CN113671421A - Transformer state evaluation and fault early warning method - Google Patents
Transformer state evaluation and fault early warning method Download PDFInfo
- Publication number
- CN113671421A CN113671421A CN202110976585.0A CN202110976585A CN113671421A CN 113671421 A CN113671421 A CN 113671421A CN 202110976585 A CN202110976585 A CN 202110976585A CN 113671421 A CN113671421 A CN 113671421A
- Authority
- CN
- China
- Prior art keywords
- transformer
- model
- early warning
- sequence data
- fault diagnosis
- 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
- 238000011156 evaluation Methods 0.000 title claims abstract description 56
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000004458 analytical method Methods 0.000 claims abstract description 84
- 238000012544 monitoring process Methods 0.000 claims abstract description 51
- 230000008859 change Effects 0.000 claims abstract description 48
- 238000007781 pre-processing Methods 0.000 claims abstract description 18
- 238000003745 diagnosis Methods 0.000 claims description 86
- 239000003795 chemical substances by application Substances 0.000 claims description 38
- 238000013527 convolutional neural network Methods 0.000 claims description 31
- 238000012549 training Methods 0.000 claims description 27
- 238000005070 sampling Methods 0.000 claims description 16
- 238000000605 extraction Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 5
- 238000003860 storage Methods 0.000 claims description 5
- 210000002569 neuron Anatomy 0.000 claims description 4
- 238000013486 operation strategy Methods 0.000 claims description 4
- 230000033228 biological regulation Effects 0.000 claims description 3
- 238000001816 cooling Methods 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 238000009413 insulation Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 230000006641 stabilisation Effects 0.000 claims description 3
- 238000011105 stabilization Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 2
- 238000007726 management method Methods 0.000 abstract description 28
- 238000012423 maintenance Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 9
- 230000015654 memory Effects 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 6
- 230000005540 biological transmission Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 4
- 230000004913 activation Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000008447 perception Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000007787 long-term memory Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 230000006855 networking Effects 0.000 description 2
- 230000002265 prevention Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 230000006403 short-term memory Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004587 chromatography analysis Methods 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 230000008713 feedback mechanism Effects 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012067 mathematical method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/62—Testing of transformers
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/72—Testing of electric windings
Abstract
The invention provides a transformer state evaluation and fault early warning method, which comprises the following steps: the monitoring parameters are collected through the collection module and transmitted to an intelligent analysis model in the edge agent module, data preprocessing and time sequence prediction are carried out in the intelligent analysis model to obtain sequence data, the sequence data are transmitted to a comprehensive analysis model in the edge agent module, the sequence data are diagnosed and analyzed through the comprehensive analysis model to obtain a transformer state change trend and an early warning message, the monitoring parameters, the transformer state change trend and the early warning message are uploaded to a cloud management platform at regular time, the cloud management platform carries out state evaluation on the transformer according to the monitoring parameters, the transformer state change trend and the early warning message to obtain a transformer evaluation report and returns the transformer evaluation report to the edge agent module. The transformer state evaluation and fault early warning method provided by the invention improves the management method of the transformer and improves the monitoring efficiency of the transformer.
Description
Technical Field
The invention relates to the technical field of transformers, in particular to a transformer state evaluation and fault early warning method.
Background
A cloud-edge cooperative ubiquitous power internet of things is built, and strategic deployment of national power grid companies is realized. Along with the implementation of the ubiquitous power internet of things strategy by a company, the national grid equipment department sets the top layer design of the ubiquitous power internet of things on the equipment side, and develops the task construction of the power transmission and transformation internet of things, and the state perception level of the equipment is improved by applying new technologies such as edge calculation and wireless transmission according to the four-layer architecture of a perception layer, a network layer, a platform layer and an application layer. The power transformer is an important device in power transmission and transformation and power supply and distribution systems, and the operation state of the power transformer has an important influence on the safe and stable operation of a power grid. In the long-term operation of the transformer, the problems of abnormal operation state and the like are inevitable, the accurate operation state evaluation and management of the power transformer are carried out, and the smooth operation of transformer state maintenance work and the formulation of a reasonable maintenance strategy are facilitated. Conventional transformer health assessment has proposed a number of successful research methods, for example, applying mathematical methods such as bayesian network, neural network, fuzzy theory, and information fusion to the transformer state assessment research. In the new situation, in the face of the vigorous development of the ubiquitous power internet of things, under the new cloud-edge cooperative system, research on improving the state perception level of equipment and optimizing a transformer configuration management method is necessary to a certain extent.
At present, the real-time and dynamic analysis of equipment in the configuration management of national grids and various provincial equipment becomes a normalized work, the maintenance condition of the equipment of the power system is improved to a certain extent, and the problems of long equipment state evaluation period, short information analysis level, poor instantaneity and the like are exposed. Along with the current continuous construction of ubiquitous electric power thing networking, novel check out test set and sensor insert in a large number, and on-line monitoring data in the electric wire netting database is the explosive growth with the off-line test, has possessed the work basis of developing the high-efficient utilization of data and analysis, carries out conventional statistics or simply fuses data in addition and can not adapt to the high-efficient intelligent demand of ubiquitous electric power thing networking. At present, the state evaluation of each unit device of the national network only carries out unified evaluation on recent monitoring and test data, then an evaluation result is obtained, only the recent comprehensive condition of the transformer can be reflected, dynamic refined monitoring of the transformer cannot be realized, the problems of long evaluation time and unobvious key defects are difficult to solve, multidimensional targeted analysis and management of transformer information are not facilitated, and point-to-point timely understanding and maintenance of the transformer by operation and maintenance evaluators are not facilitated. Therefore, it is necessary to design a transformer state evaluation and fault early warning method based on an intelligent algorithm and a cloud-edge cooperative system.
