CN111259912B - Instrument image recognition method based on AE-SVM substation inspection robot - Google Patents
Instrument image recognition method based on AE-SVM substation inspection robot Download PDFInfo
- Publication number
- CN111259912B CN111259912B CN202010032659.0A CN202010032659A CN111259912B CN 111259912 B CN111259912 B CN 111259912B CN 202010032659 A CN202010032659 A CN 202010032659A CN 111259912 B CN111259912 B CN 111259912B
- Authority
- CN
- China
- Prior art keywords
- instrument
- svm
- model
- data
- inspection robot
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/02—Recognising information on displays, dials, clocks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an instrument image recognition method based on an AE-SVM substation inspection robot. Aiming at the problems of blindness and randomness caused by manually extracting features by the image recognition algorithm of the current inspection robot, the method uses an AE model to extract image features and deep features containing more substation instrument images; then using an SVM classification model to output an instrument image classification result; and finally, reading out the data in the instrument image, and storing the data in a database to realize automatic identification of the instrument image of the substation inspection robot.
Description
Technical Field
The invention relates to the field of image recognition algorithms, in particular to an instrument image recognition method based on an AE-SVM transformer substation inspection robot.
Background
With the continuous development of the electric power technology in China, the construction and transportation quantity of the power grid is larger and the reform of the electric power system is deeper, and the electric power system in China is developing towards mechanization, automation and intellectualization. The power system is vigorously developed, and the daily monitoring, inspection, maintenance and defect elimination of the transformer substation become heavy. The traditional work such as routine and special inspection, daily maintenance and the like requires power transformation operators or team members to go to each transformer substation and converter station periodically to carry out inspection work, a large amount of data is collected in a manual collection mode, and a plurality of complicated periodic, repetitive and mechanical works are required to be carried out, so that a large amount of manpower and material resources are consumed.
The inspection work is the important part of the daily operation and maintenance work of the transformer substation, the national grid company has very strict requirements on the inspection work of the transformer substation, and the power transformation operation and maintenance profession of each grid province company also puts forward a series of actions to strengthen the power transformation inspection work, but the situations of incomplete inspection, inaccurate data, inspection flow in form and the like caused by insufficient personnel quantity configuration and the like still exist.
The intelligent inspection robot for the transformer substation is an inspection robot provided with advanced equipment such as an infrared thermometer, a high-definition camera, an audio signal receiver and the like. The intelligent inspection robot for the transformer substation can comprehensively inspect all-station equipment according to inspection tasks set in advance in time intervals and areas, various hidden danger problems can be found in time, all task modules of meter reading, temperature measurement, sound collection and functional inspection are processed together, and abnormality, defect and hidden danger of the found equipment can be identified and alarmed in time. In addition, the background of the robot intuitively forms a chart and a week and month report of each operation data index of the equipment stored in the inspection, so that operation and maintenance personnel can better carry out maintenance, fault analysis and the like on the equipment. The current inspection robot uses a traditional image recognition algorithm, the traditional image recognition algorithm uses manually extracted image features, a model is independently built for each meter, character cutting and a large number of manual modeling operations are required, and the environment change influence is large.
Disclosure of Invention
In order to solve the above-mentioned problems. The invention provides an instrument image recognition method based on an AE-SVM substation inspection robot, which solves the problem of instrument image recognition of the substation inspection robot. To achieve this object:
the invention provides an instrument image recognition method based on an AE-SVM transformer substation inspection robot, which comprises the following specific steps:
step 1, an inspection robot collects instrument images of tested equipment in each time period by using a CCD image sensor, and uploads the instrument images to an instrument data identification system of a base station system through WIFI;
step 2: the instrument data identification system extracts image features of instrument data through a pre-trained AE model;
step 3: inputting the extracted image features of the instrument data into a trained SVM model, classifying the instrument data, reading out digital instrument data according to the equipment number, and storing the digital instrument data into a database;
step 4: for the situation of error division of instrument data, the system can send the data into a classification model to perform incremental learning on the SVM model, so that the accuracy of model classification is continuously improved.
As a further improvement of the present invention, the inspection robot in the step 1 is as follows:
the control system of the substation inspection robot mainly comprises: a base station system and a patrol robot system.
