CN109165275B - Intelligent substation operation ticket information intelligent search matching method based on deep learning - Google Patents
Intelligent substation operation ticket information intelligent search matching method based on deep learning Download PDFInfo
- Publication number
- CN109165275B CN109165275B CN201810818470.7A CN201810818470A CN109165275B CN 109165275 B CN109165275 B CN 109165275B CN 201810818470 A CN201810818470 A CN 201810818470A CN 109165275 B CN109165275 B CN 109165275B
- Authority
- CN
- China
- Prior art keywords
- word
- information
- layer
- operation ticket
- neural network
- 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
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses an intelligent searching and matching method for operation ticket information of an intelligent substation based on deep learning. Similar to the work of monitoring information scheduling joint debugging of an intelligent station, verification of programmed operation in the station and the like, the work can be carried out manually one by one at present, a large amount of manpower and time are consumed, and the project progress is greatly influenced. The technical scheme adopted by the invention comprises the following steps: extracting operation object information in the operation ticket by using a keyword extraction technology; establishing a double-layer neural network model for intelligently searching and matching operation ticket information; and extracting the DO object matched with the operation order information from the SCD file by utilizing similarity calculation and sorting. The method and the device can establish the association between the operation ticket information and the SCD file based on the related technology of natural language processing, solve the intelligent identification and search matching of the transformer substation operation ticket information, and further serve the complete and efficient transformer substation intelligent debugging work.
Description
Technical Field
The invention relates to the field of artificial intelligence and power grid intelligent power transformation, in particular to an intelligent searching and matching method for operation ticket information of an intelligent substation based on deep learning.
Background
With the development of the smart power grid, the scale of the information network is continuously increased, the information exchange among different application systems is greatly increased, and the problem of influence of information links on a physical system is more prominent. The application integration of the power system is a necessary means for reducing workload, improving power quality and improving economic benefits of power enterprises, but the power enterprises are basically subjected to bar-block segmentation by depending on business departments in the informatization process, the application systems are mutually independent and highly autonomous, so that the problems of power grid modeling and inconsistent data names are serious, and the information matching is a problem which is urgently required to be solved by the application integration.
The problems are embodied as follows: in order to meet the requirements of various applications, a large amount of work needs to be carried out to verify whether the information is matched or not in the engineering debugging stage of the intelligent station. Similar to the work of monitoring information scheduling joint debugging of an intelligent station, verification of programmed operation in the station and the like, the work can be carried out manually one by one at present, a large amount of manpower and time are consumed, and the project progress is greatly influenced.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, and provides an intelligent searching and matching method for the operation ticket information of the intelligent substation based on deep learning, which realizes automatic operation to replace the conventional manual operation mode and provides technical support for the work in actual engineering production and debugging.
In order to solve the technical problems, the invention adopts the technical scheme that: the intelligent searching and matching method for the operation ticket information of the intelligent substation based on deep learning comprises the following steps:
s1, extracting the operation object information in the operation ticket by using a keyword extraction technology;
s2, establishing a double-layer neural network model for intelligently searching and matching the operation ticket information;
and S3, extracting the DO object matched with the operation ticket information from the SCD file by utilizing similarity calculation and sorting.
In addition to the above technical solution, in step S1, the keyword extraction technique uses a word formation segmentation method based on a BP neural network model and uses a recurrent neural network model to identify the operation object in the operation ticket according to the segmentation result.
As a supplement to the above technical solution, in step S1, the word segmentation is performed by the word formation word segmentation method based on the BP neural network model, and the process is as follows:
selecting K characters before and after each character of the operation ticket, and inputting 2 x K +1 characters together with the character; selecting the vector dimension of each word as L, and inputting the dimension L (2X K +1) of the neural network model; adopting three layers of BP neural networks, wherein an input layer is provided with L (2 x K +1) neurons, a hidden layer is provided with L neurons, an output layer is provided with 4 neurons, the lexeme of each neuron corresponding to a character is respectively the probability of S, B, M and E, and the probability of S, B, M and E is respectively corresponding to the character to form a word, a word head, a word middle and a word tail; and selecting the word position with the highest probability from the output word position probability values as the word position category of the word.
