CN110110095A - A kind of power command text matching technique based on shot and long term memory Recognition with Recurrent Neural Network - Google Patents
A kind of power command text matching technique based on shot and long term memory Recognition with Recurrent Neural Network Download PDFInfo
- Publication number
- CN110110095A CN110110095A CN201910355114.0A CN201910355114A CN110110095A CN 110110095 A CN110110095 A CN 110110095A CN 201910355114 A CN201910355114 A CN 201910355114A CN 110110095 A CN110110095 A CN 110110095A
- Authority
- CN
- China
- Prior art keywords
- shot
- long term
- neural network
- term memory
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 14
- 230000007787 long-term memory Effects 0.000 title claims abstract description 9
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 8
- 230000000306 recurrent effect Effects 0.000 title claims abstract description 8
- 238000013135 deep learning Methods 0.000 claims abstract description 9
- 230000005611 electricity Effects 0.000 claims abstract description 6
- 238000004458 analytical method Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 5
- 238000007689 inspection Methods 0.000 claims description 4
- 238000013461 design Methods 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims 1
- 238000004519 manufacturing process Methods 0.000 claims 1
- 238000012098 association analyses Methods 0.000 abstract description 4
- 238000005516 engineering process Methods 0.000 abstract description 4
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 239000012141 concentrate Substances 0.000 abstract description 2
- 238000000605 extraction Methods 0.000 abstract description 2
- 239000000203 mixture Substances 0.000 abstract description 2
- 238000000513 principal component analysis Methods 0.000 abstract description 2
- 230000009467 reduction Effects 0.000 abstract description 2
- 230000005856 abnormality Effects 0.000 abstract 1
- 238000012549 training Methods 0.000 description 15
- 230000011218 segmentation Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000013502 data validation Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000011022 operating instruction Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 206010028916 Neologism Diseases 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000001364 causal effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 230000002688 persistence Effects 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
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
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The invention discloses a kind of modules such as power command text matching technique, including feature extraction, association analysis based on shot and long term memory Recognition with Recurrent Neural Network.Concentrate the time series for the electricity consumption data for extracting user as initial characteristics collection from initial data first, then nondimensionalization is carried out to feature set and feature selecting is handled, and dimensionality reduction is carried out to initial characteristics collection using Principal Component Analysis and autoencoder network and obtains validity feature collection, isolated forest algorithm is finally used to calculate the abnormality score of each user to determine that it is without exception that user data has.Compared with prior art, beneficial effects of the present invention make full use of current each professional advanced technology, under open information technology frameworks system, it blends and is connected with existing related system, intelligence based on deep learning is made out an invoice application, solve the problems, such as two, first is that the primitive character of operation order indicates;Second is that selecting suitable learning algorithm.
Description
Technical field
The present invention relates to a kind of electric power based on shot and long term memory Recognition with Recurrent Neural Network for field of electric power automation to refer to
Enable text matching technique.
Background technique
With deepening continuously for electric Power Reform, power grid scale is growing, and the complicated degree of dispatching of power netwoks operation is continuous
Increase, it is a urgent problem to be solved that the scheduling accident as caused by various human factors, which happens occasionally,.However current power grid
Operation is still carried out according to original mode, is dispatched anti-error still without proposing as a basic demand and according to corresponding industry
Specification implements in power grid and system Construction.The anti-misoperation method for developing the human work of a kind of pair of electric operating is technical staff
Main target.
Summary of the invention
It is a kind of based on shot and long term memory circulation nerve net the purpose of the invention is to overcome the deficiencies of the prior art and provide
The power command text matching technique of network is issued electricity system operational order ticket and is operated item by item applied to scheduling, transport inspection department door personnel
Instruct ticket.