The base station system mainly comprises a background engineering machine, an instrument data identification system, wireless communication equipment, a base station navigation positioning planning system and the like. The background engineering machine mainly provides an operation interaction interface for operators, monitors various state information of the robot and plans a movement route of the robot. The instrument data recognition system mainly stores instrument images acquired by the CCD sensor and completes the functions of recognizing the instrument images, saving data and the like.
The inspection robot system is mainly divided into a mobile control system and an instrument detection system. The base station system mainly comprises a background engineering machine, an instrument data identification system, wireless communication equipment, a base station navigation positioning planning system and the like. The mobile control system mainly collects navigation positioning information of the inspection robot through a sensor, and meanwhile feeds back the navigation positioning information to the background engineering machine to control a movement route of the inspection robot, and the movement of the robot is controlled according to a control command of the background engineering machine, and state information of the inspection robot and the movement state of the inspection robot are uploaded. The instrument detection system mainly transmits instrument image data acquired by the sensor to the base station system.
As a further improvement of the present invention, the AE model in the step 2 is as follows:
the AE model network structure is set to be 256-80-256, the hidden layer weight penalty coefficient lambda is 0.003, the input layer activation function is a log sig function, the output layer activation function is a purelin function, the maximum iteration number is set to be 300, the learning rate is set to be 0.05, and the error function adopts root mean square error.
As a further improvement of the invention, the method for extracting the image features of the instrument from the coding model in the step 2 is as follows:
the instrument data identification system inputs an original image shot by the inspection robot into a trained AE model, and extracts image features h at an hidden layer d The method comprises the steps of carrying out a first treatment on the surface of the For data set containing M image samplesAny sample x of (2) d Coding vector h of hidden layer d Reconstruction output from coding network>Can be expressed as:
h d =f(W (1) x d +b (1) ) (1)
in which W is (1) 、b (1) Weight matrix and bias term of coding network, W (2) 、b (2) Respectively a weight matrix and a bias term of the decoding network; input x d And reconstruct the outputThe reconstruction error between, i.e. the cost function of a single sample, can be expressed as:
the overall cost function for an AE for a dataset containing M samples can be expressed as
In equation 4, the first term is the mean square error of the data, which is used to measure the approximation degree of the original input and the reconstruction input; the second term is the weight decay term: l2 regularization to suppress overfitting; lambda is the weight penalty coefficient, n l For the total layer number of AE network, S l For the number of neurons in the first layer of the self-encoding network,the connection weight between the ith neuron of the first layer and the jth neuron of the first layer and the (1) th layer of the self-coding network;
the cost function of self-coding is continuously reduced by a gradient descent method, the weights and the bias of a coding network and a decoding network are updated, and AE tries to learn a constant function, so that the output obtained by reconstructing the coding vector of an hidden layer is infinitely approximate to the original input, and the coding vector of the hidden layer is the image characteristic extracted by the original input.
As a further improvement of the present invention, the SVM classification model algorithm in the step 3 is as follows:
let the linear separable sample image feature set be (x i ,y i ) Wherein i=1, 2, n; x epsilon R d Y epsilon { -1,1} is the class label, and the optimal hyperplane for classification obtained by learning the "interval maximization" learning strategy is:
ω·x+b=0 (5)
where ω is the normal vector, determining the direction of the hyperplane. b is the displacement, and the distance between the hyperplane and the origin is determined. The general form of the corresponding linear classification function is:
f(x)=ω·x+b (6)
simultaneously, the optimal hyperplane enables all sample points to meet the following conditions:
|f(x)|≥1 (7)
the support vector of the SVM is a sample point for establishing the expression 7, and after the support vector is found, the data can be divided into two types; the SVM model employs a Gaussian radial basis function, wherein the Gaussian radial basis function is as follows:
in classification, the SVM algorithm nonlinearly maps sample data from a low-dimensional feature space to a high-dimensional feature space through a kernel function, and finds out an optimal hyperplane with maximized interval in the high-dimensional feature space, wherein the nonlinear classification function is as follows:
the SVM algorithm is a two-class model, a plurality of two-class models are constructed at the same time, the multi-class problem can be solved, and when n classes exist, n (n-1)/2 SVM models need to be designed.
