CN117235655B - Intelligent heat supply abnormal condition identification method and system based on federal learning - Google Patents
Intelligent heat supply abnormal condition identification method and system based on federal learning Download PDFInfo
- Publication number
- CN117235655B CN117235655B CN202311515645.4A CN202311515645A CN117235655B CN 117235655 B CN117235655 B CN 117235655B CN 202311515645 A CN202311515645 A CN 202311515645A CN 117235655 B CN117235655 B CN 117235655B
- Authority
- CN
- China
- Prior art keywords
- data
- abnormal
- heat supply
- working condition
- acquiring
- 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
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 302
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000010438 heat treatment Methods 0.000 claims abstract description 64
- 238000012549 training Methods 0.000 claims abstract description 54
- 238000009826 distribution Methods 0.000 claims abstract description 42
- 238000012545 processing Methods 0.000 claims abstract description 16
- 238000001514 detection method Methods 0.000 claims description 105
- 239000011159 matrix material Substances 0.000 claims description 22
- 230000032683 aging Effects 0.000 claims description 15
- 230000002776 aggregation Effects 0.000 claims description 10
- 238000004220 aggregation Methods 0.000 claims description 10
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 10
- 238000013210 evaluation model Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 4
- 238000013135 deep learning Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 abstract description 2
- 238000003745 diagnosis Methods 0.000 description 9
- 238000003860 storage Methods 0.000 description 9
- 230000003042 antagnostic effect Effects 0.000 description 6
- 230000008014 freezing Effects 0.000 description 4
- 238000007710 freezing Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000005856 abnormality Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000008485 antagonism Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Abstract
The invention discloses a federal learning-based intelligent heat supply abnormal condition identification method and system, comprising the following steps: acquiring historical heat supply data of a heat supply pipe network of a target area, and processing the preprocessed historical heat supply data by using a clustering algorithm to acquire abnormal working condition data; acquiring potential characteristic distribution by using a self-encoder network to generate reconstruction data, and carrying out data enhancement on abnormal working condition data through countermeasure training to output an abnormal working condition data set; obtaining an abnormal working condition identification model by utilizing federal learning and model selection; and extracting multidimensional heat supply data of the current time stamp heat supply pipe network as model input, identifying heat supply abnormal working conditions of the current time stamp, and positioning the heat supply abnormal working conditions. The method and the device have the advantages that the potential characteristics of the abnormal heating working condition are sampled for data enhancement, the problem of low recognition accuracy caused by the imbalance of the proportion of the positive and negative samples is avoided, the abnormal recognition is accelerated by utilizing the federal learning algorithm, the extraction accuracy of the characteristics is improved, and the robustness of the network model is enhanced.
Description
Technical Field
The invention relates to the technical field of intelligent heat supply, in particular to a federal learning-based intelligent heat supply abnormal condition identification method and system.
Background
Along with the rapid development of economy and the continuous improvement of the living standard of people, the urban central heating system is rapidly developed, and the construction of intelligent heat supply realizes the optimal operation of the system and improves the conveying capacity of the system through the comprehensive informationized acquisition of heat supply parameters and the multidimensional statistical analysis of data. The intelligent heating system has the functions of monitoring, adjusting, metering, predicting and the like, and ensures the environment-friendly, safe, economical and efficient operation of the system. The intelligent heating system acquires mass operation data by acquiring data information of the intelligent heating system, and the operation state of the heating system can be more comprehensively known by means of the mass data, so that intelligent and intelligent operation regulation and control are realized. And then accurately and reliably identifying the acquired data, and diagnosing the running state of the equipment or the abnormal working condition of the heating system.
The abnormal condition diagnosis of the heating system is the key of the intelligent heating system to run safely and stably. The diagnosis of abnormal working conditions of a heating system is influenced by the problems of lack of data information, unclear perception of system state, low diagnosis precision and the like for a long time, and when the abnormal working conditions occur, because the diagnosis system depends on manual experience for a long time, an expert is required to conduct field investigation and diagnosis, not only manpower and material resources are consumed, but also the problems of low maintenance efficiency, long response time, insufficient diagnosis depth and the like are caused. The diagnosis system relying on the manual experience cannot summarize enough similar abnormal conditions in a short time, and misjudgment and missed judgment of the abnormal conditions can occur. Therefore, how to screen the diagnosis features, and accurately identify the abnormal heating condition by using the diagnosis features is a problem to be solved.