Realizing a kind of technical solution of above-mentioned purpose is: a kind of power command based on shot and long term memory Recognition with Recurrent Neural Network
Text matching technique, which is characterized in that include the following steps,
Step 1, the electricity system operation order and operational order ticket of scheduling, transport inspection department door are collected, analysis is semantic, and Rule Summary mentions
Rule is taken, is established a kind of suitable for the actual text matches data set of work about electric power;
Step 2, based on the text matches data set, existing all kinds of deep learning algorithms is analyzed and are applicable in scene, in conjunction with
Produced on-site is practical, considers device type, equipment running status, the power station mode of connection, the influence of the factors such as line construction, design
Meet the deep learning algorithm model of mission requirements;
Step 3, using existing history ticket, model is trained, test model is ultimately formed, then uses truthful data
Validation test model, and tuning is carried out to model, until ultimately forming optimal models.
A kind of power command text matching technique based on shot and long term memory Recognition with Recurrent Neural Network of the invention, including feature
The modules such as extraction, association analysis.Concentrate the time series for the electricity consumption data for extracting user as initial special from initial data first
Then collection carries out nondimensionalization to feature set and feature selecting is handled, and uses Principal Component Analysis and autoencoder network
Dimensionality reduction is carried out to initial characteristics collection and obtains validity feature collection, the exception point of each user is finally calculated using isolated forest algorithm
Number is to determine that it is without exception that user data has.Compared with prior art, beneficial effects of the present invention make full use of current each profession first
Into technology, under open information technology frameworks system, blends and be connected with existing related system, based on deep learning
Intelligence is made out an invoice application, solves the problems, such as two, first is that the primitive character of operation order indicates;Second is that selecting suitable learning algorithm.
Specific embodiment
In order to preferably understand technical solution of the present invention, carried out in detail below by specifically embodiment
Illustrate:
A kind of power command text matching technique based on shot and long term memory Recognition with Recurrent Neural Network of the invention, including it is as follows
Step,
Step 1, the electricity system operation order and operational order ticket of scheduling, transport inspection department door are collected, analysis is semantic, and Rule Summary mentions
Rule is taken, is established a kind of suitable for the actual text matches data set of work about electric power;
Step 2, based on the text matches data set, existing all kinds of deep learning algorithms is analyzed and are applicable in scene, in conjunction with
Produced on-site is practical, considers device type, equipment running status, the power station mode of connection, the influence of the factors such as line construction, design
Meet the deep learning algorithm model of mission requirements;
Step 3, using existing history ticket, model is trained, test model is ultimately formed, then uses truthful data
Validation test model, and tuning is carried out to model, until ultimately forming optimal models.
1, text segments
For the text data of electric grid operating instruction, the mature self-contained public sphere of participle tool can use
The dictionary for word segmentation of electrical network field is constructed on the basis of dictionary.Firstly, being named Entity recognition using conditional random field models, select
Use the Open-Source Tools of CRF++ as the basis of implementation tool, feature templates are special plus beginning word and end word using individual character feature
The method optimizing feature selection approach of sign.Name entity extension electrical network field vocabulary is extracted based on the above method, imports participle
Dictionary is segmented.It is the difficult point and emphasis that this part works that identified entity, which is solved, in the professional accordance of electrical network field,
Need to rely on artificial means to be screened, supplement dictionary for word segmentation, then carry out in next step training and screening, it may be found that neologisms be added
The update of dictionary is completed into dictionary.And optimize dictionary and essentially consist in update dictionary, to optimize dictionary, it is accurate to be continuously improved
Rate reduces rate of false alarm, is excessively taken turns iteration until Entity recognition performance is stablized.Participle word is realized based on this semi-supervised mode
The optimization of allusion quotation address only not to the utmost supervised learning method labor workload is huge and infeasible problem, algorithm can also be passed through
Optimization be continuously improved participle accuracy rate.
Multidimensional association analysis
Multidimensional association analysis is realized to electric grid operating instruction text word segmentation result.Based on Principle of Statistics different dimensions it
Between be associated with after do data analysis.Common multidimensional analysis method has regression analysis, variance analysis etc..Carrying out multidimensional data analysis
When, it can also often use relation analysis model.Correlation rule reflects dependence or associated knowledge, association rule mining between things
It is to find association, correlation in the Item Sets in database or information knowledge library or object set or have causal information, relates to
And two important parameters be minimum support and the minimum confident degree.Support indicates regular former piece and consequent simultaneously in data set
The probability of middle appearance;Degree of belief indicates the probability that consequent can be released when regular former piece is set up, in other words consequent
Credibility relative to regular former piece.Correlation refers to a kind of connection that can occur of other things when certain things occurs, can be with
It is described by associated support and confidence level.
