CN115344717B - Method and device for constructing regulation and control operation knowledge graph for multi-type energy supply and consumption system - Google Patents
Method and device for constructing regulation and control operation knowledge graph for multi-type energy supply and consumption system Download PDFInfo
- Publication number
- CN115344717B CN115344717B CN202211272027.7A CN202211272027A CN115344717B CN 115344717 B CN115344717 B CN 115344717B CN 202211272027 A CN202211272027 A CN 202211272027A CN 115344717 B CN115344717 B CN 115344717B
- Authority
- CN
- China
- Prior art keywords
- knowledge graph
- data
- regulation
- constructing
- control operation
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
-
- 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)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Supply And Distribution Of Alternating Current (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a method and a device for constructing a regulation and control operation knowledge graph for a multi-type energy supply and utilization system, wherein the method comprises the following steps: determining operation technical indexes of a regulation and control operation strategy of the multi-type energy supply system, and sequentially detailing and constructing a knowledge graph mode layer by adopting a top-down method; preprocessing the regulation and control operation strategy data, and dividing the regulation and control operation strategy data into training samples and testing samples according to characteristic indexes; constructing different neural network models to realize the knowledge extraction of the regulation and control operation strategy data and obtain triple data containing entity-attribute-entity; adding the ternary group data into the knowledge graph and constructing a knowledge graph data layer by adopting a bottom-up method; and classifying the data of the knowledge graph data layer, combining the data with a knowledge graph mode layer to form a strategy set, and constructing and storing a complete knowledge graph. The method realizes the rapid structured extraction of the power grid regulation strategy text and the construction and storage processing of the knowledge graph.
Description
Technical Field
The invention belongs to the technical field of energy scheduling, and particularly relates to a method and a device for constructing a regulation and control operation knowledge graph for a multi-type energy supply and utilization system.
Background
The large-scale new energy power generation is connected into the power grid, and the power grid has intermittence, randomness and volatility, so that the power grid presents more complex nonlinear random characteristics, multi-state variable coupling and multi-time scale dynamic characteristics. Meanwhile, the system form and the operation characteristic are increasingly complex, the coupling relation between alternating current and direct current of a power distribution network and between direct current transmission and receiving end power grids is complex, the network structure is large in change, the operation mode is complex and changeable, the number of monitoring sections is increasingly increased, and the control rule is increasingly complex. These pose a serious challenge to the stable operation of the power grid and more elaborate requirements on the power dispatching operation.
Therefore, the power grid dispatching operation automation system plays an indispensable role in the safe operation of the power grid. The existing dispatching automation system has more and more applications and more complex business knowledge, most related business personnel only know local business knowledge but not related business logic, and only few expert-level personnel can clearly know the whole business process. Therefore, when a problem occurs in the complex business logic, all business personnel need to be temporarily mobilized to clear the business logic relationship, and the reason causing the problem can be possibly found out. If the knowledge graph of the related services of the whole dispatching automation system is established, all the operations and data flows can be inquired clearly according to the related service logic expressed by the knowledge graph, and therefore the optimal control operation strategy is found out.
Therefore, it is necessary to digitally process the regulation and control contents and operation strategies in various text forms, and the existing related researches mainly translate the stability regulation rules one by adopting a computer script language, and store the calculation results of the formed stable operation control strategies in a D5000 real-time library for online operation monitoring. For example, in an article "intelligent analysis-based power grid operation mode and security control strategy intelligent compilation method and application", a C language definition mode is referred, and a script file is used for describing power grid operation control conditions, means contents and the association relationship between the power grid operation control conditions and the means contents, and the association relationship between the security automatic device action conditions and the action strategy contents; however, the method is not visual enough in logic relation, uncontrollable in accuracy rate, complex in scripted entry and daily maintenance, huge in workload, and challenging because the forming and maintenance work of script documents needs to have professional knowledge and programming skills of the power system at the same time. The knowledge graph can effectively organize, manage and utilize mass information to realize intelligent knowledge extraction, reasoning, storage and retrieval, and the characteristics and application scenes of the knowledge graph are very matched with the requirements of an electric power system, so that a method for constructing the knowledge graph is provided in the field of electric power system automation aiming at the relevant business logic of an operation control automation system, the knowledge graph of the automation system is established, the summary, the search and the application of a regulation and control strategy are facilitated, and the method has high research value.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a method and a device for constructing a knowledge graph for regulating and controlling operation of a multi-type energy supply system aiming at the defects of the prior art, wherein knowledge extraction of regulating and controlling operation strategy data is realized by establishing a knowledge graph mode layer and constructing different neural network models, ternary group data comprising an entity, an attribute and an entity is obtained, a knowledge graph data layer is constructed based on the ternary group data, rapid structural extraction of a power grid regulating and controlling strategy text and knowledge graph construction and warehousing processing are realized, and a large amount of manual extraction and warehousing time is saved.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows:
a construction method of a regulation and control operation knowledge graph for a multi-type energy supply system comprises the following steps:
step 1: determining operation technical indexes of a regulation and control operation strategy of the multi-type energy supply and utilization system, sequentially refining, and constructing a knowledge graph mode layer by adopting a top-down method;
step 2: carrying out data preprocessing on the regulation and control operation strategy data;
and step 3: constructing different neural network models to realize the knowledge extraction of the regulation and control operation strategy data and obtain triple data containing 'entity-attribute-entity';
and 4, step 4: adding the ternary group data into the knowledge graph and constructing a knowledge graph data layer by adopting a bottom-up method;
and 5: and classifying the data of the knowledge graph data layer, combining the data with a knowledge graph mode layer to form a strategy set, and constructing and storing a complete knowledge graph.