Disclosure of Invention
In order to solve the technical problems, the invention provides a federal learning-based intelligent heat supply abnormal condition identification method and system.
The first aspect of the invention provides a federal learning-based intelligent abnormal heating condition identification method, which comprises the following steps:
acquiring historical heat supply data of a heat supply pipe network of a target area, preprocessing the historical heat supply data, and processing the preprocessed historical heat supply data by using a clustering algorithm to acquire abnormal working condition data in the historical heat supply data;
acquiring potential characteristic distribution of the abnormal working condition data by using a self-encoder network, generating reconstruction data according to the potential characteristic distribution, performing countermeasure training on the reconstruction data to judge deviation between the reconstruction data and the abnormal working condition data, and outputting an abnormal working condition data set according to the deviation;
training a local abnormal detection model based on the abnormal working condition data set by utilizing federal learning, acquiring an error of a global abnormal detection model according to the error of the local abnormal detection model, and acquiring an abnormal working condition identification model through model selection;
and extracting multidimensional heat supply data of the current time stamp heat supply pipe network, inputting the multidimensional heat supply data as a model of the abnormal condition identification model, identifying the abnormal heat supply condition of the current time stamp, and positioning the abnormal heat supply condition.
In the scheme, the clustering algorithm is utilized to process the preprocessed historical heat supply data, and abnormal working condition data in the historical heat supply data is obtained, specifically:
extracting primary side water supply temperature and pressure data and secondary side water supply temperature and pressure data according to the preprocessed historical heat supply data, and dividing the extracted data by taking a heat supply station in a target area as a unit to generate a training set and a testing set;
clustering the training samples in the training set, presetting a parameter initial value, randomly selecting heat supply station data, clustering by using the parameter initial value, and judging whether the profile coefficient meets a preset standard;
when the data of the current heating station is satisfied, outputting a clustering result of the data of the current heating station, and when the data of the current heating station is not satisfied, resetting a parameter initial value, and clustering again until the profile coefficient satisfies a preset standard;
after traversing all the heat station data, obtaining a clustering result of each heat station, comparing the data quantity of different clusters according to the clustering result, and obtaining an abnormal working condition data set through the comparison result;
and acquiring abnormal characteristics of abnormal heat supply working conditions by using a big data method, calculating Manhattan distance between the abnormal characteristics and the abnormal working condition data set, matching the abnormal characteristics with the minimum abnormal working condition data set, and setting an abnormal working condition type label for the abnormal working condition data set.
In this scheme, utilize the unusual operating mode data to obtain potential characteristic distribution from the encoder network, according to potential characteristic distribution generates reconstruction data, specifically is:
acquiring the labeled abnormal working condition data in the abnormal working condition data set, carrying out normalization processing, carrying out coding learning on the normalized data by using an encoder, and extracting the average mean value and the average variance of the abnormal working condition data to acquire potential characteristics corresponding to the abnormal working condition data;
sampling the potential characteristics to obtain potential characteristic distribution, storing the potential characteristic distribution into a hidden layer to serve as hidden layer characteristics, obtaining characteristic reconstruction errors through layer-by-layer training, and carrying out back propagation by utilizing the reconstruction errors to optimize network parameters of a self-encoder;
and according to the data sample obtained after the reconstruction of the abnormal working condition data from the encoder network, carrying out characteristic correction on the data sample, correcting the data sample points with overlarge deviation from the main characteristic sample in the characteristic space, and obtaining final reconstruction data.
In this scheme, to the reconstruction data carries out the antagonism training and judges the deviation of reconstruction data and unusual operating mode data, according to the deviation output unusual operating mode data, specifically does:
Introducing a generated countermeasure network, taking the self-encoder network as a generator network, acquiring reconstruction data with the same potential characteristic distribution as the abnormal working condition data by utilizing the generator, and inputting the reconstruction data into a discriminator network of the generated countermeasure network;
acquiring a characteristic covariance matrix of the input reconstruction data and abnormal working condition data of different types of labels, judging based on the characteristic covariance matrix through a discriminator network, and reading deviation information of the reconstruction data and the abnormal working condition data;
and alternately training the generator network and the discriminator network according to the deviation information until the discriminator network cannot distinguish the reconstruction network and the abnormal working condition data, and generating a large number of abnormal working condition data sets which are provided with labels and have the same potential characteristic distribution as the abnormal working condition data through the trained generation countermeasure network.
