, anomaly detection, population classification, gene classification, image segmentation, etc.

[0040] (3) Figure 4In the reinforcement learning library shown, the agent interacts with the environment to gain experience, and learns from experience to obtain the optimal strategy that maximizes the cumulative reward (return). The agent interacts with the real environment to learn experience, updates the network weights online, and directly reinforces learning; for decision-making problems that are difficult to reproduce, a model is built to simulate the real environment, the agent interacts with the simulated environment to learn experience, updates the network weights offline, and indirectly reinforces learning. The reinforcement learning library includes DP, MC, TD, REINFORCE, AC, DQN, DDPG and other model algorithms.

[0041] Its functional scenarios include: recommendation system, dialogue system, quantitative investment, high-frequency trading, etc.

[0042] (4) Figure 5 The deep learning library shown in the figure contains hierarchical machine learning methods with multi-level nonlinear transformations, with deep neural networks as the main form. It uses multiple processing layers with complex structures or composed of multiple nonlinear transformations to learn high-level abstract representations of data. It includes RNN, LSTM, GRU, DCN, GAN, DBN and other model algorithms.

[0043] Functional scenarios include: face recognition, image recognition, intelligent monitoring, voice recognition, simultaneous interpretation, machine translation, robot customer service, etc.

[0044] As described above, the embodiments of the present application exemplarily provide different types of algorithm libraries, including: supervised learning library, unsupervised learning library, reinforcement learning library, and deep learning library. In each algorithm library, a variety of model algorithms can be constructed as model templates, and each model template can be associated with at least one functional scenario, which is used to realize the use of the model template to train a model, with the purpose of supporting the provision of scenario services in the functional scenario.

[0045] 102. Configure at least one power data service mode for different entities. The power data service mode corresponds to at least one service demand. The entities include entities within the power market and entities outside the power market.

[0046] The members of the power market include power generation, power consumption, power grid companies, power trading institutions, regulatory agencies, etc. Among them, the power consumption side is composed of power users and power sales companies. The power generation side and the power consumption side are collectively referred to as market entities. The differentiated needs of market entities have brought complexity to the setting of value-added service models. Therefore, the following will analyze the personalized and diversified value-added service needs of market entities inside and outside the market, so as to provide a basis for the setting of multi-subject data value-added service models.

[0047] (1) Market demand

[0048] The market players in the electricity market environment mainly include power generation companies, power grid companies, power trading centers, power sales companies and power users.

[0049] 1) Power generation companies. With the accelerated advancement of my country's power system reform, the implementation of "separation of power plants from the grid, competitive bidding for access to the grid" and the ultimate goal of building a fair and open power market have become the development goals of the power system reform, and the operating environment of power generation companies has undergone tremendous changes. Faced with the introduction of a competitive mechanism, making perfect and reasonable bidding decisions and obtaining maximum profits are the main requirements for power generation companies to participate in market transactions. Therefore, the demand for value-added services for power generation companies lies in predicting transaction electricity prices, comparing various bidding strategies, and making optimal bidding decisions.

[0050] 2) Grid companies. Grid companies in the power market environment are mainly responsible for power transmission and distribution, grid dispatching, and ensuring the basic power demand of various industries. How to reduce the cost of power transmission and distribution and reasonably dispatch power resources are the main needs of grid companies. Therefore, it is necessary to carry out corresponding market transaction data analysis to cooperate with the power transmission and distribution dispatching work of the grid.

[0051] 3) Power trading center. In the market environment, the power trading center is not only the main provider of value-added services, but also the recipient of value-added services from other power trading centers, such as receiving business training from other power trading centers and learning advanced construction experience. Therefore, as the organizing and operating entity of power trading, the power trading center needs to expand its main business scope, simplify the power trading process, improve the business level of related practitioners, and provide corresponding value-added services in a targeted manner.

[0052] 4) Power sales companies. According to the framework of the new round of power system reform, power sales companies will become the most important market players on the power consumption side, and the current profit model of power sales companies is mainly to earn the difference through self-built power sources or external power purchases. After the electricity price is liberalized, how to purchase low-cost electricity, how to solve the difficulties in order financing, how to forecast and manage loads and make optimal decisions on power purchases and sales, how to optimize electricity price packages, and how to innovate business and become a comprehensive enterprise of energy supply and services are the main value-added business needs of new power sales companies. Therefore, it is necessary to carry out corresponding load forecasting, decision support and transaction loan value-added services.