Further preferably, in the step 1, according to the actual operation state and the regulation and control demand of the power distribution network of the multi-type energy supply and consumption system, the operation technical indexes are determined to be resource types, combination modes, user compositions and operation indexes; and refining downwards to complete the construction of the knowledge graph mode layer.
Preferably, the data preprocessing comprises cleaning, deleting and supplementing the control operation strategy data, and dividing the control operation strategy data into training samples and testing samples according to characteristic indexes; and segmenting the regulation operation strategy data to obtain a regulation word bank, and constructing a space vector according to the regulation word bank.
Further preferably, the process of word segmentation and space vector construction is as follows: firstly, determining each document contained after query text preprocessing, segmenting each document according to a domain word bank to which the document belongs, substituting each alternative keyword belonging to each document into the obtained analysis model, obtaining word vectors of a plurality of dimensions formed by the alternative keywords in each document correspondingly, and outputting the word vectors as space vectors.
The process of step 3 comprises:
step 3.1: performing text classification on the control operation strategy data by using a TextCNN model based on the word vector;
step 3.2: using an LR-CNN model to realize named entity identification of the regulation operation strategy data;
step 3.3: and (3) extracting the relationship among entities in the regulation and control operation strategy data by using a BERT-BilSTM-CRF model.
Further preferably, the named entity recognition in the step 3.2 is to perform boundary determination and category recognition on entities with specific meanings in the regulation and control operation strategy data text, and divide the entities into equipment names, resource names, investment money, recycling periods and operation places by using an LR-CNN model according to the entity attributes and categories contained in the rural power distribution network.
Preferably, the LR-CNN model extracts the character and candidate word features of the sentence by using the CNN layer, and combines the character and candidate word information by using an attention mechanism module; and meanwhile, a Rethinking mechanism is introduced, a feedback layer is added to each CNN layer, and the weight of a low-level attention mechanism module is adjusted by using high-level word information.
Further preferably, the extracting of the relationship between the entities in the step 3.3 is to judge whether predefined relationship exists between the entities on the basis of the named entity identification, so as to form a series of triple knowledge; and predefining relationships among the entities through the operation specifications and states of the power distribution network, and constructing a series of triples in the knowledge graph by using the relationships formed after extraction.
Preferably, the BERT-BilSTM-CRF model uses a pretrained BERT model, the BilSTM-CRF model is connected behind the BERT model, and meanwhile, the output of the BERT model and the output of the BilSTM layer are subjected to feature series connection and subjected to fine tuning training, so that the accuracy of relation extraction is further improved.
Further preferably, the step 5 specifically comprises:
step 5.1: determining a typical scene based on the resource types and characteristics of the power distribution network of the multi-type energy supply and consumption system;
step 5.2: regulating and controlling requirements in the knowledge graph data layer are arranged into a strategy set;
step 5.3: specific data in the knowledge graph data layer are summarized into strategies and are put into a strategy set;
step 5.4: combining the knowledge graph data layer with the knowledge graph mode layer to form a complete knowledge graph;
step 5.5: and storing the knowledge graph.
The invention provides a device for regulating and controlling a construction method of a running knowledge graph for a multi-type energy supply and utilization system, which comprises a knowledge graph mode layer module, a data preprocessing module, a knowledge extraction module, a knowledge graph data layer module, a strategy set module and a storage module;
the knowledge graph mode layer module is used for determining operation technical indexes of a regulation and control operation strategy of the multi-type energy supply system and sequentially refining the operation technical indexes to form a knowledge graph mode layer;
the data preprocessing module is used for preprocessing the data of the regulation and control operation strategy;
the knowledge extraction module is used for constructing different neural network models to realize knowledge extraction of the regulation and control operation strategy data and obtain triple data containing entity-attribute-entity;
the knowledge graph data layer module adds the triple data into a knowledge graph and adopts a bottom-up method to construct a knowledge graph data layer;
the strategy set module classifies the knowledge graph data layer data and combines the knowledge graph data layer with a knowledge graph mode layer to form a strategy set; the storage module is used for storing the knowledge graph.