In this scheme, utilize federal study to train the local anomaly detection model based on the anomaly condition dataset, obtain the error of global anomaly detection model according to the error of local anomaly detection model to obtain anomaly condition identification model through model selection, specifically:
constructing a local anomaly detection model by generating an countermeasure network, training the local anomaly detection model based on the anomaly condition data set according to federal learning, and acquiring the deviation degree of input data and reconstruction data in the local anomaly detection model;
Presetting a deviation degree threshold, judging that input data is abnormal data when the deviation degree is smaller than the deviation degree threshold, acquiring errors of a local detection model, and acquiring global abnormal detection model errors by aggregation average according to the errors of the local detection model;
screening local detection models smaller than the global abnormal detection model error according to the local detection model error, marking, obtaining scoring information of the marked local detection models according to the error, and selecting a preset number of local abnormal detection models to aggregate by using the scoring information;
and uploading model parameters of the aggregation model to compress, updating parameters of the global anomaly detection model, obtaining optimal model parameters after iteration, and outputting the anomaly condition identification model.
In this scheme, will multidimensional heat supply data is as the model input of unusual operating mode identification model, discerns the unusual operating mode of heat supply of current timestamp, and the location of the unusual operating mode of heat supply is specifically:
acquiring multidimensional heat supply data of different heat supply stations with current time stamps in a target area to construct a multidimensional heat supply sequence, importing the multidimensional heat supply sequence into an abnormal working condition identification model, and judging whether the deviation degree of reconstructed data of the multidimensional heat supply data and original multidimensional heat supply data is smaller than a preset deviation degree threshold value;
When the heat supply time stamp is smaller than the preset time stamp, proving that the multidimensional heat supply sequence of the current time stamp is abnormal working condition data, importing the multidimensional heat supply sequence into a full-connection layer, calculating the probability of the category of the abnormal heat supply working condition, and taking the category label with the maximum probability as the category of the abnormal heat supply working condition;
extracting related heat supply equipment according to the category of abnormal heat supply working conditions corresponding to the multidimensional heat supply sequence, performing coarse positioning according to the heat supply equipment, extracting working condition characteristics of the related heat supply equipment in a normal state, and obtaining deviation of the multidimensional heat supply sequence and the working condition characteristics;
and acquiring weight information of the related heating equipment according to the deviation, and positioning the current timestamp heating abnormal working condition according to the weight information.
The second aspect of the invention also provides an intelligent abnormal heating condition identification system based on federal learning, which comprises: the intelligent abnormal heating condition recognition system comprises a memory and a processor, wherein the memory comprises an intelligent abnormal heating condition recognition method program based on federal learning, and the intelligent abnormal heating condition recognition method program based on federal learning realizes the following steps when being executed by the processor:
acquiring historical heat supply data of a heat supply pipe network of a target area, preprocessing the historical heat supply data, and processing the preprocessed historical heat supply data by using a clustering algorithm to acquire abnormal working condition data in the historical heat supply data;
Acquiring potential characteristic distribution of the abnormal working condition data by using a self-encoder network, generating reconstruction data according to the potential characteristic distribution, performing countermeasure training on the reconstruction data to judge deviation between the reconstruction data and the abnormal working condition data, and outputting an abnormal working condition data set according to the deviation;
training a local abnormal detection model based on the abnormal working condition data set by utilizing federal learning, acquiring an error of a global abnormal detection model according to the error of the local abnormal detection model, and acquiring an abnormal working condition identification model through model selection;
and extracting multidimensional heat supply data of the current time stamp heat supply pipe network, inputting the multidimensional heat supply data as a model of the abnormal condition identification model, identifying the abnormal heat supply condition of the current time stamp, and positioning the abnormal heat supply condition.
The invention discloses a federal learning-based intelligent heat supply abnormal condition identification method and system, comprising the following steps: acquiring historical heat supply data of a heat supply pipe network of a target area, and processing the preprocessed historical heat supply data by using a clustering algorithm to acquire abnormal working condition data; acquiring potential characteristic distribution by using a self-encoder network to generate reconstruction data, and carrying out data enhancement on abnormal working condition data through countermeasure training to output an abnormal working condition data set; obtaining an abnormal working condition identification model by utilizing federal learning and model selection; and extracting multidimensional heat supply data of the current time stamp heat supply pipe network as model input, identifying heat supply abnormal working conditions of the current time stamp, and positioning the heat supply abnormal working conditions. The method and the device have the advantages that the potential characteristics of the abnormal heating working condition are sampled for data enhancement, the problem of low recognition accuracy caused by the imbalance of the proportion of the positive and negative samples is avoided, the abnormal recognition is accelerated by utilizing the federal learning algorithm, the extraction accuracy of the characteristics is improved, and the robustness of the network model is enhanced.