Disclosure of Invention
The invention aims to provide a transformer state evaluation and fault early warning method, which solves the problems that the traditional evaluation algorithm cannot realize dynamic fine monitoring of a transformer, is long in evaluation time and is not beneficial to operation and maintenance evaluators to timely know and maintain the transformer point to point, improves the management method of the transformer, and improves the monitoring efficiency of the transformer.
In order to achieve the purpose, the invention provides the following scheme:
a transformer state assessment and fault early warning method is applied to a transformer state assessment and fault early warning system, the system comprises an acquisition module, an edge agent module and a cloud management platform, the acquisition module is connected with the edge agent module, the edge agent module is connected with the cloud management platform, the acquisition module comprises a plurality of sensors and is used for acquiring monitoring parameters of a transformer and transmitting the monitoring parameters to the edge agent module, the edge agent module comprises an intelligent analysis model and a comprehensive analysis model, the intelligent analysis model is used for carrying out data preprocessing and time sequence prediction on the monitoring parameters of the transformer to obtain sequence data, the comprehensive analysis module is used for analyzing and judging the sequence data to obtain a transformer state change trend and early warning information, and the cloud management platform is used for realizing real-time storage, real-time storage and early warning information of the monitoring parameters of the acquisition module, Real-time recording of early warning information, evaluation and information archiving of the state of the transformer and regulation and control of the operation strategy of the transformer;
the method comprises the following steps:
step 1: monitoring parameters are acquired through an acquisition module and transmitted to an intelligent analysis model in an edge agent module, and data preprocessing and time sequence prediction are carried out in the intelligent analysis model to obtain sequence data;
step 2: transmitting the sequence data to a comprehensive analysis model in the edge agent module, and carrying out diagnosis and analysis on the sequence data through the comprehensive analysis model to obtain a transformer state change trend and an early warning message;
and step 3: and uploading the monitoring parameters, the transformer state change trend and the early warning message to a cloud management platform at regular time, and carrying out state evaluation on the transformer by the cloud management platform according to the monitoring parameters, the transformer state change trend and the early warning message to obtain a transformer evaluation report and returning the transformer evaluation report to the edge agent module.
Optionally, in step 1, the acquisition module acquires the monitoring parameters, and transmits the monitoring parameters to the intelligent analysis model in the edge agent module, and performs data preprocessing and sequence prediction in the intelligent analysis model to obtain sequence data, specifically:
establishing an LSTM prediction model, and performing data preprocessing and time sequence prediction in an intelligent analysis model by combining the LSTM prediction model to obtain sequence data, wherein the intelligent analysis model comprises 7 standardized models which are respectively a load model, a sleeve model, a cooling model, an insulation model, a partial discharge model, a mechanical structure model and an on-load tap-changer.
Optionally, an LSTM prediction model is established, and data preprocessing and sequence prediction are performed in the intelligent analysis model in combination with the LSTM prediction model to obtain sequence data, specifically:
dividing monitoring parameters into structured data and unstructured data, carrying out stabilization processing on an original time sequence by empirical mode decomposition aiming at the structured data, respectively constructing an LSTM prediction model facing each subsequence component after the processing is finished, inputting the processed structured data into the LSTM prediction model for prediction, and outputting the structured sequence data in the future time through a full connection layer;
and aiming at unstructured data, extracting effective characteristic information of original data through a convolutional neural network, establishing a characteristic vector book, integrating according to the characteristic vector book, memorizing and screening the integrated characteristics, performing fitting prediction, and outputting unstructured sequence data in future time through a full connection layer.
Optionally, in step 2, the sequence data is transmitted to a comprehensive analysis model in the edge agent module, and the sequence data is diagnosed and analyzed through the comprehensive analysis model to obtain a transformer state change trend and an early warning message, specifically:
the comprehensive analysis model comprises an SAE transformer fault diagnosis model and a CNN transformer fault diagnosis model, the structured sequence data are transmitted to the SAE transformer fault diagnosis model for diagnosis and analysis to obtain a transformer state change trend and an early warning message, and the unstructured sequence data are transmitted to the CNN transformer fault diagnosis model for diagnosis and analysis to obtain the transformer state change trend and the early warning message.
Optionally, the structured sequence data is transmitted to an SAE transformer fault diagnosis model for diagnosis and analysis, so as to obtain a transformer state change trend and an early warning message, specifically:
and establishing an SAE transformer fault diagnosis model, training the SAE transformer fault diagnosis model, inputting the structured sequence data into the trained SAE transformer fault diagnosis model, carrying out fault diagnosis on the structured sequence data through the SAE transformer fault diagnosis model, outputting a transformer state change trend in a period of time in the future, and obtaining an early warning message according to the transformer state change trend.
Optionally, training the SAE transformer fault diagnosis model specifically includes the following steps:
a1: pre-training the SAE transformer fault diagnosis model, and performing iterative update on network parameters by adopting an unsupervised greedy training method layer by layer to help the SAE transformer fault diagnosis model to deeply mine the characteristic information of an input sample;
a2: after the pre-training is finished, fine tuning training is carried out on the SAE transformer fault diagnosis model, and overall optimization is carried out on network related parameters through a supervised learning method by combining class labels of data samples so as to obtain effective feature expression.
Optionally, the unstructured sequence data are transmitted to a CNN transformer fault diagnosis model for diagnosis and analysis, so as to obtain a transformer state change trend and an early warning message, specifically:
the method comprises the steps of establishing a CNN transformer fault diagnosis model, training the CNN transformer fault diagnosis model, inputting unstructured sequence data into the trained CNN transformer fault diagnosis model, carrying out fault diagnosis on the unstructured sequence data through the CNN transformer fault diagnosis model, outputting a transformer state change trend within a period of time in the future, and obtaining early warning information according to the transformer state change trend.