[0053] 5) Electricity users. With the improvement of living conditions, electricity users have gradually developed from the initial basic electricity requirements to diversified service needs. Their demand for value-added services is mainly reflected in the independent participation in market competition, reducing energy costs, ensuring power supply quality and improving energy experience. The power trading center is required to carry out corresponding value-added services such as energy analysis and energy optimization.

[0054] (2) Demand from non-market entities

[0055] In the electricity market environment, non-market entities mainly include government departments, energy investors, power equipment manufacturers, non-industry research institutions and the general public.

[0056] 1) Government departments. With the deepening of power marketization, the power trading mechanism is very different from the past. Due to the information asymmetry between government departments and power trading centers, it may cause the system or policy to be imprecise or difficult to implement. Therefore, the government's demand for value-added services is mainly reflected in timely obtaining relevant information on power trading or market development analysis reports, or regularly obtaining popularization or training on power market knowledge.

[0057] 2) Energy investors and power equipment manufacturers. With the improvement of environmental protection requirements, the development of clean energy and new power equipment, the new energy industry and the intelligent power equipment industry are booming. However, these industries are facing the problem of low product penetration. Therefore, these companies' demand for value-added services in the power industry lies in cooperating with power trading centers with rich user bases and market data information, understanding social energy consumption information, seeking ways to promote products, and increasing market share.

[0058] 3) Research institutions outside the industry. As a pillar industry closely related to the national economy, the power industry requires data analysis from research institutions outside the industry to evaluate the development of the industry and even the entire society. For example, economic think tank research institutes and securities industry research institutions need to use macro data of electricity as an indicator to measure the production and operation activities of the entire society and the development of the industry; natural gas companies, comprehensive energy companies and other industries related to the power industry need to conduct linkage analysis of power trading data and the actual situation of the industry.

[0059] 4) The public. The public’s attitude towards a company will have a huge impact on the company’s image and marketing activities. The public’s demand for value-added services from the power trading center often focuses on information disclosure from the trading center, popularization of energy-saving knowledge, and analysis of electricity fee collection rules.

[0060] 103. According to the service requirements corresponding to the power data service model, a matching target functional scenario is searched from at least one functional scenario associated with the model algorithm of the algorithm library.

[0061] In combination with the market entities described in 102 above, illustratively, the embodiment of the present application adopts the following Table 1 to display the data service requirements of multiple market entities.

[0062] Table 1

[0063]

[0064]

[0065] In the embodiment of the present application, different types of algorithm libraries are constructed in 101, and the algorithm library includes model algorithms that can be used as model templates. In order to realize the specified functional scenario, the model template can be used to train the corresponding model to realize the provision of scenario services. Based on the fine-grained planning and construction of the model algorithm when constructing the algorithm library in 101, it is possible to support a variety of functional scenarios with a large number of functions to provide corresponding scenario services.

[0066] On this basis, the embodiments of the present application can, but are not limited to, analyze the service requirements corresponding to the power data service model of the subject and refine them to determine which functional scenarios are needed to support implementation, thereby finding which target functional scenarios can be used for assembly from a large number of functional scenarios supported by model algorithms in different types of algorithm libraries pre-built in 101.

[0067] Specifically, when performing a search operation, the semantics represented by the service requirements corresponding to the power data service model can be utilized according to natural language parsing and processing, and the semantics represented in a large number of functional scenarios supported and implemented by the model algorithms in the algorithm library can be used for semantic parsing and matching, so as to determine which target functional scenarios are related to the service requirements of the power data service model, and thus find these target functional scenarios, which indirectly determines which model algorithms in the algorithm library can be used to provide scenario services, and then assist in establishing the application services provided by the power data service model.

[0068] 104. Utilize the model algorithm associated with the target functional scenario in the algorithm library to assist in the execution of the scenario service corresponding to the target functional scenario, and apply it to the establishment of application services that meet the service requirements of the power data service model.