Drawings
FIG. 1 shows a flow chart of the intelligent heating abnormality condition identification method based on federal learning of the present invention;
FIG. 2 illustrates a flow chart of the present invention for data enhancement of abnormal operating condition data;
FIG. 3 shows a flow chart of the present invention for constructing an abnormal condition recognition model;
FIG. 4 shows a block diagram of the intelligent heating anomaly condition recognition system based on federal learning of the present invention.
Detailed Description
In order that the above-recited objects, features and advantages of the present invention will be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description. It should be noted that, in the case of no conflict, the embodiments of the present application and the features in the embodiments may be combined with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those described herein, and therefore the scope of the present invention is not limited to the specific embodiments disclosed below.
FIG. 1 shows a flow chart of the intelligent heating abnormality condition identification method based on federal learning.
As shown in fig. 1, the first aspect of the present invention provides a method for identifying abnormal conditions of intelligent heat supply based on federal learning, including:
S102, acquiring historical heat supply data of a heat supply pipe network of a target area, preprocessing the historical heat supply data, and processing the preprocessed historical heat supply data by using a clustering algorithm to acquire abnormal working condition data in the historical heat supply data;
s104, acquiring potential characteristic distribution of the abnormal working condition data by using a self-encoder network, generating reconstruction data according to the potential characteristic distribution, performing countermeasure training on the reconstruction data to judge deviation between the reconstruction data and the abnormal working condition data, and outputting an abnormal working condition data set according to the deviation;
s106, training a local anomaly detection model based on the anomaly condition data set by utilizing federal learning, acquiring an error of a global anomaly detection model according to the error of the local anomaly detection model, and acquiring an anomaly condition identification model through model selection;
s108, extracting multidimensional heat supply data of the heat supply pipe network with the current time stamp, inputting the multidimensional heat supply data as a model of the abnormal condition identification model, identifying the abnormal heat supply condition with the current time stamp, and positioning the abnormal heat supply condition.
It is to be noted that, the historical heat supply data of the heat supply pipe network of the target area is obtained by the water supply and return temperature sensor and the water supply pressure sensor, and the historical heat supply data is subjected to pretreatment such as data cleaning. Extracting primary side water supply temperature and pressure data and secondary side water supply temperature and pressure data according to the preprocessed historical heat supply data, and dividing the extracted data by taking a heat supply station in a target area as a unit to generate a training set and a testing set; clustering training samples in the training set, presetting a parameter initial value, randomly selecting heating station data, clustering by using the parameter initial value, and judging whether profile coefficients meet preset standards, wherein the value range of the profile coefficients is When the value of the contour coefficient is closer to 1, the clustering is proved to be more reasonable; when the data of the current heating station is satisfied, outputting a clustering result of the data of the current heating station, and when the data of the current heating station is not satisfied, resetting a parameter initial value, and clustering again until the profile coefficient satisfies a preset standard; after traversing all the heat station data, obtaining a clustering result of each heat station, comparing the data quantity of different clusters according to the clustering result, and obtaining an abnormal working condition data set through the comparison result; and acquiring abnormal characteristics of abnormal heat supply working conditions by using a big data method, calculating Manhattan distance between the abnormal characteristics and the abnormal working condition data set, matching the abnormal characteristics with the minimum abnormal working condition data set, and setting an abnormal working condition type label for the abnormal working condition data set.
FIG. 2 illustrates a flow chart of the present invention for data enhancement of abnormal operating condition data.