In order to utilize the context information of each character of the operation ticket, the invention adopts a BP neural network in a knowledge-based word segmentation method to segment words. The method has good memory capacity of the context information of each word of the operation ticket and has good fault-tolerant effect; meanwhile, the method adopts a nonlinear parallel processing method, and can automatically process and learn the obtained knowledge and information in word segmentation, thereby solving the speed problem caused by more operation order information.
As a supplement to the above technical solution, in step S1, the recognition of the operation object in the operation ticket is realized by using the recurrent neural network model according to the word segmentation result, and the process is as follows:
sequentially converting the obtained word segmentation sequence into word vectors, and selecting the vector dimension of a word as M; setting a word number N with enough length, and filling the space behind the insufficient word number N, wherein the input of the cyclic neural network model is a vector of M x N and corresponds to an input layer of the cyclic neural network model; the second layer is a double LSTM (Long Short Term Memory Network); the output layer outputs 2-dimensional classification conditions, the output classification results are 0 and 1, whether the word is an operation object or not is represented, 0 represents no, and 1 represents yes.
In order to solve the problems of time sequence sensitivity of each word in the operation order and uncertain content length of the operation order, the invention adopts a Recurrent Neural Network (RNN) model.
In addition to the above technical solution, in step S2, the rule adopted for searching for matching is: judging whether the groups of the two pieces of information are the same or not, and if the groups of the two pieces of information are different, determining that the two pieces of information are not matched; and if the two pieces of information belong to the same group, judging whether the specific equipment information of the two pieces of information is the same or not, and giving a final matching result.
Since the general format of the line or device information in the operation ticket is "interval + specific device information". The degree of similarity between information of devices at the same interval is obviously high compared with different intervals. Thus, the present invention abstracts the intervals into groups.
In step S2, based on the search matching rule, a two-layer neural network model is used, and the first layer area is divided into an inner item and an outer item: the outside group is '0' and the inside group is '1'; the second layer distinguishes the matching term: the intra-group mismatch is '0' and the intra-group match is '1'; and the output of the model is the result of whether the operation order information is matched with the DO object in the SCD file.
As a supplement to the above technical solution, the double-layer neural network has four structures according to the difference of the neural network adopted in each layer, which are: BP _1, BP _2, the first, second layer which represents the model adopts BP model; BP _1 and LSTM _2, which represent that the first layer of the model adopts a BP model and the second layer adopts an LSTM model; LSTM _1 and BP _2, which represent that the first layer of the model adopts an LSTM model and the second layer adopts a BP model; LSTM _1 and LSTM _2 represent that the first and second layers of the model both adopt LSTM models.
As a supplement to the above technical solution, in step S3, according to the probabilities that the output result of the neural network in step S2 is '0' and '1', it is determined whether the operation order information matches all DO objects in the SCD file in a traversal manner; the probability of '1' is the similarity or the matching degree, and the size of the probability represents the matching degree of the operation ticket information and the SCD information, so that the DO object in the SCD file with the maximum matching degree is selected as the best matching result of a certain operation ticket information.
Compared with the prior art, the invention has the beneficial effects that:
(1) chinese word segmentation is realized by using a BP neural network.
(2) And the operation ticket keyword extraction is completed by utilizing the recurrent neural network.
(3) And the searching and matching of the operation order information are realized by utilizing a double-layer deep learning model (namely, a double-layer neural network model).
The method and the device can establish the association between the operation ticket information and the SCD file based on the related technology of natural language processing, solve the intelligent identification and search matching of the transformer substation operation ticket information, and further serve the complete and efficient transformer substation intelligent debugging work.
Drawings
FIG. 1 is a flow chart of an intelligent searching and matching method for operation ticket information of an intelligent substation based on deep learning according to the invention;
fig. 2 is a flowchart illustrating the operation order information and the DO object search matching in the SCD file according to the present invention.
Detailed Description
The drawings are for illustration purposes only and are not to be construed as limiting the invention; it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted. The positional relationships depicted in the drawings are for illustrative purposes only and are not to be construed as limiting the invention.