Optionally, training a fault diagnosis model of the CNN transformer specifically includes:
inputting a one-dimensional or multi-dimensional array into an input layer, performing feature extraction and sampling treatment through a convolutional layer C1 and a sampling layer S1, performing feature extraction and sampling treatment through a convolutional layer C2 and a sampling layer S2 again after the feature extraction and sampling treatment is completed, outputting a two-dimensional image matrix after the feature extraction and sampling treatment is completed, converting the output two-dimensional image matrix through a full connection layer to obtain a one-dimensional feature vector, fully connecting the one-dimensional feature vector with neurons of the output layer for classification, and completing training.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects: the invention provides a transformer state evaluation and fault early warning method, which specifically comprises the steps of collecting monitoring parameters through a collection module, transmitting the monitoring parameters to an intelligent analysis model in an edge agent module, carrying out data preprocessing and time sequence prediction in the intelligent analysis model to obtain sequence data, transmitting the sequence data to a comprehensive analysis model in the edge agent module, carrying out diagnosis and analysis on the sequence data through the comprehensive analysis model to obtain a transformer state change trend and early warning information, uploading the monitoring parameters, the transformer state change trend and the early warning information to a cloud management platform at regular time, carrying out state evaluation on a transformer by the cloud management platform according to the monitoring parameters, the transformer state change trend and the early warning information to obtain a transformer evaluation report, and returning the transformer evaluation report to the edge agent module; the method fully considers the efficient configuration and management mode of the transformer under the background of the ubiquitous power Internet of things, improves the management method of the transformer and improves the monitoring efficiency of the transformer; the problems that the traditional evaluation algorithm cannot realize dynamic fine monitoring of the transformer, the evaluation time is long, and operation and maintenance evaluation personnel cannot know and maintain the transformer point to point in time and the like are solved; early warning information in the edge agent module is uploaded to the cloud in a timing mode, and quick and efficient prevention and maintenance are facilitated.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a schematic flow chart of a transformer state evaluation and fault early warning method according to an embodiment of the present invention;
FIG. 2 is a schematic structural diagram of an LSTM memory module;
FIG. 3 is a block flow diagram of LSTM-based transformer sequence data prediction;
FIG. 4 is a schematic diagram of a basic structure of a self-encoder;
FIG. 5 is a schematic diagram of the basic structure of a convolutional neural network;
fig. 6 is a block diagram of a transformer state evaluation and fault warning system model according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a transformer state evaluation and fault early warning method, which solves the problems that the traditional evaluation algorithm cannot realize dynamic fine monitoring of a transformer, is long in evaluation time and is not beneficial to operation and maintenance evaluators to timely know and maintain the transformer point to point, improves the management method of the transformer and improves the monitoring efficiency of the transformer.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Fig. 1 is a schematic structural diagram of a transformer state assessment and fault early warning method according to an embodiment of the present invention, and as shown in fig. 1, the transformer state assessment and fault early warning method according to the embodiment of the present invention is applied to a transformer state assessment and fault early warning system, the system includes an acquisition module, an edge agent module, and a cloud management platform, the acquisition module is connected to the edge agent module, the edge agent module is connected to the cloud management platform, the acquisition module includes a plurality of sensors, and is used for acquiring monitoring parameters of a transformer and transmitting the monitoring parameters to the edge agent module, the edge agent module includes an intelligent analysis model, a comprehensive analysis model, and an internet of things intelligent terminal, the intelligent analysis model is used for performing data preprocessing and time sequence prediction on the monitoring parameters of the transformer to obtain sequence data, the comprehensive analysis module is used for analyzing and judging the sequence data, the intelligent terminal of the internet of things is used for checking the state change trend and the early warning information of the transformer, and the cloud management platform is used for realizing real-time storage of monitoring parameters of the acquisition module, real-time recording of the early warning information, evaluation and information archiving of the state of the transformer and regulation and control of an operation strategy of the transformer;
the method comprises the following steps:
step 1: monitoring parameters are acquired through an acquisition module and transmitted to an intelligent analysis model in an edge agent module, and data preprocessing and time sequence prediction are carried out in the intelligent analysis model to obtain sequence data;
step 2: transmitting the sequence data to a comprehensive analysis model in the edge agent module, and carrying out diagnosis and analysis on the sequence data through the comprehensive analysis model to obtain a transformer state change trend and an early warning message;
and step 3: the monitoring parameters, the transformer state change trend and the early warning message are uploaded to a cloud management platform in a timing mode, the cloud management platform carries out state evaluation on the transformer according to the monitoring parameters, the transformer state change trend and the early warning message, a transformer evaluation report is obtained and returned to the edge agent module, and therefore managers can adjust the operation strategy of the transformer and the sensor signal acquisition period.
In step 1, monitoring parameters are acquired through an acquisition module and transmitted to an intelligent analysis model in an edge agent module, data preprocessing and time sequence prediction are carried out in the intelligent analysis model, and sequence data are obtained, wherein the method specifically comprises the following steps:
establishing an LSTM prediction model, and performing data preprocessing and time sequence prediction in an intelligent analysis model by combining the LSTM prediction model to obtain sequence data, wherein the intelligent analysis model comprises 7 standardized models which are respectively a load model, a sleeve model, a cooling model, an insulation model, a partial discharge model, a mechanical structure model and an on-load tap-changer.