Adopt the produced beneficial effect of above-mentioned technical scheme to lie in: based on the related technology in the fields of deep learning and knowledge maps, knowledge extraction of the regulation and control operation strategy data is realized by establishing a knowledge map mode layer and constructing different neural network models, the triple data comprising entity-attribute-entity is obtained, and a knowledge map data layer is constructed based on triple data, so that the rapid structured extraction of the power grid regulation and control strategy text, the knowledge map construction and the warehousing processing are realized, and a large amount of manual extraction and warehousing time is saved. When the knowledge graph is used for regulating and controlling information retrieval, the corresponding regulating and controlling strategy can be quickly and accurately positioned, retrieval content pushing is realized, specific operation information can be provided, searching and handling of workers are facilitated, the accuracy is high, and the limitation of the existing knowledge graph is overcome; meanwhile, the power grid regulation and control decision knowledge graph can carry out multi-level and multi-stage decision support information pushing, and the data query and retrieval efficiency is improved.
Drawings
FIG. 1 is a flow chart for constructing a regulation and control operation knowledge graph for a multi-type energy supply system provided by the invention.
FIG. 2 is a diagram of a knowledge graph pattern layer structure provided in the present invention;
FIG. 3 is a diagram of a knowledge-graph data layer structure provided by an embodiment of the invention;
FIG. 4 is a flow diagram of a knowledge graph application provided by an embodiment of the present invention;
fig. 5 is a diagram illustrating an example knowledge graph according to an embodiment of the present invention.
Detailed Description
The following detailed description of embodiments of the present invention is provided in connection with the accompanying drawings and examples. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.
As shown in fig. 1, the method for constructing the knowledge graph for regulating and controlling operation of the multi-type energy supply system according to this embodiment is as follows.
Step 1: determining operation technical indexes of a multi-type energy supply system regulation and control operation strategy, sequentially refining, and constructing a knowledge graph mode layer by a top-down method. The knowledge graph mode layer stores concepts, rules, axioms and constraint conditions, entities in the mode layer are generally abstracted nouns which are extracted from a power system and are also called as bodies, the logic relationship of detailed services under the regulation and control operation strategy of the multi-type energy supply system is complex, the number of the entities is large, the relationship between the bodies needs to be abstracted firstly, and then the final services are constructed through the learning of the entities. Fig. 2 is a basic structure of a knowledge graph pattern layer, and according to actual operation states and regulation and control requirements of a power distribution network of a multi-type energy supply system, operation technical indexes are determined to be resource types, combination modes, user compositions, operation indexes and the like, and are downwards refined to complete construction of the knowledge graph pattern layer. The resource types are refined into photovoltaic power generation, wind power generation, biomass energy and the like. The combination mode is refined into a power grid + photovoltaic power generation, a power grid + cogeneration power generation + grid-connected internet access and the like. The user composition is refined into country residents, processing workshops, farms and the like. The operation index is refined into economic operation and stable operation.
Step 2: and carrying out data preprocessing on the regulation and control operation strategy data. The regulation and control operation strategy data comprises operation requirements, a regulation and control plan and an operation report, and in order to accelerate the training speed of the neural network model and improve the accuracy of the neural network model, the regulation and control operation strategy data needs to be cleaned, deleted and supplemented, and is divided into training samples and testing samples according to characteristic indexes. And segmenting the regulation operation strategy data to obtain a regulation word bank, and constructing a space vector according to the regulation word bank. Firstly, determining each document contained after query text preprocessing, segmenting each document according to a domain word bank to which the document belongs, substituting each alternative keyword belonging to each document into the obtained analysis model, obtaining word vectors of a plurality of dimensions formed by the alternative keywords in each document correspondingly, and outputting the word vectors as space vectors.
And step 3: and constructing different neural network models to realize the knowledge extraction of the regulation and control operation strategy data and obtain the triple data containing the entity-attribute-entity.
Step 3.1: the control run policy data is text classified using a TextCNN model based on the word vectors. And the text classification is to automatically classify and mark the regulating and controlling operation strategy data text according to a predefined category. And the TextCNN model divides the power grid operation state in the data text of the control operation strategy.
Step 3.2: and using an LR-CNN model to realize named entity identification of the regulation operation strategy data. And the named entity identification is to perform boundary determination and category identification on entities with specific meanings in the regulation and control operation strategy data text, and divide the entities into equipment names, resource names, investment money, recovery periods, operation places and the like by using an LR-CNN model according to entity attributes and types contained in the rural power distribution network. The LR-CNN model used in the invention uses a CNN model with the size of a stacked window being 2 to extract the character and candidate word characteristics of the sentence, and uses an attention mechanism module to combine the character and candidate word information. And meanwhile, a Rethinking mechanism is introduced, a feedback layer is added to each CNN layer, the weight of a low-level attention mechanism module is adjusted by using high-level word information, and the weight of an error candidate word is reduced to solve the problem of candidate word conflict.