The technical scheme of the invention is described in detail in the following with reference to the attached drawings. The implementation of the technical scheme of the invention mainly comprises the following steps (the flow chart is shown in the attached figure 1):
s1, extracting the operation object information in the operation ticket by using the key word extraction technology
(1) And carrying out Chinese word segmentation by using a BP neural network model.
The front and back 3 words of each word in the operation ticket are taken, and the front and back 3 words and the word are input to total 7 words. The vector dimension of each word is selected to be 100, and the dimension 100 (2 x 3+1) of the input neural network model is 700. The method adopts a three-layer BP neural network, wherein an input layer comprises 700 neurons, a hidden layer comprises 100 neurons, an output layer comprises 4 neurons, and the lexeme of each neuron corresponding to a character is respectively S, B, M and E (S, B, M and E are respectively the probability of independent word formation (S), the head of the word (B), the middle of the word (M) and the tail of the word (E)).
And selecting the word position with the highest probability from the output word position probability values as the word position category of the word, and sequentially carrying out the same treatment on each word of the operation ticket so as to finish the word segmentation of the whole operation ticket. For example, the operation ticket 'the closing immortal line 50121 knife switch' is obtained after the word segmentation.
(2) And identifying the information of the operation object in the operation ticket by utilizing a recurrent neural network model according to the word segmentation result.
And (3) sequentially converting the word segmentation sequence obtained in the step (1) into word vectors, and selecting the vector dimension of the word as 10. The number of words 30 long enough (the space after the deficiency is filled) is taken as the dimension of fixed input, and then the input of the neural network model is a vector of 30 x 10, which corresponds to the input layer of the recurrent neural network model; the second layer is double LSTM, and the hidden layer is 128 neurons; the output layer outputs 2-dimensional classification cases represented by 0,1 (0 and 1 respectively correspond to whether the word is a device or a line, 1 represents yes, and 0 represents no).
After the operation, the line and equipment information indicated by 'xian yong 50122 knife switch' is extracted from the operation ticket word segmentation result.
S2, establishing a double-layer neural network model for intelligently searching and matching the operation ticket information;
(1) the DO object in the SCD file is input as information 1.
(2) A search matching rule is determined. The information after the operation ticket extraction is taken as information 2. The matching rule of the two pieces of information is as follows: firstly, judging whether the groups of the two pieces of information are the same or not, and if the groups of the two pieces of information are different, determining that the two pieces of information are not matched; if the two pieces of information belong to the same group, whether the specific equipment information of the two pieces of information is the same or not is judged, and a final matching result is given (a searching rule flow chart is shown in an attached figure 2).
Taking the operation order as the 'closing immortal line 50121 knife switch' example, the search matching result is shown in table 1.
Table 1 operation ticket information search matching SCD example
S3, extracting DO object matched with operation ticket information from SCD file by utilizing similarity calculation and sorting
The best result of searching matching is 'xian permanent line 5012 switch, 50121 knife switch closing', which is obtained by sequencing and comparing the calculated results of matching degree.
It should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.