As a further improvement of the present invention, the incremental learning of the SVM model in the step 4 is as follows:
and making the misclassified data set as a training set, correctly modifying the label of the training set, performing incremental learning on the training set on the basis of the trained SVM model, updating the threshold value and the weight of the model, finally realizing optimization upgrading of the SVM model, and increasing the robustness, generalization capability and recognition rate of the model.
The instrument image recognition method based on the AE-SVM substation inspection robot has the beneficial effects that: the invention has the technical effects that:
1. according to the invention, by combining an AE and an SVM algorithm, the instrument image features extracted through the AE model can contain more instrument image information, so that blindness and randomness of manually extracting the features are avoided;
2. the SVM is applied to instrument image classification of the transformer substation, so that the accuracy and the efficiency of image classification are improved;
3. according to the invention, the SVM model is optimized and upgraded by utilizing the error-divided image data, so that the incremental learning of the SVM model is realized, and the robustness and generalization capability of the model are enhanced;
4. the AE-SVM model provided by the invention provides an automatic instrument image recognition method for the transformer substation instrument image.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a system configuration diagram of the inspection robot;
FIG. 3 is a view of an AE network configuration;
FIG. 4 is a support vector machine classification optimal hyperplane.
Detailed Description
The invention is described in further detail below with reference to the attached drawings and detailed description:
the invention provides an instrument image recognition method based on an AE-SVM substation inspection robot, which aims to realize efficient classification of substation instrument images and on-line optimization and upgrading of a classification model. FIG. 1 is a flow chart of the present invention, and the steps of the present invention are described in detail below in conjunction with the flow chart.
Step 1, an inspection robot collects instrument images of tested equipment in each time period by using a CCD image sensor, and uploads the instrument images to an instrument data identification system of a base station system through WIFI;
the inspection robot control system in step 1 is specifically described as follows:
the structural diagram of the substation inspection robot control system is shown in fig. 2, and the substation inspection robot control system is mainly divided into: a base station system and a patrol robot system; the base station system mainly comprises a background engineering machine, an instrument data identification system, wireless communication equipment, a base station navigation positioning planning system and the like; the background engineering machine mainly provides an operation interaction interface for operators, monitors various state information of the robot and plans a movement route of the robot; the instrument data recognition system mainly stores instrument images acquired by the CCD sensor and completes the functions of recognizing the instrument images, saving data and the like; the inspection robot system mainly comprises a mobile control system and an instrument detection system; the base station system mainly comprises a background engineering machine, an instrument data identification system, wireless communication equipment, a base station navigation positioning planning system and the like; the mobile control system mainly collects navigation positioning information of the inspection robot through a sensor, and simultaneously feeds back the navigation positioning information to the background engineering machine to control a movement route of the inspection robot, and controls movement of the robot according to a control command of the background engineering machine, and uploads state information of the inspection robot and a movement state of the inspection robot; the instrument detection system mainly transmits instrument image data acquired by the sensor to the base station system.
Step 2: the instrument data recognition system extracts the image characteristics of instrument data through a pre-trained AE model;
the AE model and image feature extraction in step 2 is specifically described as follows:
the AE network structure diagram is shown in FIG. 3, the self-coding network structure is set to be a 256-80-256 network structure, the hidden layer weight penalty coefficient lambda is 0.003, the input layer activation function is a log sig function, the output layer activation function is a purelin function, the maximum iteration number is set to be 300, the learning rate is set to be 0.05, and the error function adopts root mean square error.