As shown in fig. 2, the long-term and short-term memory neural network is a special variant of the recurrent neural network, and the topological structure thereof controls the accumulation speed of information by introducing a gate control unit while retaining a recurrent feedback mechanism, including selectively adding new information and selectively forgetting previously accumulated information, thereby solving the long-term dependence problem existing in sequence modeling, the hidden layer of the LSTM network is replaced by a memory module containing a gate control mechanism from a common neuron, the long-term and short-term memory information of a sequence are respectively stored in a memory unit state c and a hidden layer state h, and the reading and modification of information are realized by controlling a forgetting gate, an input gate and an output gate;
assume that at time t, the inputs to the memory module in the LSTM include: time series input value xtHidden layer t-1 time state ht-1The output of the memory module includes: state h at time t of hidden layertIn the memory module, the calculation formulas of the reset gate and the update gate are as follows:
rt＝σ(Wrxt+Urht-1+br) (1)
zt＝σ(Wzxt+Uzht-1+bz) (2)
in the formula, rt、ztRespectively representing the calculation results of the reset gate and the update gate at time t, Wr、UrWeight matrix representing reset gates, Wz、UzWeight matrix representing the updated gate, br、bzBias terms representing reset gates and update gates, respectively, (representing sigmoid activation functions, whose role is to map variables to intervals 0, 1]Performing the following steps;
the output result of the memory module at the moment t is the hidden layer state htThe calculation formula is as follows:
in the formula (I), the compound is shown in the specification,
candidate hidden layer states representing time t, referred to as candidate states, Wh、UhWeight matrix representing candidate states, bhA bias term representing a candidate state and tanh represents a hyperbolic tangent activation function.
Establishing an LSTM prediction model, and performing data preprocessing and sequence prediction in an intelligent analysis model by combining the LSTM prediction model to obtain sequence data, wherein the method specifically comprises the following steps:
as shown in fig. 3, the monitoring parameters are divided into structured data and unstructured data, the structured data represented by oil chromatography, oil temperature and water content in oil has nonlinear and non-stationary characteristics, the original time sequence is subjected to stabilization treatment through empirical mode decomposition, after the treatment is finished, an LSTM prediction model is respectively constructed for each subsequence component, the treated structured data is input into the LSTM prediction model for prediction, and structured sequence data in a future time are output through a full connection layer and are used for a subsequent intelligent diagnosis model;
aiming at unstructured data such as infrared maps and discharge maps, the method has strong nonlinearity, a convolutional neural network is used for extracting effective characteristic information of original data, a dense and finished characteristic vector book is established, integration is carried out according to the characteristic vector book, the integrated characteristics are memorized and screened, fitting prediction is carried out, unstructured sequence data in a period of time in the future are output through a full connection layer, and the unstructured sequence data are used for a subsequent intelligent diagnosis model.
In step 2, the sequence data are transmitted to a comprehensive analysis model in the edge agent module, and the comprehensive analysis model is used for carrying out diagnosis and analysis on the sequence data to obtain the state change trend and the early warning message of the transformer, which specifically comprises the following steps:
the comprehensive analysis model comprises an SAE transformer fault diagnosis model and a CNN transformer fault diagnosis model, the structured sequence data are transmitted to the SAE transformer fault diagnosis model for diagnosis and analysis to obtain a transformer state change trend and an early warning message, and the unstructured sequence data are transmitted to the CNN transformer fault diagnosis model for diagnosis and analysis to obtain the transformer state change trend and the early warning message.
Aiming at the structured data and the unstructured data, respectively establishing a transformer fault diagnosis model based on SAE and CNN;
as a structural basis of the SAE, the standard Auto Encoder (AE) is a 3-layer symmetric feedforward neural network composed of an input layer, a hidden layer and an output layer, and the structure of the standard Auto Encoder (AE) is shown in fig. 4, the SAE is a deep neural network model formed by stacking a plurality of AEs, wherein the output of each lower AE is used as the input of an upper AE to achieve the step-by-step transmission of network learning and the deep mining of key features, and further, a Softmax classifier is added behind the hidden layer of the highest AE to achieve the accurate judgment of the operating state of the power transformer.
Transmitting the structured sequence data to an SAE transformer fault diagnosis model for diagnosis and analysis to obtain the state change trend and the early warning message of the transformer, which specifically comprises the following steps:
and establishing an SAE transformer fault diagnosis model, training the SAE transformer fault diagnosis model, inputting the structured sequence data into the trained SAE transformer fault diagnosis model, carrying out fault diagnosis on the structured sequence data through the SAE transformer fault diagnosis model, outputting a transformer state change trend in a period of time in the future, and obtaining an early warning message according to the transformer state change trend.
Training an SAE transformer fault diagnosis model, specifically:
the training process of the SAE transformer fault diagnosis model can be mainly divided into 2 stages of pre-training and fine-tuning training, wherein the SAE transformer fault diagnosis model is pre-trained, specifically, network parameters are iteratively updated by adopting an unsupervised layer-by-layer greedy training method and the network parameters are used for helping the SAE transformer fault diagnosis model to deeply mine characteristic information of an input sample; and carrying out fine tuning training on the SAE transformer fault diagnosis model, specifically, carrying out global optimization on network related parameters by a supervised learning method in combination with class labels of data samples so as to obtain effective feature expression.
As a supervised learning algorithm, CNN, like conventional neural networks, must perform model training using labeled data to predict a sample to be recognized through the modelAs a multi-layer neural network, the basic network structure of CNN is shown in FIG. 5, where a one-dimensional or multi-dimensional array is first inputted from an input layer and then passed through a convolutional layer C1And a sampling layer S1Performing feature extraction and sampling, and then performing convolution layer C2And a sampling layer S2Operation of and C1And S1After the two layers are consistent, the two layers are spread into a one-dimensional vector F by a full connection layer3And is passed through the activation function to the output layer output.