[0069] In an embodiment of the present application, in order to meet the service requirements corresponding to the power data service model of a subject, such as 103, the embodiment of the present application searches for model algorithms from the algorithm library constructed by 101. The embodiment of the present application is equivalent to being able to efficiently recommend which model algorithms in the algorithm library are selected as model versions, so as to be used to assist in the execution of providing a scenario service provided in a target functional scenario, such as using the recommended model algorithm as a template to train a model, and using it to implement the provision of specified scenario services in a required target functional scenario, and finally using the scenario services provided in each target functional scenario to indirectly form a larger service set, thereby providing the application services required by the service requirements corresponding to this power data service model.

[0070] For example, for the subject "electricity user", the service requirements of the electricity data service model are configured, such as the user electricity consumption behavior prediction service. For such service requirements, historical experience prediction and semantic analysis can be used. Assuming that the service requirements involve functional scenarios, such as "load forecasting", "behavior prediction" and "customer portrait" functional scenarios, then based on the algorithm library listed in 101, it can be seen that the model algorithms associated with the first two functional scenarios come from the "supervised learning library", and the model algorithm associated with the latter functional scenario comes from the "unsupervised learning library", and can be pre-refined and divided in the construction of the algorithm library. What is the model algorithm associated with each functional scenario is not limited to the examples here, and the above is only an exemplary example of the embodiment of the present application. Configuring a service requirement corresponding to an electricity data service model may not be limited to the functional scenarios listed above.

[0071] However, according to the above examples, if the degree of granularity of the segmented model algorithms and functional scenarios can be improved when building the algorithm library, then when facing the service needs of different power data service models in the future, a model algorithm and its corresponding functional scenarios can be used in different power data service models. For example, power data service model A corresponds to service demand A, and power data service model B corresponds to service demand B. After analyzing service demand A and service demand B, assuming that service demand A involves covering functional scenarios a, b, c, d, and e, and service demand B involves covering functional scenarios a, c, f, and g, it can be seen that the two service demands involve covering the same functional scenarios a and c. Therefore, according to the higher the degree of granularity when building the algorithm library, some model algorithms and their corresponding functional scenarios can be used universally, so that the algorithm library can be applied to power data service models that support different entities.

[0072] Above, the embodiment of the present application provides a method for providing personalized power data services for multiple subjects. The embodiment of the present application is to pre-build different types of algorithm libraries according to different artificial intelligence learning methods, and each model algorithm in each type of algorithm library will be used as an algorithm template to apply the corresponding model for training, so that the model can be used to provide scene services on a corresponding functional scenario, that is, a model algorithm can support a functional scenario. Therefore, when constructing the algorithm library, the embodiment of the present application divides the fine-grained degree of the constructed model algorithm, and accordingly makes the supported functional scenarios more fine-grained and diverse, thereby obtaining more rich and diverse scene services. In addition, the embodiment of the present application constructs at least one power data service model for different topics, and combines the model algorithm and the supported functional scenarios in the algorithm library as above, selects the available target functional scenario from at least one functional scenario according to the service requirements of the power data service model, and then uses the target functional scenario and its corresponding model algorithm to combine to obtain richer scene services, so as to assist in the establishment of application services that can be provided by the power data service model. Above, in response to the personalized needs of multiple subjects, the embodiment of the present application provides an efficient and service-diverse solution.

[0073] In some modified embodiments, as in 102, at least one power data service model is constructed for different entities. For entities within the power market and entities outside the power market, exemplarily, the embodiment of the present application provides the following power data service model: and the following A1-A4 are in a parallel relationship.

[0074] A1 is based on electricity users as the main body of the electricity market, and the configured electricity data service model at least includes: user electricity consumption behavior prediction service.

[0075] The service requirements corresponding to the construction of user electricity consumption behavior prediction service include at least: real-time prediction of electricity load, short-term prediction of electricity load and long-term prediction of electricity load; a model template corresponding to the use is recommended for each service requirement, and the model template is used to train and build a prediction model for electricity load.