According to the embodiment of the invention, the deviation between the reconstruction data and the abnormal working condition data is judged by performing countermeasure training on the reconstruction data, and the abnormal working condition data is output according to the deviation, specifically:
s202, introducing a generated countermeasure network, taking the self-encoder network as a generator network, acquiring reconstruction data with the same potential characteristic distribution as the abnormal working condition data by utilizing the generator, and inputting the reconstruction data into a discriminator network of the generated countermeasure network;
S204, acquiring a characteristic covariance matrix of the input reconstruction data and abnormal working condition data of different types of labels, judging based on the characteristic covariance matrix through a discriminator network, and reading deviation information of the reconstruction data and the abnormal working condition data;
s206, alternately training the generator network and the discriminator network according to the deviation information until the discriminator network cannot distinguish the reconstruction network and the abnormal working condition data, and generating a large number of abnormal working condition data sets which are provided with labels and have the same potential characteristic distribution as the abnormal working condition data through the trained generation countermeasure network.
The method comprises the steps of acquiring abnormal working condition data with labels in an abnormal working condition data set, carrying out normalization processing, carrying out coding learning on the normalized data by using an encoder, and extracting the average mean value and the average variance of the abnormal working condition data to acquire potential characteristics corresponding to the abnormal working condition data; sampling the potential characteristics to obtain potential characteristic distribution, storing the potential characteristic distribution into a hidden layer to serve as hidden layer characteristics, obtaining characteristic reconstruction errors through layer-by-layer training, and carrying out back propagation by utilizing the reconstruction errors to optimize network parameters of a self-encoder; and according to the data sample obtained after the reconstruction of the abnormal working condition data by the decoder of the encoder network, carrying out characteristic correction on the data sample, correcting the data sample point with overlarge deviation from the main characteristic sample in the characteristic space, and obtaining final reconstruction data.
Freezing the network weight parameters of the generated antagonistic network generator network in the generation of the antagonistic network, training the identifier network, freezing the network weight parameters of the generated antagonistic network identifier, and training the generator; after alternate training, enabling the value of the loss function of the generated countermeasure network to reach a preset threshold range, and obtaining the trained generated countermeasure model for data enhancement of abnormal working condition data; generating a plurality of abnormal working condition data sets with labels by using the generation countermeasure network, wherein the abnormal working condition data sets have the same potential characteristic distribution as the abnormal working condition data. The characteristic covariance matrix judgment characteristic deviation information of the input reconstruction data and the abnormal working condition data of different types of labels is obtained, and the characteristic covariance matrix judgment characteristic deviation information specifically comprises the following steps:
；
wherein,representing characteristic deviation information>Representing the sum of the elements on the diagonal of the matrix, +.>Characteristic covariance matrix representing abnormal working condition data, < ->Representing the covariance matrix of the reconstructed data.
FIG. 3 shows a flow chart of the present invention for constructing an abnormal condition recognition model.
According to the embodiment of the invention, the local anomaly detection model is trained based on the anomaly condition data set by utilizing federal learning, the error of the global anomaly detection model is obtained according to the error of the local anomaly detection model, and the anomaly condition identification model is obtained through model selection, specifically:
S302, constructing a local anomaly detection model by generating an countermeasure network, training the local anomaly detection model based on the anomaly condition data set according to federal learning, and acquiring the deviation degree of input data and reconstruction data in the local anomaly detection model;
s304, presetting a deviation degree threshold, judging that input data is abnormal data when the deviation degree is smaller than the deviation degree threshold, acquiring errors of a local detection model, and carrying out aggregation average according to the errors of the local detection model to acquire errors of a global abnormal detection model;
s306, screening local detection models smaller than the global abnormal detection model error according to the local detection model error, marking, obtaining grading information of the marked local detection models according to the error, and selecting a preset number of local abnormal detection models to aggregate by utilizing the grading information;
s308, uploading model parameters of the aggregation model, compressing, updating parameters of the global anomaly detection model, obtaining optimal model parameters after iteration, and outputting an anomaly condition identification model.
It should be noted that, the scoring information is obtained according to the ratio of the error of the local detection model to the aggregate error of all the local detection models, and the local anomaly detection model, the detection error of the global anomaly detection model, and the scoring information of the local anomaly detection model are updated every time each iteration.
Acquiring multidimensional heat supply data of different heat supply stations with current time stamps in a target area to construct a multidimensional heat supply sequence, importing the multidimensional heat supply sequence into an abnormal working condition identification model, and judging whether the deviation degree of reconstructed data of the multidimensional heat supply data and original multidimensional heat supply data is smaller than a preset deviation degree threshold value; when the heat supply time stamp is smaller than the preset time stamp, proving that the multidimensional heat supply sequence of the current time stamp is abnormal working condition data, importing the multidimensional heat supply sequence into a full-connection layer, calculating the probability of the category of the abnormal heat supply working condition, and taking the category label with the maximum probability as the category of the abnormal heat supply working condition; extracting related heat supply equipment according to the category of abnormal heat supply working conditions corresponding to the multidimensional heat supply sequence, performing coarse positioning according to the heat supply equipment, extracting working condition characteristics of the related heat supply equipment in a normal state, and obtaining deviation of the multidimensional heat supply sequence and the working condition characteristics; and acquiring weight information of the related heating equipment according to the deviation, and positioning the current timestamp heating abnormal working condition according to the weight information.