Step 3.3: and (3) extracting the relationship between entities in the regulation and control operation strategy data by using a BERT-BilSTM-CRF model. The extraction of the relationship between the entities is to judge whether predefined relationship exists between the entities on the basis of the recognition of the named entities, thereby forming a series of triple knowledge. In the invention, the relationship between entities is predefined through the operation specification and the state of the power distribution network, a series of triples in a knowledge map are constructed by utilizing the relationship formed after extraction, meanwhile, a pretrained BERT model is used, a BiLSTM-CRF model is connected behind the pretrained BERT model, and meanwhile, the output of the BERT layer and the output of the BiLSTM layer are subjected to characteristic series connection and fine tuning training, so that the accuracy of relationship extraction is further improved.
And 4, step 4: adding the ternary group data into the knowledge graph and constructing a knowledge graph data layer by adopting a bottom-up method; FIG. 3 is a knowledge graph data layer structure, which mainly includes equipment status, equipment capacity, runtime, operating cost, location, etc. And after finishing text classification, named entity identification and relationship extraction among the entities of the regulation and control operation strategy data, adding the obtained triple containing the entity-attribute-entity into the corresponding position of the knowledge graph data layer.
And 5: and classifying the data of the knowledge graph data layer, combining the data with a knowledge graph mode layer to form a strategy set, and constructing and storing a complete knowledge graph.
Step 5.1: and determining a typical scene based on the resource types and characteristics of the power distribution network of the multi-type energy supply and utilization system. Based on the characteristic data of the cold, heat and electricity demand of users and the evaluation data of renewable energy resources, suitable users are aggregated to form different types of energy cooperative society, and part of the energy cooperative society is determined as a typical scene based on the operation characteristics of the power distribution network of the multi-type energy supply and use system, such as energy cooperative society in rural living areas, ecological tourist energy cooperative society, ecological greenhouse-planted energy cooperative society and the like.
Step 5.2: and regulating and controlling requirements in the knowledge graph data layer are organized into a strategy set. The regulation and control requirements in the knowledge graph data layer are evaluated and judged, such as a user combination mode, a resource operation season, a resource complementary relation, an equipment operation state and the like, and the regulation and control requirements are added into a corresponding typical scene to form a strategy set.
Step 5.3: and (5) summarizing specific data in the knowledge graph data layer into strategies and putting the strategies into a strategy set. Specific data in the knowledge graph data layer, such as cold, heat, electric design load, energy supply system installation scheme, annual energy consumption, total investment and the like, are summarized and extracted and added into a corresponding strategy set to form detailed strategy information which can assist power grid workers in power scheduling and reference decision making in the strategy set.
Step 5.4: and combining the knowledge graph data layer with the knowledge graph mode layer to form a complete knowledge graph.
And (3) determining main operation indexes in the knowledge graph mode layer, and embedding the strategy and strategy set and the corresponding knowledge graph data layer data into the structure of the corresponding knowledge graph mode layer to form a complete knowledge graph.
Step 5.5: and storing the knowledge graph. And storing the constructed complete knowledge graph in a Neo4j graph database, wherein Neo4j is a graph database compiled based on Java language, and the graph database stores information in the form of nodes and relations and provides interface-friendly visual demonstration on the basis.
The application of the regulation and control operation knowledge graph for the multi-type energy supply system is characterized in that under the condition of seasonal operation triggering, corresponding contents in the knowledge graph are found and refined and decomposed, different typical scenes are continuously generated downwards and refined into corresponding strategy sets through step-by-step construction in combination with existing data and different conditions, data of the knowledge graph data sets are added into the schemes of the corresponding scheme sets and output, and complete regulation and control operation strategies are generated to assist workers in making decisions. Referring to fig. 4, the specific steps are as follows:
the method comprises the following steps: seasonal operation triggering conditions are received and located in a knowledge graph. In the operation process of the power distribution network of the multi-type energy supply and consumption system, seasonal operation triggering condition instructions or requirements of the knowledge graph, such as requirements of summer operation state, winter operation state, holiday operation state, temporary regulations and the like, are received, after receiving the triggering conditions, power grid workers search keywords or words in a Neo4j graph database stored in the knowledge graph to find the position of the corresponding knowledge graph, and positioning of the knowledge graph under the seasonal new operation triggering instruction is completed.
Step two: and selecting the operation indexes of the power distribution network. After positioning is completed, the operation state and the regulation and control requirements of the power distribution network are combined, and the operation indexes of the power distribution network, such as indexes of economic operation, resource stabilization, line heavy overload, voltage out-of-limit, stable operation and the like, are continuously selected downwards.
Step three: and generating a corresponding typical scene under each operation index. After the technical indexes of the power grid operation are determined, the power grid operation is refined under the corresponding indexes to generate corresponding typical scenes, such as an ecological tourism cooperative, a farm energy cooperative, an agricultural product processing cooperative and the like.