Transmitting the unstructured sequence data to a CNN transformer fault diagnosis model for diagnosis and analysis to obtain a transformer state change trend and an early warning message, wherein the method specifically comprises the following steps:
the method comprises the steps of establishing a CNN transformer fault diagnosis model, training the CNN transformer fault diagnosis model, inputting unstructured sequence data into the trained CNN transformer fault diagnosis model, carrying out fault diagnosis on the unstructured sequence data through the CNN transformer fault diagnosis model, outputting a transformer state change trend within a period of time in the future, and obtaining early warning information according to the transformer state change trend.
The specific process for training the CNN transformer fault diagnosis model comprises the following steps:
inputting a sample, convolutional layer C, to the input layer1Performing convolution processing on the images of the input layer by adopting a convolution kernel to obtain a two-dimensional characteristic diagram; sampling layer S1Respectively aligned with a layer C1All the sub-blocks are subjected to pooling treatment to obtain a two-dimensional characteristic diagram (the number is unchanged, and the size is half of the original size); convolutional layer C2And a sampling layer S2The operation steps and principles of1And S1The consistency is achieved; full connection layer pair S2The output two-dimensional image matrix is converted to obtain a one-dimensional characteristic vector F3(ii) a One-dimensional feature vector F3And the neuron of the output layer is fully connected for classification, and training is completed.
The transformer state evaluation and fault early warning system model block diagram provided by the invention is shown in fig. 6, wherein a three-layer system from a parameter layer to an edge layer and then to a management layer corresponds to a terminal-edge-cloud cooperative structure. The early warning information is an early warning radar map.
The invention provides a transformer state evaluation and fault early warning method, which specifically comprises the steps of collecting monitoring parameters through a collection module, transmitting the monitoring parameters to an intelligent analysis model in an edge agent module, carrying out data preprocessing and time sequence prediction in the intelligent analysis model to obtain sequence data, transmitting the sequence data to a comprehensive analysis model in the edge agent module, carrying out diagnosis and analysis on the sequence data through the comprehensive analysis model to obtain a transformer state change trend and early warning information, uploading the monitoring parameters, the transformer state change trend and the early warning information to a cloud management platform at regular time, carrying out state evaluation on a transformer by the cloud management platform according to the monitoring parameters, the transformer state change trend and the early warning information to obtain a transformer evaluation report, and returning the transformer evaluation report to the edge agent module; the method fully considers the efficient configuration and management mode of the transformer under the background of the ubiquitous power Internet of things, improves the management method of the transformer and improves the monitoring efficiency of the transformer; the problems that the traditional evaluation algorithm cannot realize dynamic fine monitoring of the transformer, the evaluation time is long, and operation and maintenance evaluation personnel cannot know and maintain the transformer point to point in time and the like are solved; early warning information in the edge agent module is uploaded to the cloud in a timing mode, and quick and efficient prevention and maintenance are facilitated.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (8)
1. The transformer state assessment and fault early warning method is applied to a transformer state assessment and fault early warning system and is characterized by comprising an acquisition module, an edge agent module and a cloud management platform, wherein the acquisition module is connected with the edge agent module, the edge agent module is connected with the cloud management platform, the acquisition module comprises a plurality of sensors and is used for acquiring monitoring parameters of a transformer and transmitting the monitoring parameters to the edge agent module, the edge agent module comprises an intelligent analysis model and a comprehensive analysis model, the intelligent analysis model is used for carrying out data preprocessing and time sequence prediction on the monitoring parameters of the transformer to obtain sequence data, the comprehensive analysis module is used for carrying out analysis and judgment on the sequence data to obtain state change trends and early warning information of the transformer, and the cloud management platform is used for realizing real-time storage, real-time storage and early warning of the monitoring parameters of the acquisition module, Real-time recording of early warning information, evaluation and information archiving of the state of the transformer and regulation and control of the operation strategy of the transformer;
the method comprises the following steps:
step 1: monitoring parameters are acquired through an acquisition module and transmitted to an intelligent analysis model in an edge agent module, and data preprocessing and time sequence prediction are carried out in the intelligent analysis model to obtain sequence data;
step 2: transmitting the sequence data to a comprehensive analysis model in the edge agent module, and carrying out diagnosis and analysis on the sequence data through the comprehensive analysis model to obtain a transformer state change trend and an early warning message;
and step 3: and uploading the monitoring parameters, the transformer state change trend and the early warning message to a cloud management platform at regular time, and carrying out state evaluation on the transformer by the cloud management platform according to the monitoring parameters, the transformer state change trend and the early warning message to obtain a transformer evaluation report and returning the transformer evaluation report to the edge agent module.
2. The transformer state evaluation and fault early warning method according to claim 1, wherein in step 1, the monitoring parameters are collected by the collection module and transmitted to an intelligent analysis model in the edge agent module, and data preprocessing and sequence prediction are performed in the intelligent analysis model to obtain sequence data, specifically:
establishing an LSTM prediction model, and performing data preprocessing and time sequence prediction in an intelligent analysis model by combining the LSTM prediction model to obtain sequence data, wherein the intelligent analysis model comprises 7 standardized models which are respectively a load model, a sleeve model, a cooling model, an insulation model, a partial discharge model, a mechanical structure model and an on-load tap-changer.