The instrument data identification system inputs an original image shot by the inspection robot into a trained self-coding network, and extracts image features h at an hidden layer d The method comprises the steps of carrying out a first treatment on the surface of the For data set containing M image samplesAny sample x of (2) d Coding vector h of hidden layer d Reconstruction output from coding network>Can be expressed as:
h d =f(W (1) x d +b (1) ) (1)
in which W is (1) 、b (1) Weight matrix and bias term of coding network, W (2) 、b ( 2 ) Respectively a weight matrix and a bias term of the decoding network; input x d And reconstruct the outputThe reconstruction error between, i.e. the cost function of a single sample, can be expressed as:
the overall cost function for an AE for a dataset containing M samples can be expressed as
In equation 4, the first term is the mean square error of the data, which is used to measure the approximation degree of the original input and the reconstruction input; the second term is the weight decay term: l2 regularization to suppress overfitting; lambda is the weight penalty coefficient, n l For the total layer number of AE network, S l For the number of neurons in the first layer of the self-encoding network,the connection weight between the ith neuron of the first layer and the jth neuron of the first layer and the (1) th layer of the self-coding network;
the cost function of self-coding is continuously reduced by a gradient descent method, the weights and the bias of a coding network and a decoding network are updated, and AE tries to learn a constant function, so that the output obtained by reconstructing the coding vector of an hidden layer is infinitely approximate to the original input, and the coding vector of the hidden layer is the image characteristic extracted by the original input.
Step 3: inputting the extracted image features of the instrument data into a trained SVM model, classifying the instrument data, reading out digital instrument data according to the equipment number, and storing the digital instrument data into a database;
the SVM classification model in step 3 is specifically described as follows:
let the linear separable sample image feature set be (x i ,y i ) Wherein i=1, 2, n; x epsilon R d Y epsilon { -1,1} is the class label, and the optimal hyperplane for classification obtained by learning the "interval maximization" learning strategy is:
ω·x+b=0 (5)
where ω is the normal vector, determining the direction of the hyperplane. b is the displacement, the distance between the hyperplane and the origin is determined, and the hyperplane is classified as shown in fig. 4. The general form of the corresponding linear classification function is:
f(x)=ω·x+b (6)
simultaneously, the optimal hyperplane enables all sample points to meet the following conditions:
|f(x)|≥1 (7)
the support vector of the SVM is a sample point for establishing the expression 7, and after the support vector is found, the data can be divided into two types; the SVM model employs a Gaussian radial basis function, wherein the Gaussian radial basis function is as follows:
in classification, the SVM algorithm nonlinearly maps sample data from a low-dimensional feature space to a high-dimensional feature space through a kernel function, and finds out an optimal hyperplane with maximized interval in the high-dimensional feature space, wherein the nonlinear classification function is as follows:
the SVM algorithm is a two-class model, a plurality of two-class models are constructed at the same time, the multi-class problem can be solved, and when n classes exist, n (n-1)/2 SVM models need to be designed.
Step 4: for the situation of error division of instrument data, the system can send the data into a classification model to perform incremental learning on the SVM model, so that the accuracy of model classification is continuously improved.
Incremental learning of the SVM model in step 4 is described as follows:
and making the misclassified data set as a training set, correctly modifying the label of the training set, performing incremental learning on the training set on the basis of the trained SVM model, updating the threshold value and the weight of the model, finally realizing optimization upgrading of the SVM model, and increasing the robustness, generalization capability and recognition rate of the model.
The above description is only of the preferred embodiment of the present invention, and is not intended to limit the present invention in any other way, but is intended to cover any modifications or equivalent variations according to the technical spirit of the present invention, which fall within the scope of the present invention as defined by the appended claims.