Step four: and obtaining a corresponding strategy set and a regulation operation strategy. Typical users in the generated corresponding typical scenes, such as rural residents, schools, retail stores, administrative services, iron towers, network services, life power facilities, life production services, health centers, processing workshops and the like, are selected, then, the energy supply and energy utilization characteristics and resource data of the typical users are subdivided to obtain a strategy set containing information such as user combination modes, resource operation seasons, resource complementary relations, equipment operation states and the like, then, specific data in a knowledge graph data layer, such as cold, heat, electric design loads, energy supply system installation schemes, annual energy consumption, total investment and the like, are subjected to induction extraction and added into a corresponding strategy set, and detailed strategy information which can assist power grid workers in power scheduling and reference decision making is formed in the strategy set.
In the knowledge graph application process, corresponding information is gradually added to complete the updating of a knowledge graph mode layer and a knowledge graph data layer; knowledge in the power system is continuously increased and updated, the knowledge map is required to be dynamically constructed and iteratively updated after being built, new knowledge is continuously increased, old knowledge is deleted, and the structure of the knowledge map is correspondingly adjusted. The knowledge graph updating comprises the updating of a knowledge graph data layer and the updating of a knowledge graph mode layer. Relatively, the influence of knowledge graph data layer updating on the whole architecture of the knowledge graph is small, and the influence of knowledge graph mode layer updating is large; therefore, the knowledge graph data layer can be updated automatically, and the knowledge graph mode layer is updated manually.
For the updating of the knowledge graph mode layer, the updating content needed in the mode layer is determined according to the running state of the power distribution network within a period of time and the running requirement of the future power distribution network by adopting an expert evaluation mode, and meanwhile, the mode layer is updated by adopting an incremental updating method, wherein the incremental updating takes newly added data as input to update the knowledge graph, and the resource consumption is low.
The knowledge graph data layer also adopts an incremental updating method, the regulation and control operation strategy document in the knowledge graph data layer is updated every year, and is also updated seasonally in the same year, and meanwhile, the regulation and control operation strategy document may have temporary regulations on holidays and festivals. In the regulation and control operation strategy document of each version, a plurality of regulations exist, for each regulation, a plurality of layers of operation mode definitions exist, and the lowest mode uses the regulation and control operation strategy and a strategy set table to express the operation range of the regulation and control operation strategy.
In summary, in the embodiment, for the problems that the operation automation degree of the power distribution network of the multi-type energy supply and consumption system is not high and the operation regulation and control information is complicated, the knowledge map technology is used for performing knowledge extraction, representation and management on the regulation and control operation strategy information and assisting the scheduling personnel in power grid regulation and control, the decision efficiency and the decision generation speed can be improved, the association relation among different strategies can be displayed more intuitively, and the emergency processing capability and the scheduling intelligence level of the power grid are improved. A method for constructing a multi-type energy supply and consumption system regulation and control operation knowledge map combining top-down and bottom-up is provided by taking a power grid regulation and control operation strategy plan text as a research object, and the problem of knowledge extraction in the power field involved in the method is solved. Firstly, defining the knowledge organization structure, concept type and relationship among concepts of the knowledge graph from top to bottom to form a mode layer of the knowledge graph; then, aiming at the characteristics of a power grid fault handling plan text, a plurality of deep learning models are comprehensively used for extracting knowledge, and a data layer of a knowledge map is constructed from bottom to top: in order to avoid word segmentation errors, classifying the pre-arranged plan text by using a textCNN model based on a word vector; in order to solve the candidate word conflict problem, a named entity in the power field is identified by using an LR-CNN model; on the basis of named entity identification, a BERT-BilSTM-CRF model is used for extracting the relationship between entities. Then, the effectiveness of the knowledge extraction method is verified through experiments. And finally, visualizing the constructed knowledge graph and analyzing the application of the knowledge graph in the generation of the regulation and control strategy.
Taking an application knowledge graph in an ecological forest farm in a typical scene of a certain country in a power distribution network of a multi-type energy supply and utilization system as an example, fig. 5 is a schematic diagram of the application knowledge graph in the ecological forest farm provided by the embodiment of the invention.
TABLE 1 Table of specific implementation schemes of control operation strategy under economic operation scene of ecological forest farm in a certain village of a plurality of types of energy supply and utilization systems
The method comprises the steps of firstly determining the type of a main energy cooperative company in the typical scene, namely the ecological forest, namely a farm energy cooperative, then determining economic operation as a main technical index of operation of a power distribution network in the typical scene, performing text classification on regulation and control operation strategy plan data in the typical scene, identifying entities, extracting relationships among the entities to form corresponding triplets, adding the triplets into a knowledge map to form a knowledge map data layer in the typical scene, wherein the triplets mainly comprise total loads and farm energy supply resources in the cooperative society, refining the total loads into equipment types and load types, comprising heating equipment and cooling equipment, continuously and respectively subdividing the heating equipment into a lamp panel and a warm lamp, subdividing the cooling equipment into a cold air blower and a negative pressure air blower, continuously refining the load types to obtain heating of the farm, cooling of an office building and heating of the office building, refining the farm energy supply resources into biogas, geothermal and electric power, and adding specific data in each type into a corresponding position of the knowledge map, and displaying the energy supply and energy supply information of the farm by using the working staff as a visual reference for the knowledge map.