[0076] User load demand forecasting can optimize energy distribution for users and power grid companies. It mainly includes four main stages: data collection, data preprocessing, model building and load forecasting. By collecting and preprocessing data such as meteorology, historical load, date type, etc., data quality and integrity are improved. Informer abandons the sequential dependency of recurrent neural networks, uses a multi-scale time encoding-decoding structure to capture features at different time scales, and uses an adaptive length attention mechanism to automatically adjust the attention range according to the sequence length, improving the processing capability of long sequence data, while improving computing efficiency through block and parallel computing.

[0077] Through data collection, data is cleaned, fused, normalized, and exception processed. A prediction model is built based on LSTM, Attention, graph structure learning, Informer and other algorithms to learn the medium- and long-term dependencies and key moment features of time series data. The model is combined and structurally adjusted to achieve real-time, short-term or medium- and long-term prediction of load demand.

[0078] A2 is based on the power distribution enterprise as the main body of the power market, and the configured power data service model at least includes: building a simulation and deduction environment for power trading behavior that conforms to the real operating environment.

[0079] The service requirements corresponding to the construction of the power data service model include at least: power supply and demand forecasting, quotation strategy; recommend the corresponding model template for each service demand, and the model template is used to train and build a power supply and demand forecasting model and a quotation strategy forecasting model.

[0080] The continuous deepening of electricity market reform has led to increasing flexibility and uncertainty in the decisions of market players on power generation and consumption, bringing new challenges to market supervision and risk management. Constructing a simulation environment for power trading behavior that conforms to the real operating environment is the key to meeting these challenges.

[0081] Based on various data types such as historical power generation and consumption data, meteorological data, and market environment, we use algorithmic technologies such as neural networks and machine learning to build prediction models for different scenarios and realize power supply and demand forecasts. We use large model technologies such as reinforcement learning and swarm intelligence to tailor prediction models and quotation strategy generation mechanisms for market participants. In a dynamically changing market environment, we can adaptively adjust quotation strategies to ensure the authenticity and predictability of behavioral simulation deductions, and provide market participants with auxiliary reference for actual operations.

[0082] The design of an analysis and simulation forecasting application model based on the market trends of power generation and sales companies can assist them in formulating quotation strategies and bidding decisions. At the same time, it can combine historical data and real-time market information to conduct risk assessment and cost-benefit analysis, thereby assisting the actual operation of power sales companies.

[0083] A3 is based on the power trading center as the main body of the power market, and the configured power data service model includes at least: user credit evaluation service and user portrait service.

[0084] For user portrait services, the service requirements corresponding to user portrait services include at least: labeling user information; building a labeling system for power market user portraits based on labeling user information, and the labeling system includes at least: data aggregation layer, basic labeling layer, application labeling layer and service application layer; setting the data processing direction from data aggregation layer to basic labeling layer, basic labeling layer to application labeling layer, and application labeling layer to service application layer to output and provide services.

[0085] ① User credit evaluation: The credit of electricity users is closely related to their "electricity consumption" characteristics. Therefore, relying on big models and artificial intelligence algorithms, through data collection at the perception layer, the user's load curve and previous default records are obtained, and the evaluation index system is configured as needed. Data is extracted to calculate the score and rating of each scenario in the index system, and a credit evaluation report and credit evaluation ranking are generated and released to the public.

[0086] ② User portrait: Based on the unsupervised learning library of the large model, by analyzing user data, it is possible to label user information and build a virtual representation of the real user, so that energy companies can quickly and accurately understand the user's full information picture, thus laying the foundation for providing users with targeted energy services.

[0087] The data aggregation layer under the specific power market user portrait system architecture is mainly responsible for collecting user basic data, channel access data, business basic data and external data, etc., to provide basic data sources for the label library; the basic label layer is mainly responsible for the generation and setting of the basic label library, including the basic indicator library, label comparison table, indicator weight value and label calculation rules, etc.; the application label layer is mainly responsible for the construction and setting of the application label layer, including label management, user attributes, behavioral preferences, etc.; the service application layer is mainly based on basic labels and application labels, providing services for upper-level applications such as system applications and information output, such as Figure 6 .

[0088] A4 is based on the power grid enterprises as the main body of the power market, and the configured power data service model at least includes: industry analysis and market operation evaluation application.

