CN-121981123-A - Semantic analysis-based intelligent identification method, system, equipment and medium for power dispatching instruction

CN121981123ACN 121981123 ACN121981123 ACN 121981123ACN-121981123-A

Abstract

The invention discloses a semantic analysis-based intelligent recognition method, a semantic analysis-based intelligent recognition system, semantic analysis-based intelligent recognition equipment and semantic analysis-based intelligent recognition media for power dispatching instructions, which relate to the field of power dispatching instruction recognition and comprise the steps of obtaining power dispatching instructions generated by a dispatching platform and preprocessing the dispatching instructions to form standardized word sequence input; the method comprises the steps of inputting standardized word sequences, carrying out semantic coding on scheduling instructions to generate semantic representation representing overall semantic features of the scheduling instructions, carrying out instruction type identification on the scheduling instructions based on the semantic representation, determining instruction types, carrying out sequence labeling processing on the scheduling instructions under the constraint of the instruction types, extracting element information in the scheduling instructions, and outputting structural identification results based on the instruction types and the element information. The invention enables the dispatching instruction to be accurately understood and expressed in a standardized way under the condition of not depending on a fixed template and a keyword rule, thereby overcoming the problems of sensitivity to expression modes, insufficient generalization capability and unstable element extraction in the prior art.

Inventors

WANG LEI
ZHANG LIBO
HUANG LI
FU TONGFU
ZHANG JIANXING
Mou nan
YANG MINXUE
ZOU RUIRUI

Assignees

贵州电网有限责任公司

Dates

Publication Date: 20260505
Application Date: 20251218

Claims (10)

1. The intelligent power dispatching instruction recognition method based on semantic analysis is characterized by comprising the following steps of, Acquiring a power dispatching instruction in a natural language form generated by a dispatching platform, and performing text preprocessing on the dispatching instruction to form standardized word sequence input; Semantic coding is carried out on the scheduling instruction based on standardized word sequence input, and semantic representation used for representing the overall semantic features of the scheduling instruction is generated; Based on semantic representation, carrying out instruction type identification on the scheduling instruction, and determining the instruction type corresponding to the scheduling instruction; Under the constraint of the instruction type, carrying out sequence labeling processing on the scheduling instruction, and extracting element information in the scheduling instruction; Based on the instruction type and the element information, outputting a structural identification result corresponding to the scheduling instruction.
2. The intelligent power dispatching instruction recognition method based on semantic parsing according to claim 1, wherein the forming standardized word sequence input comprises, Based on a pre-constructed special dictionary in the electric power field, carrying out word segmentation processing on a scheduling instruction in a natural language form, and dividing an original character sequence into a plurality of entries to form a standardized word sequence; Aiming at each entry obtained by word segmentation, distributing corresponding preliminary semantic tags for each entry according to semantic roles in a scheduling instruction; And carrying out unified mapping processing on the vocabulary entries marked by the segmentation and semantic tags.
3. The intelligent power dispatching instruction recognition method based on semantic parsing as set forth in claim 2, wherein generating the semantic representation characterizing the overall semantic features of the dispatching instruction comprises, Converting the standardized word sequence input into a corresponding vectorization representation, so that each term becomes a term vector with uniform dimension; Semantic feature fusion is carried out on the term vectors based on the context modeling structure, context association relations among terms in the term sequences are comprehensively depicted, and intermediate representation reflecting overall semantic information of the scheduling instruction is generated; and carrying out aggregation processing on the intermediate representation to obtain semantic representation for representing the overall semantic features of the scheduling instruction.
4. The intelligent power dispatching instruction identification method based on semantic parsing according to claim 3, wherein the instruction type identification of the dispatching instruction comprises, Inputting the semantic representation into a multi-category classification model to map the semantic representation to a category space; based on the category space, judging the possibility that the semantic representation belongs to each preset instruction type, and forming an instruction type judging result; and determining the instruction type corresponding to the scheduling instruction according to the instruction type judging result.
5. The intelligent power dispatching instruction recognition method based on semantic parsing as set forth in claim 4, wherein the sequence labeling of dispatching instructions comprises, Based on the determined instruction type, constructing a corresponding element labeling constraint condition, and taking the constraint condition as a priori limit of sequence labeling processing; Under the action of element labeling constraint conditions, word-by-word labeling processing is carried out on word sequences corresponding to the scheduling instructions, and labeling results reflecting semantic roles of the words are generated; and extracting element information matched with the instruction type in the scheduling instruction according to the labeling result.
6. The intelligent power dispatching instruction recognition method based on semantic analysis of claim 5, wherein the structured recognition result corresponding to the output dispatching instruction comprises, Organizing and correlating the element information to construct an element structure matched with the instruction type; and (3) carrying out standardization processing on the element structure, and outputting structured data content as a structured recognition result of the scheduling instruction.
7. The intelligent power dispatching instruction recognition method based on semantic parsing as set forth in claim 6, wherein the structured recognition result of the dispatching instruction comprises, Transmitting the structured recognition result to a processing module corresponding to the power dispatching service; and executing the scheduling service processing logic matched with the instruction type to generate a corresponding scheduling processing result.
8. The intelligent power dispatching instruction recognition system based on semantic analysis is applied to the intelligent power dispatching instruction recognition method based on semantic analysis according to any one of claims 1-7, and is characterized by comprising a standardization module, a type recognition module, a recognition result module and a processing module; The standardized module acquires a power dispatching instruction in a natural language form generated by the dispatching platform and carries out text pretreatment on the dispatching instruction to form standardized word sequence input; The type identification module performs semantic coding on the scheduling instruction based on standardized word sequence input, generates semantic representation for representing the overall semantic features of the scheduling instruction, performs instruction type identification on the scheduling instruction based on the semantic representation, and determines the instruction type corresponding to the scheduling instruction; The recognition result module is used for carrying out sequence labeling processing on the scheduling instruction under the constraint of the instruction type, extracting element information in the scheduling instruction, and outputting a structural recognition result corresponding to the scheduling instruction based on the instruction type and the element information; and the processing module executes the scheduling service processing logic matched with the instruction type and generates a corresponding scheduling processing result.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of a semantic parsing based intelligent identification method for power scheduling instructions according to any one of claims 1 to 7.
10. A computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of a semantic parsing based intelligent identification method for power scheduling instructions as claimed in any one of claims 1 to 7.