And then, carrying out induction and arrangement on the contents related to the control operation strategies in the control operation strategy plan data and integrating the control operation strategies into a strategy set, wherein the strategy set generated under the economic operation condition of the ecological forest farm is as follows: biogas system + photovoltaic system + hurdle heat pump heating refrigeration, the specific strategy that the strategy is concentrated is: a power generation waste heat recovery device is added in the methane system, so that the methane yield in winter is improved; the photovoltaic system is paved in a full area, the self-use is the main, and the rest electricity is on the internet; the hurdle heat pump heating and refrigerating system adopts an air source heat pump for offices.
And finally, specific information in the regulation and control operation strategy is summarized and summarized into specific implementation schemes comprising resource output and installation schemes and total investment and recovery periods, and the specific implementation schemes generated in the scene are shown in the table 1, so that a complete knowledge map application process aiming at the ecological forest farm is formed.
The invention also provides a device for regulating and controlling the construction method of the operation knowledge graph for the multi-type energy supply and consumption system, which comprises a knowledge graph mode layer module, a data preprocessing module, a knowledge extraction module, a knowledge graph data layer module, a strategy set module and a storage module;
the knowledge graph mode layer module is used for determining the operation technical indexes of the regulation and control operation strategy of the multi-type energy supply system and sequentially thinning the operation technical indexes to form a knowledge graph mode layer;
the data preprocessing module is used for preprocessing the data of the regulation and control operation strategy data;
the knowledge extraction module is used for constructing different neural network models to realize knowledge extraction of the regulation and control operation strategy data and obtain triple data containing entity-attribute-entity;
the knowledge graph data layer module adds the triple data into a knowledge graph and adopts a bottom-up method to construct a knowledge graph data layer;
the strategy set module classifies the knowledge graph data layer data and combines the knowledge graph data layer with a knowledge graph mode layer to form a strategy set; the storage module is used for storing the knowledge graph.
The present invention provides a computer storage medium having computer program instructions stored thereon; when the computer program instructions are executed by the processor, the method for constructing the regulation and control operation knowledge graph for the multi-type energy supply and utilization system is realized.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; such modifications or substitutions do not depart from the spirit of the invention, which is defined by the claims.
Claims (8)
1. A construction method of a regulation and control operation knowledge graph for a multi-type energy supply system is characterized by comprising the following steps:
step 1: determining operation technical indexes of a regulation and control operation strategy of the multi-type energy supply system, sequentially refining, and constructing a knowledge graph mode layer by a top-down method;
and 2, step: carrying out data preprocessing on the regulation and control operation strategy data;
and 3, step 3: constructing different neural network models to realize knowledge extraction of the regulation and control operation strategy data and obtain triple data containing entity-attribute-entity;
step 3.1: performing text classification on the control operation strategy data by using a TextCNN model based on the word vector;
step 3.2: using an LR-CNN model to realize named entity identification of the regulation and control operation strategy data;
step 3.3: performing entity-to-entity relationship extraction on entities in the regulation and control operation strategy data by using a BERT-BilSTM-CRF model;
and 4, step 4: adding the ternary group data into the knowledge graph and constructing a knowledge graph data layer by adopting a bottom-up method;
and 5: classifying the data of the knowledge graph data layer, combining the data with a knowledge graph mode layer to form a strategy set, and constructing and storing a complete knowledge graph;
step 5.1: determining a typical scene based on the resource types and characteristics of the power distribution network of the multi-type energy supply and utilization system;
and step 5.2: regulating and controlling requirements in the knowledge graph data layer are organized into a strategy set;
step 5.3: specific data in the knowledge graph data layer are summarized into strategies and put into a strategy set;
step 5.4: combining the knowledge graph data layer with the knowledge graph mode layer to form a complete knowledge graph;
and step 5.5: and storing the knowledge graph.
2. The method for constructing the regulation and control operation knowledge graph for the multi-type energy supply system according to claim 1, wherein in the step 1, the operation technical indexes are determined to be resource types, combination modes, user compositions and operation indexes according to the actual operation state and regulation and control requirements of a power distribution network of the multi-type energy supply system; and refining downwards to complete the construction of the knowledge graph mode layer.
3. The method for constructing the regulation operation knowledge-graph for the multi-type energy supply system according to claim 1, wherein the data preprocessing comprises cleaning, deleting and supplementing the regulation operation strategy data, and dividing the regulation operation strategy data into a training sample and a testing sample according to characteristic indexes; and segmenting the regulation operation strategy data to obtain a regulation word bank, and constructing a space vector according to the regulation word bank.