3. The transformer state evaluation and fault early warning method according to claim 2, wherein an LSTM prediction model is established, and data preprocessing and sequence prediction are performed in an intelligent analysis model in combination with the LSTM prediction model to obtain sequence data, and the method specifically comprises the following steps:
dividing monitoring parameters into structured data and unstructured data, carrying out stabilization processing on an original time sequence by empirical mode decomposition aiming at the structured data, respectively constructing an LSTM prediction model facing each subsequence component after the processing is finished, inputting the processed structured data into the LSTM prediction model for prediction, and outputting the structured sequence data in the future time through a full connection layer;
and aiming at unstructured data, extracting effective characteristic information of original data through a convolutional neural network, establishing a characteristic vector book, integrating according to the characteristic vector book, memorizing and screening the integrated characteristics, performing fitting prediction, and outputting unstructured sequence data in future time through a full connection layer.
4. The transformer state evaluation and fault early warning method according to claim 3, wherein in the step 2, the sequence data is transmitted to a comprehensive analysis model in the edge agent module, and the comprehensive analysis model is used for carrying out diagnosis and analysis on the sequence data to obtain a transformer state change trend and an early warning message, and specifically comprises the following steps:
the comprehensive analysis model comprises an SAE transformer fault diagnosis model and a CNN transformer fault diagnosis model, the structured sequence data are transmitted to the SAE transformer fault diagnosis model for diagnosis and analysis to obtain a transformer state change trend and an early warning message, and the unstructured sequence data are transmitted to the CNN transformer fault diagnosis model for diagnosis and analysis to obtain the transformer state change trend and the early warning message.
5. The transformer state evaluation and fault early warning method according to claim 4, wherein the structured sequence data is transmitted to an SAE transformer fault diagnosis model for diagnosis and analysis to obtain a transformer state change trend and an early warning message, and the method specifically comprises the following steps:
and establishing an SAE transformer fault diagnosis model, training the SAE transformer fault diagnosis model, inputting the structured sequence data into the trained SAE transformer fault diagnosis model, carrying out fault diagnosis on the structured sequence data through the SAE transformer fault diagnosis model, outputting a transformer state change trend in a period of time in the future, and obtaining an early warning message according to the transformer state change trend.
6. The transformer state evaluation and fault early warning method according to claim 5, wherein the SAE transformer fault diagnosis model is trained, and the method specifically comprises the following steps:
a1: pre-training the SAE transformer fault diagnosis model, and performing iterative update on network parameters by adopting an unsupervised greedy training method layer by layer to help the SAE transformer fault diagnosis model to deeply mine the characteristic information of an input sample;
a2: after the pre-training is finished, fine tuning training is carried out on the SAE transformer fault diagnosis model, and overall optimization is carried out on network related parameters through a supervised learning method by combining class labels of data samples so as to obtain effective feature expression.
7. The transformer state evaluation and fault early warning method according to claim 4, wherein the unstructured sequence data are transmitted to a CNN transformer fault diagnosis model for diagnosis and analysis to obtain a transformer state change trend and an early warning message, and the method specifically comprises the following steps:
the method comprises the steps of establishing a CNN transformer fault diagnosis model, training the CNN transformer fault diagnosis model, inputting unstructured sequence data into the trained CNN transformer fault diagnosis model, carrying out fault diagnosis on the unstructured sequence data through the CNN transformer fault diagnosis model, outputting a transformer state change trend within a period of time in the future, and obtaining early warning information according to the transformer state change trend.
8. The transformer state evaluation and fault early warning method according to claim 7, wherein a CNN transformer fault diagnosis model is trained, specifically:
inputting one-dimensional or multi-dimensional array into input layer, passing through convolution layer C1And a sampling layer S1Performing feature extraction and sampling treatment, and passing through the convolutional layer C again2And a sampling layer S2And performing feature extraction and sampling treatment, outputting a two-dimensional image matrix after the feature extraction and sampling treatment is completed, converting the output two-dimensional image matrix through a full connection layer to obtain a one-dimensional feature vector, and classifying the one-dimensional feature vector and neurons of the output layer in a full connection manner to complete training.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110976585.0A CN113671421A (en) | 2021-08-24 | 2021-08-24 | Transformer state evaluation and fault early warning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110976585.0A CN113671421A (en) | 2021-08-24 | 2021-08-24 | Transformer state evaluation and fault early warning method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113671421A true CN113671421A (en) | 2021-11-19 |