[0089] Based on deep learning technology, it mines indicators that are difficult to quantify, such as social welfare, maturity and health of the electricity market, and calculates and displays them in real time, including high-level indexes, key indicator evaluation, risk early warning monitoring and other functions. Relying on artificial intelligence and large model analysis algorithms, it makes analysis reports on different time scales such as years, quarters and months for the transaction data of market entities. At the same time, it relies on intelligent correlation analysis algorithms to analyze the correlation between the electricity market and the industry, and reflects the development of the industry through the result data of the electricity market.

[0090] In some modified embodiments, the embodiments of the present application also provide some power data service modes such as including: intelligent consultation and preference push, and intelligent customer service.

[0091] ① For intelligent consultation and preference push, develop a data value-added service model based on the personalized needs of market entities, develop intelligent consultation and preference push services, realize user preference information push, and enhance user stickiness and experience.

[0092] Within the scope of information disclosure, it provides users with convenient and intuitive basic and advanced search functions. Based on large model technology, by analyzing user data and based on user behavior characteristics, it provides users with targeted power trading information services, including content push, product push, expert push, etc.

[0093] Build a systematic market analysis knowledge base based on the data content of power grid enterprises, or build a knowledge base for problems or topics to achieve rapid matching of user consultation content with the knowledge base. At the same time, according to the customization and retrieval habits of each market player for data, perform frequency analysis, correlation analysis, etc. on the recorded data, and push information that users are concerned about based on the analysis results, saving manual response time, improving service efficiency, and increasing user stickiness.

[0094] ② For intelligent customer service, RAG technology is introduced. By building a professional field knowledge base, before LLM answers user questions, it first identifies user intentions and retrieves information from the professional knowledge base. The retrieved results are used as a reference to organize answers, improve the professionalism of the answers, and reduce the illusion phenomenon of the model. By citing the source of knowledge, users can verify the accuracy of the answers and increase their trust in the model output, effectively solving the problem of introducing and updating professional knowledge in the power market.

[0095] Focus on user self-service and intelligent interactive services, supplemented by manual services; make full use of the advantages of robots to guide users to switch to self-service and interactive service modes, and divert manual pressure. Establish a good information channel between intelligent customer service robots and power users, so that the information dividend brought by social progress can truly benefit all entities in the power market.

[0096] Furthermore, as a response to the above Figure 1 The implementation of the method shown in the figure, the embodiment of the present application provides a device for providing multi-subject personalized power data services. The device embodiment corresponds to the aforementioned method embodiment. For ease of reading, the present device embodiment will no longer repeat the details of the aforementioned method embodiment one by one, but it should be clear that the device in this embodiment can correspond to all the contents of the aforementioned method embodiment. The device is used to efficiently assist in the establishment of rich and diverse application services that meet the needs of power data service models of different subjects, such as Figure 7 As shown, the device comprises:

[0097] A construction unit 21 is used to construct different types of algorithm libraries according to different artificial intelligence learning methods, including at least: a supervised learning library, an unsupervised learning library, a reinforcement learning library, and a deep learning library; in the algorithm library, each model algorithm is used as an algorithm template, and is correspondingly associated with at least one functional scenario, so that the model algorithm is used as an algorithm template to train the corresponding model, so as to provide scenario services in the corresponding functional scenario;

[0098] A configuration unit 22, configured to configure at least one power data service mode for different subjects, wherein the power data service mode corresponds to configuring at least one service requirement, and the subjects include subjects in the power market and subjects outside the power market;

[0099] A search unit 23, configured to search for a matching target functional scenario from at least one functional scenario associated with the model algorithm in the algorithm library according to the service requirement corresponding to the power data service mode;

[0100] The execution unit 24 is used to utilize the model algorithm associated with the target functional scenario in the algorithm library to assist in the execution of providing the scenario service corresponding to the target functional scenario, and is used to form an application service that meets the service requirements corresponding to the power data service model.