Claims (1)
1. The instrument image recognition method based on the AE-SVM substation inspection robot comprises the following specific steps of:
the control system of the substation inspection robot is divided into: a base station system and a patrol robot system; the base station system consists of a background engineering machine, an instrument data identification system, wireless communication equipment and a base station navigation positioning planning system; the background engineering machine provides an operation interaction interface for operators, monitors various state information of the robot and plans a movement route of the robot; the instrument data recognition system stores instrument images acquired by the CCD sensor and completes the functions of recognizing the instrument images and saving data; the inspection robot system is divided into a mobile control system and an instrument detection system; the base station system consists of a background engineering machine, an instrument data identification system, wireless communication equipment and a base station navigation positioning planning system; the mobile control system collects navigation positioning information of the inspection robot through a sensor, and feeds back the navigation positioning information to the background engineering machine to control a movement route of the inspection robot, and the movement of the robot is controlled according to a control command of the background engineering machine, and state information of the inspection robot and the movement state of the inspection robot are uploaded; the instrument detection system transmits instrument image data acquired by the sensor to the base station system;
step 1, an inspection robot collects instrument images of tested equipment in each time period by using a CCD image sensor, and uploads the instrument images to an instrument data identification system of a base station system through WIFI;
step 2: the instrument data identification system extracts image features of instrument data through a pre-trained AE model;
step 3: inputting the extracted image features of the instrument data into a trained self-coding SVM model, classifying the instrument data, reading out digital instrument data according to the equipment number, and storing the digital instrument data into a database;
the self-coding SVM model comprises the following steps:
the AE network structure is set as a 256-80-256 network structure, the implicit layer weight penalty coefficient is 0.003, the input layer activation function is a log sig function, the output layer activation function is a purelin function, the maximum iteration number is set as 300, the learning rate is set as 0.05, and the error function adopts root mean square error;
the AE network structure model extracts the image characteristics of the instrument:
the instrument data identification system inputs an original image shot by the inspection robot into a trained self-coding network, and extracts image features h at an hidden layer d The method comprises the steps of carrying out a first treatment on the surface of the For data set containing M image samplesAny sample x of (2) d Coding vector h of hidden layer d Reconstruction output from coding network>Can be expressed as:
h d =f(W (1) x d +b (1) ) (1)
in which W is (1) 、b (1) Weight matrix and bias term of coding network, W (2) 、b (2) Respectively a weight matrix and a bias term of the decoding network; input x d And reconstruct the outputThe reconstruction error between, i.e. the cost function of a single sample, can be expressed as:
the overall cost function for an AE for a dataset containing M samples can be expressed as
In equation 4, the first term is the mean square error of the data, which is used to measure the approximation degree of the original input and the reconstruction input; the second term is the weight decay term: l2 regularization to suppress overfitting; lambda is the weight penalty coefficient, n l For the total layer number of AE network, S l For the number of neurons in the first layer of the self-encoding network,the connection weight between the ith neuron of the first layer and the jth neuron of the first layer and the (1) th layer of the self-coding network; />
Continuously reducing a self-coding cost function by a gradient descent method, updating weights and offsets of a coding network and a decoding network, and trying to learn an identical function by AE so that the output obtained by reconstructing the coding vector of an hidden layer is infinitely approximate to the original input, wherein the coding vector of the hidden layer is the image characteristic extracted by the original input;
step 4: for the situation of error division of instrument data, the system can send the data into a classification model to perform incremental learning on the SVM classification model, so that the accuracy of model classification is continuously improved;
the SVM classification model comprises:
let the linear separable sample image feature set be (x i ,y i ) Wherein i=1, 2, n; x epsilon R d Y epsilon { -1,1} is the class label, and the optimal hyperplane for classification obtained by learning the "interval maximization" learning strategy is:
ω·x+b=0 (5)
wherein ω is a normal vector, determines the direction of the hyperplane, b is a displacement, determines the distance between the hyperplane and the origin, and the general form of the corresponding linear classification function is:
f(x)=ω·x+b (6)
simultaneously, the optimal hyperplane enables all sample points to meet the following conditions:
the support vector of the I f (x) I1 (7) SVM is a sample point which enables the expression 7 to be established, and after the support vector is found, data can be divided into two types; the SVM model employs a Gaussian radial basis function, wherein the Gaussian radial basis function is as follows:
in classification, the SVM algorithm nonlinearly maps sample data from a low-dimensional feature space to a high-dimensional feature space through a kernel function, and finds out an optimal hyperplane with maximized interval in the high-dimensional feature space, wherein the nonlinear classification function is as follows:
the SVM algorithm is a two-class model, a plurality of two-class models are constructed at the same time, so that the problem of multi-class can be solved, and when n classes exist, n (n-1)/2 SVM models need to be designed;
incremental learning of the SVM model:
and making the misclassified data set as a training set, correctly modifying the label of the training set, performing incremental learning on the training set on the basis of the trained SVM model, updating the threshold value and the weight of the model, finally realizing optimization upgrading of the SVM model, and increasing the robustness, generalization capability and recognition rate of the model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010032659.0A CN111259912B (en) | 2020-01-13 | 2020-01-13 | Instrument image recognition method based on AE-SVM substation inspection robot |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010032659.0A CN111259912B (en) | 2020-01-13 | 2020-01-13 | Instrument image recognition method based on AE-SVM substation inspection robot |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111259912A CN111259912A (en) | 2020-06-09 |
CN111259912B true CN111259912B (en) | 2023-05-12 |
Family
ID=70954095
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010032659.0A Active CN111259912B (en) | 2020-01-13 | 2020-01-13 | Instrument image recognition method based on AE-SVM substation inspection robot |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111259912B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112834959A (en) * | 2021-01-04 | 2021-05-25 | 福州大学 | Direct-current power distribution system fault detection method based on high-frequency feature extraction |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101531530B1 (en) * | 2014-12-31 | 2015-06-25 | (주)스타넥스 | Image analysis method, apparatus and computer readable medium |
CN105184309B (en) * | 2015-08-12 | 2018-11-16 | 西安电子科技大学 | Classification of Polarimetric SAR Image based on CNN and SVM |
CN105913090B (en) * | 2016-04-14 | 2019-03-26 | 西安电子科技大学 | SAR image objective classification method based on SDAE-SVM |
CN105930837A (en) * | 2016-05-17 | 2016-09-07 | 杭州申昊科技股份有限公司 | Transformer station instrument equipment image recognition method based on autonomous routing inspection robot |
-
2020
- 2020-01-13 CN CN202010032659.0A patent/CN111259912B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN111259912A (en) | 2020-06-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107817404B (en) | Portable metering automation terminal fault diagnosis device and diagnosis method thereof | |
CN112990558B (en) | Meteorological temperature and illumination prediction method based on deep migration learning | |
CN111008726B (en) | Class picture conversion method in power load prediction | |
CN110555551A (en) | air quality big data management method and system for smart city | |
CN117540908B (en) | Agricultural resource integration method and system based on big data | |
CN116127065A (en) | Simple and easy-to-use incremental learning text classification method and system | |
CN111259912B (en) | Instrument image recognition method based on AE-SVM substation inspection robot | |
CN113947013A (en) | Boiler short-term NO based on hybrid deep neural network modelingxEmission prediction method | |
CN117391257A (en) | Road congestion condition prediction method and device | |
CN111985782A (en) | Automatic tramcar driving risk assessment method based on environment perception | |
CN115130599A (en) | Semi-supervision method for strip mine card state recognition under time series GAN data enhancement | |
CN115293249A (en) | Power system typical scene probability prediction method based on dynamic time sequence prediction | |
CN116363498A (en) | Loess plateau terrace automatic identification method based on deep migration learning | |
CN113537405A (en) | Blue algae image classification and identification tool adopting artificial intelligence technology | |
CN112434750B (en) | Dam monitoring data development mode identification method based on convolutional neural network | |
CN110837932A (en) | Thermal power prediction method of solar heat collection system based on DBN-GA model | |
CN118070134B (en) | Method and system for predicting electricity consumption of users in park | |
Alshawabkeh et al. | Automated Pavement Crack Detection Using Deep Feature Selection and Whale Optimization Algorithm. | |
CN117473173B (en) | Patent achievement transformation management system and method based on big data | |
CN118012977B (en) | AI and GIS fusion-based two-dimensional multi-mode data processing method | |
CN113780439B (en) | Multi-cloud identification system of different types of meteorological satellites based on unsupervised domain adaptation | |
CN117932347B (en) | PM2.5 prediction method and system based on resistance transfer learning | |
Wang et al. | Research on Daily Tourist Flow Prediction of Scenic Spots Based on Similar Day Clustering and LSSVM Model | |
Hofmann et al. | Use of machine-learning for monitoring solar thermal plants | |
CN118213987A (en) | Short-term photovoltaic power generation power prediction method, system, equipment and medium based on similar weather discovery |
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 | ||
TR01 | Transfer of patent right |
Effective date of registration: 20230717 Address after: No. 1 Dongji Avenue, Jiangning District, Nanjing, Jiangsu Province, 210000 (Jiangning Development Zone) Patentee after: NANJING YOUKUO ELECTRICAL TECHNOLOGY CO.,LTD. Address before: No. 99 Jiangning Road, Nanjing District hirokage 210000 cities in Jiangsu Province Patentee before: JINLING INSTITUTE OF TECHNOLOGY |
|
TR01 | Transfer of patent right |