Model construction and training
By patterned way or from coding developing algorithm.Other than supporting specified algorithm, also support user is customized
Algoritic module, the editing area at interface can be added to by way of dragging, and be modified.
Big data algorithm running environment provides distributed algorithm training management mechanism, simplifies the programming of client's intelligent algorithm
Complexity, by deep learning training use distributed arithmetic mode submission system running background, promoted computational efficiency, accelerate knot
Fruit generates.
(1) training mission is passed to customized training mission parameter by modes such as environmental variance or command line parameters, or
Local mirror image can also be used to start customized environmental training task in person user.
(2) training mission priority adjusts
Platform supports training mission priority query, according to the needs of real work, can dynamically adjust all training and appoint
The priority of business.When resource quota is not enough to be supported on all training missions of operation, platform by according to the priority of setting,
Preferentially allocate resources to the high training mission of priority.
(3) training log details are checked
The training log details management of persistence is provided, can check the operating condition of each training mission.
Those of ordinary skill in the art it should be appreciated that more than embodiment be intended merely to illustrate the present invention,
And be not used as limitation of the invention, as long as the change in spirit of the invention, to embodiment described above
Change, modification is all fallen within the scope of claims of the present invention.
Claims (1)
1. a kind of power command text matching technique based on shot and long term memory Recognition with Recurrent Neural Network, which is characterized in that including such as
Lower step,
Step 1, the electricity system operation order and operational order ticket of scheduling, transport inspection department door are collected, analysis is semantic, Rule Summary, extracts rule
Then, it establishes a kind of suitable for the actual text matches data set of work about electric power;
Step 2, based on the text matches data set, existing all kinds of deep learning algorithms is analyzed and are applicable in scene, in conjunction with scene
Production is practical, considers device type, equipment running status, the power station mode of connection, the influence of the factors such as line construction, design satisfaction
The deep learning algorithm model of mission requirements;
Step 3, using existing history ticket, model is trained, test model is ultimately formed, is then verified with truthful data
Test model, and tuning is carried out to model, until ultimately forming optimal models.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910355114.0A CN110110095A (en) | 2019-04-29 | 2019-04-29 | A kind of power command text matching technique based on shot and long term memory Recognition with Recurrent Neural Network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910355114.0A CN110110095A (en) | 2019-04-29 | 2019-04-29 | A kind of power command text matching technique based on shot and long term memory Recognition with Recurrent Neural Network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110110095A true CN110110095A (en) | 2019-08-09 |
Family
ID=67487375
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910355114.0A Pending CN110110095A (en) | 2019-04-29 | 2019-04-29 | A kind of power command text matching technique based on shot and long term memory Recognition with Recurrent Neural Network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110110095A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111260338A (en) * | 2020-02-19 | 2020-06-09 | 云南电网有限责任公司昆明供电局 | Intelligent generation method, device and platform for substation operation ticket |
CN111444704A (en) * | 2020-03-27 | 2020-07-24 | 中南大学 | Network security keyword extraction method based on deep neural network |
CN112926633A (en) * | 2021-02-01 | 2021-06-08 | 长江慧控科技(武汉)有限公司 | Abnormal energy consumption detection method, device, equipment and storage medium |
CN113095422A (en) * | 2021-04-21 | 2021-07-09 | 广东电网有限责任公司 | Method and device for realizing automatic programming order issuing |
CN113408525A (en) * | 2021-06-17 | 2021-09-17 | 成都崇瑚信息技术有限公司 | Multilayer ternary pivot and bidirectional long-short term memory fused text recognition method |
CN113627514A (en) * | 2021-08-05 | 2021-11-09 | 南方电网数字电网研究院有限公司 | Data processing method and device of knowledge graph, electronic equipment and storage medium |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108763542A (en) * | 2018-05-31 | 2018-11-06 | 中国华戎科技集团有限公司 | A kind of Text Intelligence sorting technique, device and computer equipment based on combination learning |