Claims (3)
1. The intelligent searching and matching method for the operation ticket information of the intelligent substation based on deep learning is characterized by comprising the following steps of:
s1, extracting the operation object information in the operation ticket by using a keyword extraction technology;
s2, establishing a double-layer neural network model for intelligently searching and matching the operation ticket information;
s3, extracting a DO object matched with the operation order information from the SCD file by utilizing similarity calculation and sorting;
in step S1, the keyword extraction technology utilizes a word formation and segmentation method based on a BP neural network model and utilizes a cyclic neural network model to realize the identification of an operation object in an operation ticket according to the segmentation result;
in step S1, the word segmentation is performed by the word formation word segmentation method based on the BP neural network model, and the process is as follows:
selecting K characters before and after each character of the operation ticket, and inputting 2 x K +1 characters together with the character; selecting the vector dimension of each word as L, and inputting the dimension L (2X K +1) of the neural network model; adopting three layers of BP neural networks, wherein an input layer is provided with L (2 x K +1) neurons, a hidden layer is provided with L neurons, an output layer is provided with 4 neurons, the lexeme of each neuron corresponding to a character is respectively the probability of S, B, M and E, and the probability of S, B, M and E is respectively corresponding to the character to form a word, a word head, a word middle and a word tail; selecting the word position with the maximum probability from the output word position probability values as the word position category of the word;
in step S1, the cyclic neural network model is used to identify the operation object in the operation ticket according to the word segmentation result, and the process is as follows:
sequentially converting the obtained word segmentation sequence into word vectors, and selecting the vector dimension of a word as M; setting a word number N with enough length, and filling the space behind the insufficient word number N, wherein the input of the cyclic neural network model is a vector of M x N and corresponds to an input layer of the cyclic neural network model; the second layer is double LSTM; the output layer outputs 2-dimensional classification conditions, the output classification results are 0 and 1, whether the word is an operation object or not is represented, 0 represents no, and 1 represents yes;
in step S2, the rule used for searching for matching is: judging whether the groups of the two pieces of information are the same or not, and if the groups of the two pieces of information are different, determining that the two pieces of information are not matched; if the two pieces of information belong to the same group, judging whether the specific equipment information of the two pieces of information is the same or not, and giving a final matching result;
in step S2, based on the search matching rule, a two-layer neural network model is used, and the first layer area is divided into an inner item and an outer item: the outside group is '0' and the inside group is '1'; the second layer distinguishes the matching term: the intra-group mismatch is '0' and the intra-group match is '1'; and the output of the model is the result of whether the operation order information is matched with the DO object in the SCD file.
2. The intelligent searching and matching method for the operation ticket information of the intelligent substation based on the deep learning of claim 1 is characterized in that according to the difference of the neural networks adopted in each layer, the double-layer neural network has four structures, which are respectively: BP _1, BP _2, the first, second layer which represents the model adopts BP model; BP _1 and LSTM _2, which represent that the first layer of the model adopts a BP model and the second layer adopts an LSTM model; LSTM _1 and BP _2, which represent that the first layer of the model adopts an LSTM model and the second layer adopts a BP model; LSTM _1 and LSTM _2 represent that the first and second layers of the model both adopt LSTM models.
3. The intelligent searching and matching method for the operation ticket information of the intelligent substation based on the deep learning of claim 2, wherein in step S3, according to the probability that the output result of the neural network in step S2 is '0' and '1', it is determined through traversal whether the operation ticket information matches with all DO objects in the SCD file; the probability of '1' is the similarity or the matching degree, and the size of the probability represents the matching degree of the operation ticket information and the SCD information, so that the DO object in the SCD file with the maximum matching degree is selected as the best matching result of a certain operation ticket information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810818470.7A CN109165275B (en) | 2018-07-24 | 2018-07-24 | Intelligent substation operation ticket information intelligent search matching method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810818470.7A CN109165275B (en) | 2018-07-24 | 2018-07-24 | Intelligent substation operation ticket information intelligent search matching method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109165275A CN109165275A (en) | 2019-01-08 |