[0101] Further, such as Figure 8 As shown, the configuration unit 22 includes:

[0102] The first configuration module 221 is used to configure the power data service mode based on the power users as the power market entities, including at least: user power consumption behavior prediction service;

[0103] The second configuration module 222 is used to configure the power data service mode based on the power selling enterprise as the power market subject, which at least includes: building a power transaction behavior simulation environment that conforms to the real operating environment;

[0104] The third configuration module 223 is used to configure the power data service mode based on the power trading center as the power market subject, including at least: user credit evaluation service and user portrait service;

[0105] The fourth configuration module 224 is used to configure the power data service model based on the power grid enterprise as the power market entity, including at least: industry analysis and market operation evaluation application.

[0106] Further, such as Figure 8 As shown, the first configuration module 221 is specifically used for:

[0107] The service requirements corresponding to the user electricity consumption behavior prediction service include at least: real-time prediction of electricity load, short-term prediction of electricity load and long-term prediction of electricity load;

[0108] A model template is associated with each of the service requirements and used for training and constructing a prediction model for power load.

[0109] Further, such as Figure 8 As shown, the second configuration module 222 is specifically used for:

[0110] The service requirements corresponding to the power data service model include at least: power supply and demand forecasting, quotation strategy;

[0111] A model template corresponding to the service demand recommendation is associated with each of the service demand recommendations, and the model template is used to train and construct an electricity supply and demand prediction model and a quotation strategy prediction model.

[0112] Further, such as Figure 8 As shown, the third configuration module 223 is specifically used for:

[0113] The service requirements corresponding to the construction of the user portrait service at least include: labeling user information;

[0114] According to the labeling of user information, a labeling system for power market user portraits is constructed, wherein the labeling system at least includes: a data aggregation layer, a basic labeling layer, an application labeling layer, and a service application layer;

[0115] The data processing direction is set from the data aggregation layer to the basic tag layer, from the basic tag layer to the application tag layer, and from the application tag layer to the service application layer to output and provide services.

[0116] Furthermore, the entities within the electricity market include at least: electricity users, electricity sales companies, power trading centers, and power grid companies; and the entities outside the electricity market include at least: government departments, energy investors and power equipment manufacturers, non-industry research institutions, and the general public.

[0117] In summary, the device for providing multi-agent personalized power data services includes a processor and a memory. The above-mentioned construction unit, configuration unit, search unit and execution unit are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions.

[0118] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set to assist in realizing the scene services of functional scenarios by adjusting kernel parameters, so as to meet the service requirements of the power data service model of different subjects, and can quickly and accurately recommend available model templates, based on which the scene services corresponding to the target functional scenarios can be assisted in the execution, so that a larger service set can be formed based on multiple scene services, which is used to meet the application services corresponding to the service requirements of the power data service model. In the above, in response to the personalized needs of different subjects, this application provides an efficient and service-diverse solution.

[0119] An embodiment of the present application provides a storage medium having a program stored thereon, and a method for providing a multi-agent personalized power data service is implemented when the program is executed by a processor.

[0120] An embodiment of the present application provides a processor, which is used to run a program, wherein the program executes a method for providing a multi-agent personalized power data service when the program is running.

[0121] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program of initializing the steps of a method for providing a multi-agent personalized power data service.

[0122] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0123] In a typical configuration, the device includes one or more processors (CPU), memory and bus. The device may also include input / output interface, network interface and the like.

[0124] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip. The memory is an example of a computer-readable medium.

[0125] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules 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, 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. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0126] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0127] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0128] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for providing multi-agent personalized power data services, characterized in that: The method comprises: According to different artificial intelligence learning methods, different types of algorithm libraries are constructed, including at least: supervised learning library, unsupervised learning library, reinforcement learning library, and deep learning library; in the algorithm library, each model algorithm is used as an algorithm template, and is correspondingly associated with at least one functional scenario, so that the model algorithm is used as an algorithm template to train the corresponding model, which is used to provide scenario services in the corresponding functional scenario; At least one power data service mode is configured for different subjects, and at least one service demand is configured for the power data service mode, and the subjects include subjects in the power market and subjects outside the power market; According to the service requirement corresponding to the power data service mode, searching for a matching target functional scenario from at least one functional scenario associated with the model algorithm in the algorithm library; The model algorithm associated with the target functional scenario in the algorithm library is used to assist in the execution of providing the scenario service corresponding to the target functional scenario, which is applied to forming an application service that meets the service requirements corresponding to the power data service model.