Family
ID=78545750
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110976585.0A Pending CN113671421A (en) | 2021-08-24 | 2021-08-24 | Transformer state evaluation and fault early warning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113671421A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115327286A (en) * | 2022-10-17 | 2022-11-11 | 国能大渡河检修安装有限公司 | Transformer monitoring method and system applied to power station |
| CN115361061A (en) * | 2022-08-24 | 2022-11-18 | 中铁电气化局集团有限公司 | Optical fiber fault monitoring method |
| CN115754416A (en) * | 2022-11-16 | 2023-03-07 | 国能大渡河瀑布沟发电有限公司 | Edge calculation-based partial discharge analysis system and method for hydraulic generator |
| CN117630758A (en) * | 2024-01-24 | 2024-03-01 | 国能大渡河检修安装有限公司 | Method and system for monitoring health state of power station transformer |
| CN117630758B (en) * | 2024-01-24 | 2024-05-03 | 国能大渡河检修安装有限公司 | Method and system for monitoring health state of power station transformer |
Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108388950A (en) * | 2018-01-29 | 2018-08-10 | 杭州安脉盛智能技术有限公司 | Intelligent transformer O＆M method and system based on big data |
| CN108829908A (en) * | 2018-04-11 | 2018-11-16 | 浙江工业大学 | A kind of circuit structure Reliability Prediction Method based on depth self-encoding encoder network |
| CN108960488A (en) * | 2018-06-13 | 2018-12-07 | 国网山东省电力公司经济技术研究院 | A kind of accurate prediction technique of saturation loading spatial distribution based on deep learning and Multi-source Information Fusion |
| US20190087726A1 (en) * | 2017-08-30 | 2019-03-21 | The Board Of Regents Of The University Of Texas System | Hypercomplex deep learning methods, architectures, and apparatus for multimodal small, medium, and large-scale data representation, analysis, and applications |
| CN109948166A (en) * | 2019-03-25 | 2019-06-28 | 腾讯科技（深圳）有限公司 | Text interpretation method, device, storage medium and computer equipment |
| CN110263944A (en) * | 2019-05-21 | 2019-09-20 | 中国石油大学(华东) | A kind of multivariable failure prediction method and device |
| CN110297179A (en) * | 2018-05-11 | 2019-10-01 | 宫文峰 | Diesel-driven generator failure predication and monitoring system device based on integrated deep learning |
| CN110531266A (en) * | 2019-10-14 | 2019-12-03 | 国网山东省电力公司电力科学研究院 | A kind of synchronous machinery excitation winding interturn short-circuit fault early warning method |
| US20190391574A1 (en) * | 2018-06-25 | 2019-12-26 | Nec Laboratories America, Inc. | Early anomaly prediction on multi-variate time series data |
| CN110763929A (en) * | 2019-08-08 | 2020-02-07 | 浙江大学 | Intelligent monitoring and early warning system and method for convertor station equipment |
| CN110794301A (en) * | 2019-10-12 | 2020-02-14 | 上海雷纳姆电气技术有限公司 | Motor life value judgment method |
| US20200090034A1 (en) * | 2018-09-18 | 2020-03-19 | Salesforce.Com, Inc. | Determining Intent from Unstructured Input to Update Heterogeneous Data Stores |
| US20200151222A1 (en) * | 2018-11-09 | 2020-05-14 | Accenture Global Solutions Limited | Dark web content analysis and identification |
| CN111161883A (en) * | 2019-12-31 | 2020-05-15 | 杭州知盛数据科技有限公司 | Disease prediction system based on variational self-encoder and electronic equipment thereof |
| AU2020101108A4 (en) * | 2020-06-24 | 2020-07-30 | Institute Of Geology And Geophysics, Chinese Academy Of Sciences | A method for geophysical observation information fusion |
| WO2020197533A1 (en) * | 2019-03-22 | 2020-10-01 | General Electric Company | Surrogate of a simulation engine for power system model calibration |
| US20200349430A1 (en) * | 2019-05-03 | 2020-11-05 | Webroot Inc. | System and method for predicting domain reputation |
| WO2021022042A1 (en) * | 2019-07-30 | 2021-02-04 | William Marsh Rice University | System and method for design of rapid excavating and wear-resistant drill bits |
| US20210050029A1 (en) * | 2019-08-12 | 2021-02-18 | Massachusetts Institute Of Technology | Methods and Apparatus for Reducing Stuttering |
| CN112580198A (en) * | 2020-12-03 | 2021-03-30 | 国网山西省电力公司晋城供电公司 | Improved optimization classification method for transformer state evaluation |
| CN112763967A (en) * | 2020-12-11 | 2021-05-07 | 国网辽宁省电力有限公司鞍山供电公司 | BiGRU-based intelligent electric meter metering module fault prediction and diagnosis method |
-
2021
- 2021-08-24 CN CN202110976585.0A patent/CN113671421A/en active Pending
Patent Citations (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190087726A1 (en) * | 2017-08-30 | 2019-03-21 | The Board Of Regents Of The University Of Texas System | Hypercomplex deep learning methods, architectures, and apparatus for multimodal small, medium, and large-scale data representation, analysis, and applications |
| CN108388950A (en) * | 2018-01-29 | 2018-08-10 | 杭州安脉盛智能技术有限公司 | Intelligent transformer O＆M method and system based on big data |
| CN108829908A (en) * | 2018-04-11 | 2018-11-16 | 浙江工业大学 | A kind of circuit structure Reliability Prediction Method based on depth self-encoding encoder network |
| CN110297179A (en) * | 2018-05-11 | 2019-10-01 | 宫文峰 | Diesel-driven generator failure predication and monitoring system device based on integrated deep learning |
| CN108960488A (en) * | 2018-06-13 | 2018-12-07 | 国网山东省电力公司经济技术研究院 | A kind of accurate prediction technique of saturation loading spatial distribution based on deep learning and Multi-source Information Fusion |
| US20190391574A1 (en) * | 2018-06-25 | 2019-12-26 | Nec Laboratories America, Inc. | Early anomaly prediction on multi-variate time series data |
| US20200090034A1 (en) * | 2018-09-18 | 2020-03-19 | Salesforce.Com, Inc. | Determining Intent from Unstructured Input to Update Heterogeneous Data Stores |
| US20200151222A1 (en) * | 2018-11-09 | 2020-05-14 | Accenture Global Solutions Limited | Dark web content analysis and identification |
| WO2020197533A1 (en) * | 2019-03-22 | 2020-10-01 | General Electric Company | Surrogate of a simulation engine for power system model calibration |