| CN109165275B true CN109165275B (en) | 2021-03-02 |
Family
ID=64898207
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810818470.7A Active CN109165275B (en) | 2018-07-24 | 2018-07-24 | Intelligent substation operation ticket information intelligent search matching method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109165275B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111353487A (en) * | 2019-08-20 | 2020-06-30 | 国网浙江省电力有限公司台州供电公司 | Equipment information extraction method for transformer substation |
| CN111523241B (en) * | 2020-04-28 | 2023-06-13 | 国网浙江省电力有限公司湖州供电公司 | Construction method of power load logic information model |
| CN112017411A (en) * | 2020-08-21 | 2020-12-01 | 广东电网有限责任公司清远供电局 | Substation operation anti-error early warning system |
| CN112749509B (en) * | 2020-12-30 | 2022-06-10 | 西华大学 | Intelligent substation fault diagnosis method based on LSTM neural network |
| CN112700008B (en) * | 2021-01-06 | 2022-06-28 | 青岛弯弓信息技术有限公司 | Model matching processing method and system for cloud configuration platform |
| CN113656532B (en) * | 2021-08-16 | 2024-05-07 | 雅砻江流域水电开发有限公司 | Intelligent retrieval system for work ticket and operation ticket |
| CN117093675A (en) * | 2023-10-18 | 2023-11-21 | 江苏泰坦智慧科技有限公司 | Operation ticket searching and matching method, device and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5630020A (en) * | 1989-06-09 | 1997-05-13 | U.S. Philips Corporation | Learning method and neural network structure |
| CN105224511A (en) * | 2015-09-23 | 2016-01-06 | 山东网聪信息科技有限公司 | A kind of method of calibration of intelligent substation SCD file |
| CN105975456A (en) * | 2016-05-03 | 2016-09-28 | 成都数联铭品科技有限公司 | Enterprise entity name analysis and identification system |
| CN107153946A (en) * | 2017-04-28 | 2017-09-12 | 国家电网公司 | Intelligent station is secondary to pacify automatic generation method and the system of arranging |
| CN108021552A (en) * | 2017-11-09 | 2018-05-11 | 国网浙江省电力公司电力科学研究院 | A kind of power system operation ticket method for extracting content and system |
-
2018
- 2018-07-24 CN CN201810818470.7A patent/CN109165275B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5630020A (en) * | 1989-06-09 | 1997-05-13 | U.S. Philips Corporation | Learning method and neural network structure |
| CN105224511A (en) * | 2015-09-23 | 2016-01-06 | 山东网聪信息科技有限公司 | A kind of method of calibration of intelligent substation SCD file |
| CN105975456A (en) * | 2016-05-03 | 2016-09-28 | 成都数联铭品科技有限公司 | Enterprise entity name analysis and identification system |
| CN107153946A (en) * | 2017-04-28 | 2017-09-12 | 国家电网公司 | Intelligent station is secondary to pacify automatic generation method and the system of arranging |
| CN108021552A (en) * | 2017-11-09 | 2018-05-11 | 国网浙江省电力公司电力科学研究院 | A kind of power system operation ticket method for extracting content and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109165275A (en) | 2019-01-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109165275B (en) | Intelligent substation operation ticket information intelligent search matching method based on deep learning | |
| CN111914644B (en) | Dual-mode cooperation based weak supervision time sequence action positioning method and system | |
| Mei et al. | Sgnn: A graph neural network based federated learning approach by hiding structure | |
| WO2018218707A1 (en) | Neural network and attention mechanism-based information relation extraction method | |
| CN110309331A (en) | A kind of cross-module state depth Hash search method based on self-supervisory | |
| CN103345656B (en) | A kind of data identification method based on multitask deep neural network and device | |
| CN113610540B (en) | River crab anti-counterfeiting tracing method and system | |
| CN111597420B (en) | Rail transit standard relation extraction method based on deep learning | |
| CN110188192B (en) | Multi-task network construction and multi-scale criminal name law enforcement combined prediction method | |
| CN107918633B (en) | Sensitive public opinion content identification method and early warning system based on semantic analysis technology | |
| CN112419096B (en) | Automatic circulation method of user electricity appeal work order based on NLP information extraction and few-sample self-learning | |
| CN111881677A (en) | Address matching algorithm based on deep learning model | |
| CN111522965A (en) | Question-answering method and system for entity relationship extraction based on transfer learning | |
| CN111753024A (en) | Public safety field-oriented multi-source heterogeneous data entity alignment method | |
| CN113407660B (en) | Unstructured text event extraction method | |
| CN102289522A (en) | Method of intelligently classifying texts | |
| CN112395417A (en) | Network public opinion evolution simulation method and system based on deep learning | |
| CN113434688B (en) | Data processing method and device for public opinion classification model training | |
| CN109871449A (en) | A kind of zero sample learning method end to end based on semantic description | |
| CN113849653A (en) | Text classification method and device | |
| CN116010581A (en) | Knowledge graph question-answering method and system based on power grid hidden trouble shooting scene | |
| Parvathi et al. | Identifying relevant text from text document using deep learning | |
| CN109543038B (en) | Emotion analysis method applied to text data | |
| CN114238524B (en) | Satellite frequency-orbit data information extraction method based on enhanced sample model | |
| CN111428502A (en) | Named entity labeling method for military corpus |
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 |