Description

Semantic analysis-based intelligent identification method, system, equipment and medium for power dispatching instruction Technical Field The present invention relates to the field of power dispatch instruction identification, in particular to a semantic analysis-based intelligent identification method, system, equipment and medium for power dispatching instructions. Background With the continuous expansion of the power grid scale, the increasing complexity of the operation mode and the diversification of the scheduling scenes, the number and the complexity of the scheduling orders are obviously increased. A large number of dispatching instructions in natural language form are formed in the current power grid dispatching process, and the dispatching instructions relate to various kinds of information such as plant names, equipment numbers, voltage levels, action types, running states and the like. However, the existing dispatching instruction analysis system still highly depends on modes such as manual experience, fixed template matching or keyword retrieval, and the like, so that the method is difficult to effectively adapt to the writing habit, the word order difference and the diversity of expression modes of different dispatchers. Under the background, an intelligent analysis method capable of carrying out semantic understanding, automatically identifying instruction types, accurately extracting equipment and action parameters and outputting a unified structural result on natural language dispatching orders is urgently needed to support business scenes such as intelligent ticket formation, automatic dispatching execution and dispatching safety check. Aiming at the limitation of the prior art in power dispatching instruction analysis, the invention provides an intelligent power dispatching instruction identification method based on semantic analysis and sequence labeling, which is used for improving the automation degree and analysis accuracy of dispatching business. Disclosure of Invention In view of the above problems, the invention provides a semantic analysis-based intelligent identification method, system, equipment and medium for power dispatching instructions. Therefore, the invention aims to solve the problem that the existing dispatching instruction analysis system still highly depends on modes such as manual experience, fixed template matching or keyword retrieval, and the like, and is difficult to effectively adapt to the writing habit, the word sequence difference and the diversity of expression modes of different dispatchers. The intelligent recognition method for the power dispatching instruction based on semantic analysis comprises the steps of obtaining a power dispatching instruction in a natural language form generated by a dispatching platform, conducting text preprocessing on the dispatching instruction to form standardized word sequence input, conducting semantic coding on the dispatching instruction based on the standardized word sequence input to generate semantic representation used for representing overall semantic features of the dispatching instruction, conducting instruction type recognition on the dispatching instruction based on the semantic representation, determining an instruction type corresponding to the dispatching instruction, conducting sequence labeling processing on the dispatching instruction under constraint of the instruction type, extracting element information in the dispatching instruction, and outputting a structural recognition result corresponding to the dispatching instruction based on the instruction type and the element information. The intelligent recognition method for the power dispatching instructions based on semantic analysis is characterized in that the step of forming standardized word sequence input comprises the steps of carrying out word segmentation processing on dispatching instructions in a natural language form based on a pre-built power domain special dictionary, dividing an original character sequence into a plurality of entries to form a standardized word sequence, distributing corresponding preliminary semantic tags for the entries according to semantic roles in the dispatching instructions aiming at the entries obtained by word segmentation, and carrying out unified mapping processing on the entries marked by the word segmentation and the semantic tags. The intelligent recognition method for the power dispatching instruction based on semantic analysis comprises the steps of converting standardized word sequence input into corresponding vectorization representation to enable each term to be changed into a uniform-dimension term vector, carrying out semantic feature fusion on the term vector based on a context modeling structure, comprehensively describing context association relations among terms in the term sequence, generating intermediate representation reflecting overall semantic information of the dispatching instruction, and car