4. The method for constructing the knowledge domain for the regulation and control operation of the multi-type energy supply system according to claim 3, wherein the processes of word segmentation and space vector construction are as follows: firstly, determining each document contained after query text preprocessing, segmenting each document according to a domain word bank to which the document belongs, substituting each alternative keyword belonging to each document into the obtained analysis model, obtaining word vectors of a plurality of dimensions formed by the alternative keywords in each document correspondingly, and outputting the word vectors as space vectors.
5. The method for constructing the knowledge graph of the regulation and control operation of the multi-type energy supply and consumption system according to claim 1, wherein the named entity recognition in the step 3.2 is to perform boundary determination and category recognition on entities with specific meanings in the regulation and control operation strategy data text, and divide the entities into equipment names, resource names, investment money, recycling periods and operation places by using an LR-CNN model according to the entity attributes and categories contained in the rural power distribution network.
6. The method for constructing a knowledge graph for regulating and controlling operation of a multi-type energy supply system according to claim 1, wherein the extracting of the relationship between the entities in the step 3.3 is to judge whether a predefined relationship exists between the entities based on named entity identification, so as to form a series of triple knowledge; and predefining the relation between the entities through the operation specification and the state of the power distribution network, and constructing a series of triples in the knowledge graph by using the relation formed after extraction.
7. The method for constructing the regulation operation knowledge graph for the multi-type energy supply system as claimed in claim 1, wherein the BERT-BilSTM-CRF model uses a pre-trained BERT model, the BilSTM-CRF model is connected behind the BERT model, and the output of the BERT model and the output of the BilSTM layer are connected in series by features.
8. The device for realizing the method for constructing the knowledge graph for regulating and controlling the operation of the multi-type energy supply system according to any one of claims 1 to 7 is characterized by comprising a knowledge graph mode layer module, a data preprocessing module, a knowledge extraction module, a knowledge graph data layer module, a strategy set module and a storage module;
the knowledge graph mode layer module is used for determining operation technical indexes of a regulation and control operation strategy of the multi-type energy supply system and sequentially refining the operation technical indexes to form a knowledge graph mode layer;
the data preprocessing module is used for preprocessing the data of the regulation and control operation strategy;
the knowledge extraction module is used for constructing different neural network models to realize knowledge extraction of the regulation and control operation strategy data and obtain triple data containing entity-attribute-entity;
the knowledge graph data layer module adds the triple data into a knowledge graph and adopts a bottom-up method to construct a knowledge graph data layer;
the strategy set module classifies the data of the knowledge graph data layer and combines the data with a knowledge graph mode layer to form a strategy set; the storage module is used for storing the knowledge graph.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211272027.7A CN115344717B (en) | 2022-10-18 | 2022-10-18 | Method and device for constructing regulation and control operation knowledge graph for multi-type energy supply and consumption system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211272027.7A CN115344717B (en) | 2022-10-18 | 2022-10-18 | Method and device for constructing regulation and control operation knowledge graph for multi-type energy supply and consumption system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115344717A CN115344717A (en) | 2022-11-15 |
| CN115344717B true CN115344717B (en) | 2023-02-17 |
Family
ID=83957099
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211272027.7A Active CN115344717B (en) | 2022-10-18 | 2022-10-18 | Method and device for constructing regulation and control operation knowledge graph for multi-type energy supply and consumption system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115344717B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112148893A (en) * | 2020-09-25 | 2020-12-29 | 南方电网数字电网研究院有限公司 | Energy analysis knowledge graph construction method and energy analysis visualization method |
| CN112612902A (en) * | 2020-12-23 | 2021-04-06 | 国网浙江省电力有限公司电力科学研究院 | Knowledge graph construction method and device for power grid main device |
| CN112905804A (en) * | 2021-02-22 | 2021-06-04 | 国网电力科学研究院有限公司 | Dynamic updating method and device for power grid dispatching knowledge graph |
| CN114077674A (en) * | 2021-10-31 | 2022-02-22 | 国电南瑞科技股份有限公司 | Power grid dispatching knowledge graph data optimization method and system |