2. The method according to claim 1, characterized in that The configuring at least one power data service mode for different subjects includes: Based on the electricity users as the main body of the electricity market, the configuration of the electricity data service model at least includes: user electricity consumption behavior prediction service; Based on the electricity sales enterprise as the main body of the electricity market, the configuration of the electricity data service model at least includes: building a simulation environment for electricity trading behavior that conforms to the real operating environment; Based on the power trading center as the main body of the power market, the configuration of the power data service model includes at least: user credit evaluation service and user portrait service; Based on the fact that power grid enterprises are the main players in the power market, the configuration of the power data service model at least includes: industry analysis and market operation evaluation applications.

3. The method according to claim 2, characterized in that The configuration of the power data service model based on power users as power market entities at least includes: user power consumption behavior prediction service, including: The service requirements corresponding to the user electricity consumption behavior prediction service include at least: real-time prediction of electricity load, short-term prediction of electricity load and long-term prediction of electricity load; A model template is associated with each of the service requirements and used for training and constructing a prediction model for power load.

4. The method according to claim 2, characterized in that: The configuration of the power data service model based on the power distribution enterprise as the power market subject at least includes: building a power transaction behavior simulation environment that conforms to the real operating environment, including: The service requirements corresponding to the power data service model include at least: power supply and demand forecasting, quotation strategy; A model template corresponding to the service demand recommendation is associated with each of the service demand recommendations, and the model template is used to train and construct an electricity supply and demand prediction model and a quotation strategy prediction model.

5. The method according to claim 2, characterized in that: The configuration of the power data service model based on the power trading center as the power market entity at least includes: user portrait service, including: The service requirements corresponding to the construction of the user portrait service at least include: labeling user information; According to the labeling of user information, a labeling system for power market user portraits is constructed, wherein the labeling system at least includes: a data aggregation layer, a basic labeling layer, an application labeling layer, and a service application layer; The data processing direction is set from the data aggregation layer to the basic tag layer, from the basic tag layer to the application tag layer, and from the application tag layer to the service application layer to output and provide services.

6. The method according to any one of claims 1 to 5, characterized in that The entities in the power market include at least: power users, power distribution companies, power trading centers, and power grid companies. The entities outside the power market include at least: government departments, energy investors and power equipment manufacturers, non-industry research institutions, and the general public.

7. A device for providing multi-agent personalized power data services, characterized in that: The device comprises: A construction unit is used to construct different types of algorithm libraries according to different artificial intelligence learning methods, including at least: supervised learning library, unsupervised learning library, reinforcement learning library, and deep learning library; in the algorithm library, each model algorithm is used as an algorithm template, and is correspondingly associated with at least one functional scenario, so that the model algorithm is used as an algorithm template to train the corresponding model, so as to provide scenario services on the corresponding functional scenario; A configuration unit, configured to configure at least one power data service mode for different subjects, wherein the power data service mode corresponds to configuring at least one service demand, and the subjects include subjects in the power market and subjects outside the power market; A searching unit, configured to search for a matching target functional scenario from at least one functional scenario associated with the model algorithm in the algorithm library according to the service requirement corresponding to the power data service mode; An execution unit is used to utilize the model algorithm associated with the target function scenario in the algorithm library to assist in the execution of the scenario service corresponding to the target function scenario, and is used to form an application service that meets the service requirements corresponding to the power data service model.

8. The device according to claim 7, characterized in that The configuration unit comprises: The first configuration module is used to configure the power data service mode based on the power users as the power market subjects, including at least: user power consumption behavior prediction service; The second configuration module is used to configure the power data service model based on the power sales enterprise as the power market subject, which at least includes: building a power transaction behavior simulation environment that conforms to the real operating environment; The third configuration module is used to configure the power data service mode based on the power trading center as the power market entity, including at least: user credit evaluation service and user portrait service; The fourth configuration module is used to configure the power data service model based on the power grid enterprise as the main body of the power market, which at least includes: industry analysis and market operation evaluation application.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for providing a multi-agent personalized power data service according to any one of claims 1 to 6 is implemented.

10. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the method for providing a multi-agent personalized power data service as claimed in any one of claims 1 to 6 is implemented.

Images