| CN109948166A (en) * | 2019-03-25 | 2019-06-28 | 腾讯科技（深圳）有限公司 | Text interpretation method, device, storage medium and computer equipment |
| CN111241855A (en) * | 2019-03-25 | 2020-06-05 | 腾讯科技（深圳）有限公司 | Text translation method, text translation device, storage medium and computer equipment |
| US20200349430A1 (en) * | 2019-05-03 | 2020-11-05 | Webroot Inc. | System and method for predicting domain reputation |
| CN110263944A (en) * | 2019-05-21 | 2019-09-20 | 中国石油大学(华东) | A kind of multivariable failure prediction method and device |
| WO2021022042A1 (en) * | 2019-07-30 | 2021-02-04 | William Marsh Rice University | System and method for design of rapid excavating and wear-resistant drill bits |
| CN110763929A (en) * | 2019-08-08 | 2020-02-07 | 浙江大学 | Intelligent monitoring and early warning system and method for convertor station equipment |
| US20210050029A1 (en) * | 2019-08-12 | 2021-02-18 | Massachusetts Institute Of Technology | Methods and Apparatus for Reducing Stuttering |
| CN110794301A (en) * | 2019-10-12 | 2020-02-14 | 上海雷纳姆电气技术有限公司 | Motor life value judgment method |
| CN110531266A (en) * | 2019-10-14 | 2019-12-03 | 国网山东省电力公司电力科学研究院 | A kind of synchronous machinery excitation winding interturn short-circuit fault early warning method |
| CN111161883A (en) * | 2019-12-31 | 2020-05-15 | 杭州知盛数据科技有限公司 | Disease prediction system based on variational self-encoder and electronic equipment thereof |
| AU2020101108A4 (en) * | 2020-06-24 | 2020-07-30 | Institute Of Geology And Geophysics, Chinese Academy Of Sciences | A method for geophysical observation information fusion |
| CN112580198A (en) * | 2020-12-03 | 2021-03-30 | 国网山西省电力公司晋城供电公司 | Improved optimization classification method for transformer state evaluation |
| CN112763967A (en) * | 2020-12-11 | 2021-05-07 | 国网辽宁省电力有限公司鞍山供电公司 | BiGRU-based intelligent electric meter metering module fault prediction and diagnosis method |
Non-Patent Citations (3)
| Title |
|---|
| 唐文虎;牛哲文;赵柏宁;季天瑶;李梦诗;吴青华;: "数据驱动的人工智能技术在电力设备状态分析中的研究与应用", 高电压技术, no. 09 * |
| 曹成;王丽华;: "变压器智能运维管理平台技术浅析", 山东工业技术, no. 12 * |
| 沈贝敏;周小平;孙卫国;叶韶光;: "基于深度学习的精神病风险预测研究", 计算机仿真, no. 10 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115361061A (en) * | 2022-08-24 | 2022-11-18 | 中铁电气化局集团有限公司 | Optical fiber fault monitoring method |
| CN115327286A (en) * | 2022-10-17 | 2022-11-11 | 国能大渡河检修安装有限公司 | Transformer monitoring method and system applied to power station |
| CN115754416A (en) * | 2022-11-16 | 2023-03-07 | 国能大渡河瀑布沟发电有限公司 | Edge calculation-based partial discharge analysis system and method for hydraulic generator |
| CN115754416B (en) * | 2022-11-16 | 2023-06-27 | 国能大渡河瀑布沟发电有限公司 | Partial discharge analysis system and method for hydro-generator based on edge calculation |
| CN117630758A (en) * | 2024-01-24 | 2024-03-01 | 国能大渡河检修安装有限公司 | Method and system for monitoring health state of power station transformer |
| CN117630758B (en) * | 2024-01-24 | 2024-05-03 | 国能大渡河检修安装有限公司 | Method and system for monitoring health state of power station transformer |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111798051B (en) | Air quality space-time prediction method based on long-term and short-term memory neural network | |
| CN113671421A (en) | Transformer state evaluation and fault early warning method | |
| Ke et al. | Short-term electrical load forecasting method based on stacked auto-encoding and GRU neural network | |
| CN108445752B (en) | Random weight neural network integrated modeling method for self-adaptively selecting depth features | |
| CN111340282B (en) | DA-TCN-based method and system for estimating residual service life of equipment | |
| Dedovic et al. | Forecasting PM10 concentrations using neural networks and system for improving air quality | |
| Liu et al. | Fault diagnosis of water quality monitoring devices based on multiclass support vector machines and rule-based decision trees | |
| CN112598180A (en) | Distributed regional wind power prediction method | |
| CN113391607A (en) | Hydropower station gate control method and system based on deep learning | |
| CN114297907A (en) | Greenhouse environment spatial distribution prediction method and device | |
| CN111222992A (en) | Stock price prediction method of long-short term memory neural network based on attention mechanism | |
| CN112232543A (en) | Multi-site prediction method based on graph convolution network | |
| CN115796434A (en) | Management and control method and device for power distribution network, electronic equipment and storage medium | |
| CN112733440A (en) | Intelligent fault diagnosis method, system, storage medium and equipment for offshore oil-gas-water well | |
| CN116092294A (en) | Expressway traffic flow prediction method based on Transformer and graph meaning network | |
| CN116845874A (en) | Short-term prediction method and device for power load | |
| CN115128978A (en) | Internet of things environment big data detection and intelligent monitoring system | |
| Zhang et al. | Gated recurrent unit-enhanced deep convolutional neural network for real-time industrial process fault diagnosis | |
| CN114444811A (en) | Aluminum electrolysis mixing data superheat degree prediction method based on attention mechanism | |
| Li et al. | PV resource evaluation based on Xception and VGG19 two-layer network algorithm | |
| Tu et al. | Longer time span air pollution prediction: The attention and autoencoder hybrid learning model | |
| CN114995248A (en) | Intelligent maintenance and environmental parameter big data internet of things system | |
| CN113469013A (en) | Motor fault prediction method and system based on transfer learning and time sequence | |
| Chen et al. | Research on load forecasting method of distribution transformer based on deep learning | |
| CN113723670A (en) | Photovoltaic power generation power short-term prediction method with variable time window |
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: 20211119 |