| CN115033705A (en) * | 2022-04-21 | 2022-09-09 | 国网江苏省电力有限公司南京供电分公司 | Power grid regulation and control risk early warning information knowledge graph design method and system |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200226133A1 (en) * | 2016-10-18 | 2020-07-16 | Hithink Financial Services Inc. | Knowledge map building system and method |
| US11327989B2 (en) * | 2017-08-02 | 2022-05-10 | Accenture Global Solutions Limited | Multi-dimensional industrial knowledge graph |
| CN111930784B (en) * | 2020-07-23 | 2022-08-09 | 南京南瑞信息通信科技有限公司 | Power grid knowledge graph construction method and system |
| CN112242187B (en) * | 2020-10-26 | 2023-06-27 | 平安科技（深圳）有限公司 | Medical scheme recommendation system and method based on knowledge graph characterization learning |
| CN115170344A (en) * | 2022-05-30 | 2022-10-11 | 国网江苏省电力有限公司南京供电分公司 | Intelligent processing method and device, medium and equipment for operation events of regulation and control system |
-
2022
- 2022-10-18 CN CN202211272027.7A patent/CN115344717B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112148893A (en) * | 2020-09-25 | 2020-12-29 | 南方电网数字电网研究院有限公司 | Energy analysis knowledge graph construction method and energy analysis visualization method |
| CN112612902A (en) * | 2020-12-23 | 2021-04-06 | 国网浙江省电力有限公司电力科学研究院 | Knowledge graph construction method and device for power grid main device |
| CN112905804A (en) * | 2021-02-22 | 2021-06-04 | 国网电力科学研究院有限公司 | Dynamic updating method and device for power grid dispatching knowledge graph |
| CN114077674A (en) * | 2021-10-31 | 2022-02-22 | 国电南瑞科技股份有限公司 | Power grid dispatching knowledge graph data optimization method and system |
| CN115033705A (en) * | 2022-04-21 | 2022-09-09 | 国网江苏省电力有限公司南京供电分公司 | Power grid regulation and control risk early warning information knowledge graph design method and system |
Non-Patent Citations (6)
| Title |
|---|
| Construction technology of large scale knowledge maps in customer service field of Power Grid;Shuang Shuang Guo等;《2020 International Conference on Intelligent Computing and Human-Computer Interaction (ICHCI)》;20210510;全文 * |
| 基于知识图谱技术的风电数据管理与应用研究;吕梦平;《CNKI优秀硕士学位论文全文库》;20220515;全文 * |
| 基于知识图谱的电力知识平台关键技术研究;俞阳;《CNKI优秀硕士学位论文全文库》;20190515;全文 * |
| 知识图谱在智能电网的应用;王琼等;《电子元器件与信息技术》;20200120(第01期);全文 * |
| 调度自动化***知识图谱的构建与应用;李新鹏等;《中国电力》;20181208(第02期);全文 * |
| 面向智能调控领域的知识图谱构建与应用;余建明等;《电力***保护与控制》;20200201(第03期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115344717A (en) | 2022-11-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ntakolia et al. | Machine learning applied on the district heating and cooling sector: A review | |
| CN110851892B (en) | Park level comprehensive energy system aided design method and system | |
| Mate et al. | A hybrid integrated architecture for energy consumption prediction | |
| CN112559765B (en) | Semantic integration method for multi-source heterogeneous database | |
| CN112612819A (en) | Big data analysis and mining method and system for pumped storage power station | |
| CN115757810A (en) | Method for constructing standard ontology of knowledge graph | |
| Liu et al. | Varying condition SCOPF based on deep learning and knowledge graph | |
| Zhang et al. | The power big data-based energy analysis for intelligent community in smart grid | |
| Richarz et al. | Graph-based research field analysis by the use of natural language processing: An overview of German energy research | |
| CN115344717B (en) | Method and device for constructing regulation and control operation knowledge graph for multi-type energy supply and consumption system | |
| Lin et al. | Overview of multi-criteria decision analysis and its applications on energy systems | |
| CN116630088A (en) | Digital system architecture for electric power system science and technology attack | |
| CN115937881A (en) | Method for automatically identifying content of knowledge graph construction standard form | |
| Wang et al. | Integrated energy system multi-level planning model based on scenario reasoning, equipment selection, and capacity optimization | |
| Liu et al. | Review of multiple load forecasting method for integrated energy system | |
| CN115759253A (en) | Power grid operation and maintenance knowledge map construction method and system | |
| Cheng et al. | Reactive Power Load Forecasting based on K-means Clustering and Random Forest Algorithm | |
| Aktas et al. | A hesitant fuzzy linguistic group decision making model for energy storage unit selection | |
| CN115705364A (en) | Power grid monitoring knowledge graph construction method based on equipment operation characteristics | |
| Salleh et al. | A systematic literature review of electricity load forecasting using long short-term memory | |
| Xiong et al. | Experimental Evaluation of Intelligent e-Government System Based on Text Mining | |
| Ehlinger et al. | SKILLS OR A DEGREE?: THE RISE OF SKILL-BASED HIRING FOR AI AND GREEN JOBS. | |
| Gonzalez Ehlinger et al. | Finding Talent for the Twin Transition: Are UK Firms Shifting Towards Skill-Based Hiring for AI and Green Jobs? | |
| Zheng et al. | Research on text classification method of distribution network terminal debugging job based on hierarchical comprehensive context | |
| Shen et al. | Architecture and Implementation of Electricity Statistics Knowledge Graphs |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |