CN109344976A - A kind of electrical system operating status intelligent analysis method based on Keras - Google Patents
A kind of electrical system operating status intelligent analysis method based on Keras Download PDFInfo
- Publication number
- CN109344976A CN109344976A CN201810971417.0A CN201810971417A CN109344976A CN 109344976 A CN109344976 A CN 109344976A CN 201810971417 A CN201810971417 A CN 201810971417A CN 109344976 A CN109344976 A CN 109344976A
- Authority
- CN
- China
- Prior art keywords
- electrical system
- operating status
- system operating
- data
- 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.)
- Pending
Links
Classifications
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
-
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Strategic Management (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The present invention discloses a kind of electrical system operating status intelligent analysis method based on Keras, by acquiring power station electrical system operation history data, is stored in OpenTSDB database, constructs the history vectors database of electrical system operating status;Determine that every group of sample number batch_size carries out MLP parameter initialization from history vectors database;The data for loading history vectors database, are classified as including training set and test set;Multilayer neural network structure is constructed, electrical system operating status intelligent measurement model is constructed;Training set data is passed to model as input training data, exports the predicted value and the corresponding threshold value of predicted value of electrical system operating status;Obtain the measured value of electrical system operating status, compare measured value threshold value corresponding with predicted value, predicted value, carry out prediction alarm, operating condition is prompted to deviate, guidance restores in time, it avoids causing safety problem, it is accurate to the prediction of electrical each system difference operating condition, improve the safety and stability of power station electrical system operation.
Description
Technical field
The present invention relates to power station electrical system running technology application field more particularly to a kind of electrical systems based on Keras
System operating status intelligent analysis method.
Background technique
The long-term heavy-duty service of electrical system in large-scale power station unit, especially during peak load regulation is run, machine
Group start and stop or variable load operation can make electrical system mutually should bear significantly load variation and alternating electromagnetic field, fortune with equipment
Line mode can not only influence the stability of unit operation with adjustment, if adjustment is improper, can also generate damage to electrical equipment,
Influence the service life of electrical component.For supercritical parameter unit, generator metal temperature, transformer unit system, hydrogen gas system,
The work such as chilled water system, sealing oil system is determined under conditions of close to the operating condition limit is allowed, this safety in operation problem is just more
It is prominent and important.It according to existing technology, is mainly protected with automated system, operator supervises disk and adjusted in time to realize machine
Group stable operation, but in actual production process, due to limit by a variety of conditions, operator be difficult judgement reply it is current with
The corresponding method of operation of following operating condition, can not also continue long-term real-time monitoring system, thus can not be to these method of operation parameters
Carry out real time on-line monitoring, often not up to alarm limits due to certain local parameters deviate and operate normally threshold value, and be not easy by
It was found that causing a variety of security risks, safety accident is in turn resulted in.
The intelligent algorithms such as machine learning at present have gradually been applied in fields such as intelligent measurements, in intelligent Power Station dependent module
Also it is commonly used, is mainly used for the aid decision of management aspect, be also not directed to live electrical system operation control plane.And this hair
It is bright, the neural network analysis technology based on Keras is carried out to the electrical system running mode data of acquisition, is greatly improved
The value of power plant electrical operation data does not find that the one kind proposed in the present invention is based on also through document Investigations such as patents at present
The electrical system operating status intelligent analysis method of Keras precedent used in the analysis of power station electrical operation condition intelligent.
Summary of the invention
The purpose of the present invention is to provide a kind of electrical system operating status intelligent analysis method based on Keras, to
Solve the problems, such as that above-mentioned background technique proposes.
A kind of electrical system operating status intelligent analysis method based on Keras, which comprises the following steps:
Power station electrical system operation history data is acquired, and is stored in OpenTSDB database;
The historical data being stored in OpenTSDB database is obtained, the history vectors data of electrical system operating status are constructed
Library;
Determine the sample number batch_size of each gradient updating when training, from history vectors database with each gradient updating
Sample number batch_size be grouped for radix, and MLP parameter initialization is carried out to every group of sample number batch_size;
The data of history vectors database are loaded, and are classified as including training set and test set, a part of conduct of total number of samples
Training set is used for model training, and remainder is used for model measurement as test set;
Multilayer neural network structure is constructed using Keras, constructs electrical system operating status intelligent measurement model;
To the training set data in history vectors database as input training data, the incoming electrical system operation constructed
Condition intelligent detection model exports the predicted value and the corresponding threshold value of predicted value of electrical system operating status;
The measured value for obtaining electrical system operating status compares measured value threshold value corresponding with predicted value, predicted value, carries out pre-
Survey alarm.
Preferably, the power station electrical system operation history data includes the metal structure of generator, transformer unit system, hydrogen
Gas system determines the thermal parameter and/or electrical parameter data of chilled water system and the multiple positions of sealing oil system under different operating conditions.
Preferably, the MLP parameter initialization specifically: according to power station electrical system operation history data in different works
Classification number under condition assigns the initial value of the initial value of classification number and the frequency of training of every group of sample number batch_size.
Preferably, the training set and test set are to carry out layering according to the classification number of the history vectors database to adopt
Sample specifically sets the ratio of training set and test set as 0.8:0.2.
Preferably, described to construct multilayer neural network structure using Keras, it further includes steps of
(1) sequence model of Keras is set;
(2) it using the sample size of training set and the 2D tensor that forms of input dimension as input data, builds neural network and connects entirely
Connect layer;
(3) neural network active coating is built, neural network active coating adds activation primitive to the output of the full articulamentum of neural network;
(4) Dropout layers of neural network are built, and adds Dropout for input data, Dropout will be in model training process
In the neuron connection of certain percentage input data is disconnected when updating input data every time at random;
(5) multilayer neural network structure is compiled and configures, and constructs electrical system operating status intelligent measurement model.
Preferably, it can continue successively to add neural network after the step (4) in the multilayer neural network structure and connect entirely
Layer, neural network active coating and Dropout layers of neural network.
Preferably, relatively corresponding with predicted value, the predicted value threshold value of measured value specifically: if actual value without departing from
The corresponding threshold value of predicted value, then electrical system operating status intelligent measurement model prompts electrical system normal operation;If there is reality
Actual value exceeds threshold value, then electrical system operating status intelligent measurement model prompts system operation to deviate nominal situation, then through people
Work judgement, however, it is determined that be non-faulting, the corresponding electrical system operating status feature of the measured value is saved into history vectors data
Label is added, and electrical system operating status intelligent measurement model is trained again in library.
Due to taking above-mentioned technical solution, the beneficial effects of the present invention are: power station electrical system majority needs peace throughout the year
The timely adjustment of full stable operation, operating status is also extremely important, short-circuit especially between power station generator lock, caused by endanger very
Greatly;Pass through generator metal temperature, transformer unit system, hydrogen gas system, the top chilled water system, Seal Oil system to power station electrical system
Historical data of uniting modeling, establishes operating condition model of each electrical operation System History vector under different operating conditions, passes through communication
Interface obtains the newest the points of measurement evidence of each system, by having constructed in each system of model prediction parameter under different operating conditions
Predicted value and threshold value, when measured value exceeds model nominal situation history threshold value, progress early warning prompts operating condition to deviate, and guidance is timely
Restore, avoids causing safety problem;And have disclosure of the invention implementation, it can effectively to power station, electrically each system carries out state
Deviate early warning, and by the utilization of the intelligent Conditions Matching technology based on Keras, more to the prediction of electrical each system difference operating condition
Add precisely, the prediction of smaller deviation in the early time can be shifted to an earlier date, improves the safety and stability of power station electrical system operation.
Detailed description of the invention
Fig. 1 is the flow diagram of the electrical system operating status intelligent analysis method of the invention based on Keras;
Fig. 2 is the flow diagram that Keras of the invention constructs multilayer neural network structure.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
A kind of electrical system operating status intelligent analysis method based on Keras shown in Fig. 1 flow diagram is please referred to,
The following steps are included:
In step s101, power station electrical system operation history data is acquired, and is stored in OpenTSDB database.
It is to be understood that step S101 is implemented are as follows: by communication interface, obtain power station electrical system history run number
According to power station electrical system operation history data includes the metal structure of generator, transformer unit system, hydrogen gas system, determines cold water system
System and thermal parameter and/or electrical parameter data of the multiple positions of sealing oil system under different operating conditions, wherein the metal knot of generator
Structure, hydrogen gas system, determines chilled water system and the multiple positions of sealing oil system at transformer unit system, includes but are not limited to generator, send
Blower, electricity generator stator coil, leakproof fuel cell, generator unit stator, hydrogen cooler, stator core and stator coil, in difference
Thermal parameter under operating condition is specially temperature parameter, and electrical parameter includes power, electric current, voltage and frequency, and the thermal parameter and electricity are joined
Number data be stored in OpenTSDB database, in embodiments of the present invention, for ensure data can accuracy and referential, at least
The electrical system operation history data in multiple and different seasons under each operating condition should be acquired.
In step s 102, the historical data being stored in OpenTSDB database is obtained, building electrical system runs shape
The history vectors database of state.
It is to be understood that step S102 is implemented are as follows: in an embodiment of the present invention, acquisition is returned from on-site transfer
The thermal parameter and/or electricity being stored under the multiple position difference operating conditions of power station electrical system operating status in OpenTSDB database
Parameters history data use to cooperate the study of Keras neural network analysis technology and comparing analysis, by data assembled arrangement shape
At vector matrix, it is built into the history vectors database of electrical system operating status, with for subsequent modeling use.
In step s 103, the sample number batch_ of each gradient updating when training is determined from history vectors database
Size is grouped using the sample number batch_size of each gradient updating as radix, and to every group of sample number batch_size
Carry out MLP parameter initialization.
It is to be understood that step S103 is implemented are as follows: determine that every subgradient is more when training from history vectors database
New sample number batch_size, assign in practical applications batch_size==128, small quantities of gradient decline, this method
Data in vector data library are divided into several batches, specifically using the sample number batch_size of each gradient updating as radix
It is grouped and carrys out undated parameter, as soon as in this way, group data in one batch have codetermined the direction of this subgradient, decline is got up
It is not easy sideslip, reduces randomness;Moreover, it is much smaller compared to entire data set capacity per a batch of sample number, for mould
The decline of type calculation amount, promotes operational efficiency.
Wherein, in one embodiment of the invention, MLP parameter initialization specifically: run according to power station electrical system
The classification number under different operating conditions of historical data assigns the initial value and every group of sample number batch_size of classification number
The initial value of frequency of training.It is determined here according to operation history data and assigns classification number initial value nb_classes=100, every group
The frequency of training of sample number batch_size refers to neural network model frequency of training, is denoted as nb_epoch, assigns here just
Initial value nb_epoch=20, can according to building model training after loss function and accuracy come re-start adjustment.
In step S104, the data of history vectors database are loaded, and are classified as including training set and test set, gross sample
A part of this number is used for model training as training set, and remainder is used for model measurement as test set.
It is to be understood that step S104 is implemented are as follows: when the data of load history vectors database, OpenTSDB history
A part in database is for training, and for a part for testing the accuracy of training pattern, training set and test set are bases
The classification number of the history vectors database carries out stratified sampling, specifically sets the ratio of training set and test set as 0.8:0.2,
By (X_train, y_train), (X_test, y_test)=gwcy.load_data () obtains the data of database
Classification realizes that wherein X_train, y_train are training data, and X_test, y_test are test set data, above four variables
It is tensor.
In step s105, multilayer neural network structure is constructed using Keras, constructs electrical system operating status intelligence
Detection model.
Wherein, in one embodiment of the invention, as shown in Fig. 2, constructing multilayer neural network structure using Keras,
It further includes steps of
(1) sequence model of Keras is set, sequence model is the linear stacking of multiple network layers.
(2) using the sample size of training set and the 2D tensor that forms of input dimension as input data, neural network is built
Full articulamentum.It is to be understood that sample size nb_samples and input dimension input_dim composition (nb_samples,
Input_dim input data of the 2D tensor as the full articulamentum of neural network) is exported shaped like (nb_samples, output_
Dim 2D tensor), wherein output_dim indicates the input dimension of output data.
(3) neural network active coating is built, neural network active coating adds activation to the output of the full articulamentum of neural network
Function.
(4) Dropout layers of neural network are built, and adds Dropout for input data, Dropout will be in model training
The neuron connection for disconnecting certain percentage input data when updating input data every time in the process at random is used in embodiment
In preventing over-fitting, Dropout takes positive value 0.2 herein, realizes to model.add (Dropout (0.2)) to input data
Random opening operation.
(5) multilayer neural network structure is compiled and configures, and constructs electrical system operating status intelligent measurement model.
It can continue successively to add the full articulamentum of neural network, neural network after step (4) in multilayer neural network structure
Active coating and Dropout layers of neural network, in embodiment, can need successively constantly to add neural network complete according to the actual situation
Articulamentum, neural network active coating and Dropout layers of neural network.
In step s 106, it is passed to the training set data in history vectors database as input training data
The electrical system operating status intelligent measurement model of building, predicted value and the predicted value for exporting electrical system operating status are corresponding
Threshold value.
In step s 107, the measured value for obtaining electrical system operating status, compares measured value and predicted value, predicted value phase
Corresponding threshold value carries out prediction alarm.
Wherein, in one embodiment of the invention, it is specific to compare measured value threshold value corresponding with predicted value, predicted value
If are as follows: actual value threshold value corresponding without departing from predicted value, the electrical system of electrical system operating status intelligent measurement model prompt
System normal operation;If there is actual value beyond threshold value, electrical system operating status intelligent measurement model prompts system operation inclined
From nominal situation, then through artificial judgment, however, it is determined that be non-faulting, the corresponding electrical system operating status feature of the measured value is protected
History vectors database is deposited into, adds label, and be trained again to electrical system operating status intelligent measurement model.
In practical applications, pass through the electrical system operating status intelligent measurement model of building, collected each electrical system
System runs real time data and is predicted, the electrical system operating status obtained the measured value of observation point and prediction under different operating conditions
Value, the data of the corresponding threshold value of predicted value, this sentences the operating status of No. 1 generator in power plant as observation point, obtains real
Measured value, predicted value, upper threshold and bottom threshold are as shown in table 1.
Table 1 models observation point and corresponding parameter value
As shown in Table 1, the A phase current of No. 1 generator, stator voltage (A-C), stator voltage (A-B), stator voltage (B-C), frequency
Rate, exciting current, excitation voltage and reactive power measured value and the upper threshold and bottom threshold phase of corresponding predicted value
Compare, the measured value of each observation point is without departing from upper threshold and bottom threshold, then electrical system operating status is intelligently examined at this time
It surveys model and prompts electrical system normal operation, model continuation is trained according to neural network model frequency of training nb_epoch.
In practical applications, continue the operating status of No. 1 generator using in power plant as observation point, in addition obtain one group of reality
Measured value, predicted value, the parameter value of upper threshold and bottom threshold are as shown in table 2.
Table 2 models observation point and corresponding parameter value
Table 2 as above it is found that the A phase current of No. 1 generator, stator voltage (A-C), stator voltage (A-B), stator voltage (B-C),
Frequency, exciting current, the measured value of excitation voltage and reactive power and corresponding predicted value upper threshold and bottom threshold
It compares, wherein the measured value 688.34MW of the generator power of observation point exceeds bottom threshold 697.44MW, No. 1 generator
The measured value 3609.1A of exciting current exceeds upper threshold 3562.87A, then electrical system operating status intelligent measurement mould at this time
Type prompt operation deviates nominal situation, then through artificial judgment, however, it is determined that and be non-faulting, belong to wrong report, then it is the measured value is corresponding
Electrical system operating status, i.e. generator power and No. 1 exciter current of generator feature be saved into history vectors database,
Label is added, as mark, and according to neural network model frequency of training nb_epoch, frequency of training nb_epoch is by loss letter
Several and accuracy is again trained electrical system operating status intelligent measurement model to be adjusted.
Particular embodiments described above, pair present invention solves the technical problem that, technical scheme and beneficial effects carry out
It is further described, it should be understood that the above is only a specific embodiment of the present invention, is not limited to this
Invention, all within the spirits and principles of the present invention, any modification, equivalent substitution, improvement and etc. done should be included in this hair
Within bright protection scope.
Claims (7)
1. a kind of electrical system operating status intelligent analysis method based on Keras, which comprises the following steps:
Power station electrical system operation history data is acquired, and is stored in OpenTSDB database;
The historical data being stored in OpenTSDB database is obtained, the history vectors data of electrical system operating status are constructed
Library;
Determine the sample number batch_size of each gradient updating when training, from history vectors database with each gradient updating
Sample number batch_size be grouped for radix, and MLP parameter initialization is carried out to every group of sample number batch_size;
The data of history vectors database are loaded, and are classified as including training set and test set, a part of conduct of total number of samples
Training set is used for model training, and remainder is used for model measurement as test set;
Multilayer neural network structure is constructed using Keras, constructs electrical system operating status intelligent measurement model;
To the training set data in history vectors database as input training data, the incoming electrical system operation constructed
Condition intelligent detection model exports the predicted value and the corresponding threshold value of predicted value of electrical system operating status;
The measured value for obtaining electrical system operating status compares measured value threshold value corresponding with predicted value, predicted value, carries out pre-
Survey alarm.
2. a kind of electrical system operating status intelligent analysis method based on Keras according to claim 1, feature exist
In, the power station electrical system operation history data include the metal structure of generator, it is transformer unit system, hydrogen gas system, fixed cold
The thermal parameter and/or electrical parameter data of water system and the multiple positions of sealing oil system under different operating conditions.
3. a kind of electrical system operating status intelligent analysis method based on Keras according to claim 1, feature exist
In the MLP parameter initialization specifically: according to the classification under different operating conditions of power station electrical system operation history data
Number assigns the initial value of the initial value of classification number and the frequency of training of every group of sample number batch_size.
4. a kind of electrical system operating status intelligent analysis method based on Keras according to claim 1 or 3, special
Sign is that the training set and test set are to carry out stratified sampling according to the classification number of the history vectors database, specifically sets
The ratio for determining training set and test set is 0.8:0.2.
5. a kind of electrical system operating status intelligent analysis method based on Keras according to claim 1, feature exist
In, it is described to construct multilayer neural network structure using Keras, it further includes steps of
(1) sequence model of Keras is set;
(2) it using the sample size of training set and the 2D tensor that forms of input dimension as input data, builds neural network and connects entirely
Connect layer;
(3) neural network active coating is built, neural network active coating adds activation primitive to the output of the full articulamentum of neural network;
(4) Dropout layers of neural network are built, and adds Dropout for input data, Dropout will be in model training process
In the neuron connection of certain percentage input data is disconnected when updating input data every time at random;
(5) multilayer neural network structure is compiled and configures, and constructs electrical system operating status intelligent measurement model.
6. a kind of electrical system operating status intelligent analysis method based on Keras according to claim 5, feature exist
In can continue successively to add the full articulamentum of neural network after the step (4) in the multilayer neural network structure, neural network swashs
Layer living and Dropout layers of neural network.
7. a kind of electrical system operating status intelligent analysis method based on Keras according to claim 1, feature exist
In relatively measured value threshold value corresponding with predicted value, predicted value specifically: if actual value is corresponding without departing from predicted value
Threshold value, then electrical system operating status intelligent measurement model prompt electrical system normal operation;If there is actual value to exceed threshold value,
Then electrical system operating status intelligent measurement model prompts system operation to deviate nominal situation, then through artificial judgment, however, it is determined that
For non-faulting, the corresponding electrical system operating status feature of the measured value is saved into history vectors database, adds label, and
Again electrical system operating status intelligent measurement model is trained.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810971417.0A CN109344976A (en) | 2018-08-24 | 2018-08-24 | A kind of electrical system operating status intelligent analysis method based on Keras |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810971417.0A CN109344976A (en) | 2018-08-24 | 2018-08-24 | A kind of electrical system operating status intelligent analysis method based on Keras |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109344976A true CN109344976A (en) | 2019-02-15 |
Family
ID=65291532
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810971417.0A Pending CN109344976A (en) | 2018-08-24 | 2018-08-24 | A kind of electrical system operating status intelligent analysis method based on Keras |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109344976A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112381376A (en) * | 2020-11-10 | 2021-02-19 | 易显智能科技有限责任公司 | Method and device for evaluating driving ability process |
CN115169709A (en) * | 2022-07-18 | 2022-10-11 | 华能汕头海门发电有限责任公司 | Power station auxiliary machine fault diagnosis method and system based on data driving |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104102773A (en) * | 2014-07-05 | 2014-10-15 | 山东鲁能软件技术有限公司 | Equipment fault warning and state monitoring method |
CN104573850A (en) * | 2014-12-18 | 2015-04-29 | 湖南大唐先一科技有限公司 | Method for evaluating state of thermal power plant equipment |
CN105117591A (en) * | 2015-08-15 | 2015-12-02 | 国家电网公司 | Power device running status analysis method |
CN105137354A (en) * | 2015-08-27 | 2015-12-09 | 江苏科技大学 | Motor fault detection method based on nerve network |
US20160048757A1 (en) * | 2006-03-10 | 2016-02-18 | Power Analytics Corporation | Systems and Methods for Real-Time Forecasting and Predicting of Electrical Peaks and Managing the Energy, Health, Reliability, and Performance of Electrical Power Systems Based on an Artificial Adaptive Neural Network |
CN106682781A (en) * | 2016-12-30 | 2017-05-17 | 山东鲁能软件技术有限公司 | Power equipment multi-index prediction method |
CN107203809A (en) * | 2017-04-20 | 2017-09-26 | 华中科技大学 | A kind of deep learning automation parameter adjustment method and system based on Keras |
-
2018
- 2018-08-24 CN CN201810971417.0A patent/CN109344976A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160048757A1 (en) * | 2006-03-10 | 2016-02-18 | Power Analytics Corporation | Systems and Methods for Real-Time Forecasting and Predicting of Electrical Peaks and Managing the Energy, Health, Reliability, and Performance of Electrical Power Systems Based on an Artificial Adaptive Neural Network |
CN104102773A (en) * | 2014-07-05 | 2014-10-15 | 山东鲁能软件技术有限公司 | Equipment fault warning and state monitoring method |
CN104573850A (en) * | 2014-12-18 | 2015-04-29 | 湖南大唐先一科技有限公司 | Method for evaluating state of thermal power plant equipment |
CN105117591A (en) * | 2015-08-15 | 2015-12-02 | 国家电网公司 | Power device running status analysis method |
CN105137354A (en) * | 2015-08-27 | 2015-12-09 | 江苏科技大学 | Motor fault detection method based on nerve network |
CN106682781A (en) * | 2016-12-30 | 2017-05-17 | 山东鲁能软件技术有限公司 | Power equipment multi-index prediction method |
CN107203809A (en) * | 2017-04-20 | 2017-09-26 | 华中科技大学 | A kind of deep learning automation parameter adjustment method and system based on Keras |
Non-Patent Citations (2)
Title |
---|
FIRELING: "深度学习之Keras入门", 《HTTP://WWW.LINING0806.COM/%E6%B7%B1%E5%BA%A6%E5%AD%A6%E4%B9%A0%E4%B9%8BKERAS%E5%85%A5%E9%97%A8/》 * |
LC19861217: "深度学习:Keras入门(一)之基础篇", 《HTTPS://WWW.CNBLOGS.COM/LC1217/P/7132364.HTML》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112381376A (en) * | 2020-11-10 | 2021-02-19 | 易显智能科技有限责任公司 | Method and device for evaluating driving ability process |
CN115169709A (en) * | 2022-07-18 | 2022-10-11 | 华能汕头海门发电有限责任公司 | Power station auxiliary machine fault diagnosis method and system based on data driving |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zollanvari et al. | Transformer fault prognosis using deep recurrent neural network over vibration signals | |
CN102663412B (en) | Power equipment current-carrying fault trend prediction method based on least squares support vector machine | |
Mao et al. | Investigation of polymer electrolyte membrane fuel cell internal behaviour during long term operation and its use in prognostics | |
CN110095744A (en) | A kind of electronic mutual inductor error prediction method | |
CN110320892A (en) | The sewage disposal device fault diagnosis system and method returned based on Lasso | |
Tomin et al. | Machine learning techniques for power system security assessment | |
CN107843718A (en) | A kind of method that transformer insulation oil ageing state is assessed | |
CN111898669B (en) | Abnormal event early warning system of direct-current submerged arc furnace based on machine learning | |
CN111525587B (en) | Reactive load situation-based power grid reactive voltage control method and system | |
Gupta et al. | Neural network based early warning system for an emerging blackout in smart grid power networks | |
Li et al. | Real-time monitoring of long-term voltage stability via convolutional neural network | |
CN106997513A (en) | The distribution net equipment state evaluation system analyzed based on big data | |
CN116169675B (en) | Power system dynamic stability online evaluation method considering operation mode change | |
Jiao | Application and prospect of artificial intelligence in smart grid | |
CN109344976A (en) | A kind of electrical system operating status intelligent analysis method based on Keras | |
CN113612237A (en) | Method for positioning resonance-induced subsynchronous oscillation source in offshore wind farm | |
CN110705831A (en) | Power angle instability mode pre-judgment model construction method after power system fault and application thereof | |
Halilcevic et al. | Prediction of power system security levels | |
Shi et al. | Intelligent power system stability assessment and dominant instability mode identification using integrated active deep learning | |
CN116245033A (en) | Artificial intelligent driven power system analysis method and intelligent software platform | |
CN115603446A (en) | Power distribution station area operation monitoring system based on convolution neural network and cloud edge synergistic effect | |
Won et al. | Prediction of remaining useful lifetime of membrane using machine learning | |
Negri et al. | Artificial Intelligence and Smart Grids: the Cable Joint Test Case | |
Zheng et al. | A fast stability assessment scheme based on classification and regression tree | |
Zhang et al. | Spatio-temporal fusion model of natural gas pipeline condition monitoring based on convolutional neural network and long short-term memory neural network |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190215 |