CN109446314A (en) * | 2018-11-14 | 2019-03-08 | 沈文策 | A kind of customer service question processing method and device |
-
2019
- 2019-04-29 CN CN201910355114.0A patent/CN110110095A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108763542A (en) * | 2018-05-31 | 2018-11-06 | 中国华戎科技集团有限公司 | A kind of Text Intelligence sorting technique, device and computer equipment based on combination learning |
CN109446314A (en) * | 2018-11-14 | 2019-03-08 | 沈文策 | A kind of customer service question processing method and device |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111260338A (en) * | 2020-02-19 | 2020-06-09 | 云南电网有限责任公司昆明供电局 | Intelligent generation method, device and platform for substation operation ticket |
CN111260338B (en) * | 2020-02-19 | 2022-03-29 | 云南电网有限责任公司昆明供电局 | Intelligent generation method, device and platform for substation operation ticket |
CN111444704A (en) * | 2020-03-27 | 2020-07-24 | 中南大学 | Network security keyword extraction method based on deep neural network |
CN111444704B (en) * | 2020-03-27 | 2023-09-19 | 中南大学 | Network safety keyword extraction method based on deep neural network |
CN112926633A (en) * | 2021-02-01 | 2021-06-08 | 长江慧控科技(武汉)有限公司 | Abnormal energy consumption detection method, device, equipment and storage medium |
CN113095422A (en) * | 2021-04-21 | 2021-07-09 | 广东电网有限责任公司 | Method and device for realizing automatic programming order issuing |
CN113408525A (en) * | 2021-06-17 | 2021-09-17 | 成都崇瑚信息技术有限公司 | Multilayer ternary pivot and bidirectional long-short term memory fused text recognition method |
CN113627514A (en) * | 2021-08-05 | 2021-11-09 | 南方电网数字电网研究院有限公司 | Data processing method and device of knowledge graph, electronic equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110110095A (en) | A kind of power command text matching technique based on shot and long term memory Recognition with Recurrent Neural Network | |
CN106295807B (en) | A kind of method and device of information processing | |
CN111143576A (en) | Event-oriented dynamic knowledge graph construction method and device | |
CN109472024A (en) | A kind of file classification method based on bidirectional circulating attention neural network | |
CN110866093A (en) | Machine question-answering method and device | |
KR20200007969A (en) | Information processing methods, terminals, and computer storage media | |
CN110287482B (en) | Semi-automatic participle corpus labeling training device | |
CN114722839B (en) | Man-machine cooperative dialogue interaction system and method | |
CN109241030A (en) | Robot manipulating task data analytics server and robot manipulating task data analysing method | |
CN103744977A (en) | Monitoring method and monitoring system for cloud computing system platform | |
CN113435582B (en) | Text processing method and related equipment based on sentence vector pre-training model | |
CN110175272A (en) | One kind realizing the convergent control method of work order and control device based on feature modeling | |
CN115688920A (en) | Knowledge extraction method, model training method, device, equipment and medium | |
CN113095050A (en) | Intelligent ticketing method, system, equipment and storage medium | |
CN110321555A (en) | A kind of power network signal classification method based on Recognition with Recurrent Neural Network model | |
CN109614612A (en) | A kind of Chinese text error correction method based on seq2seq+attention | |
CN113434688A (en) | Data processing method and device for public opinion classification model training | |
CN112036185B (en) | Method and device for constructing named entity recognition model based on industrial enterprise | |
CN113312924A (en) | Risk rule classification method and device based on NLP high-precision analysis label | |
CN115809796B (en) | Project intelligent dispatching method and system based on user portrait | |
CN112307767A (en) | Bi-LSTM technology-based regulation and control knowledge modeling method | |
CN111352820A (en) | Method, equipment and device for predicting and monitoring running state of high-performance application | |
CN108363738B (en) | Recommendation method for industrial equipment data analysis algorithm | |
CN113554280B (en) | Training method, device, equipment and storage medium of power grid system scheduling model | |
CN115203427A (en) | Power system regulation and control regulation knowledge graph management system, knowledge graph generation method and storage medium |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190809 |
|
WD01 | Invention patent application deemed withdrawn after publication |