According to the embodiment of the invention, the working condition characteristics of the heating equipment in the heating network of the target area under the normal state are obtained, the operation characteristics of the heating equipment under the current preset step length are judged, and the operation characteristics and the working condition characteristics under the normal state are compared and analyzed to generate deviation rate information; judging whether the deviation rate information is larger than a deviation rate threshold value, and if so, generating heat supply equipment aging early warning information; acquiring initial weight of heating equipment in a target heating pipe network according to historical heating abnormal conditions; and constructing a heat supply equipment aging evaluation model by using a deep learning method, inputting the deviation of the operation characteristics and the working condition characteristics in a normal state into the heat supply equipment aging evaluation model to estimate the aging degree of the heat supply equipment, and generating early warning information according to the aging degree to replace the heat supply equipment in time.
FIG. 4 shows a block diagram of the intelligent heating anomaly condition recognition system based on federal learning of the present invention.
The second aspect of the present invention also provides a federally learned intelligent abnormal heating condition recognition system 4, which comprises: the intelligent abnormal heating condition recognition method based on the federal learning comprises a memory 41 and a processor 42, wherein the intelligent abnormal heating condition recognition method based on the federal learning is implemented when executed by the processor as follows:
acquiring historical heat supply data of a heat supply pipe network of a target area, preprocessing the historical heat supply data, and processing the preprocessed historical heat supply data by using a clustering algorithm to acquire abnormal working condition data in the historical heat supply data;
acquiring potential characteristic distribution of the abnormal working condition data by using a self-encoder network, generating reconstruction data according to the potential characteristic distribution, performing countermeasure training on the reconstruction data to judge deviation between the reconstruction data and the abnormal working condition data, and outputting an abnormal working condition data set according to the deviation;
training a local abnormal detection model based on the abnormal working condition data set by utilizing federal learning, acquiring an error of a global abnormal detection model according to the error of the local abnormal detection model, and acquiring an abnormal working condition identification model through model selection;
And extracting multidimensional heat supply data of the current time stamp heat supply pipe network, inputting the multidimensional heat supply data as a model of the abnormal condition identification model, identifying the abnormal heat supply condition of the current time stamp, and positioning the abnormal heat supply condition.
According to the embodiment of the invention, the deviation between the reconstruction data and the abnormal working condition data is judged by performing countermeasure training on the reconstruction data, and the abnormal working condition data is output according to the deviation, specifically:
introducing a generated countermeasure network, taking the self-encoder network as a generator network, acquiring reconstruction data with the same potential characteristic distribution as the abnormal working condition data by utilizing the generator, and inputting the reconstruction data into a discriminator network of the generated countermeasure network;
acquiring a characteristic covariance matrix of the input reconstruction data and abnormal working condition data of different types of labels, judging based on the characteristic covariance matrix through a discriminator network, and reading deviation information of the reconstruction data and the abnormal working condition data;
and alternately training the generator network and the discriminator network according to the deviation information until the discriminator network cannot distinguish the reconstruction network and the abnormal working condition data, and generating a large number of abnormal working condition data sets which are provided with labels and have the same potential characteristic distribution as the abnormal working condition data through the trained generation countermeasure network.
The method comprises the steps of acquiring abnormal working condition data with labels in an abnormal working condition data set, carrying out normalization processing, carrying out coding learning on the normalized data by using an encoder, and extracting the average mean value and the average variance of the abnormal working condition data to acquire potential characteristics corresponding to the abnormal working condition data; sampling the potential characteristics to obtain potential characteristic distribution, storing the potential characteristic distribution into a hidden layer to serve as hidden layer characteristics, obtaining characteristic reconstruction errors through layer-by-layer training, and carrying out back propagation by utilizing the reconstruction errors to optimize network parameters of a self-encoder; and according to the data sample obtained after the reconstruction of the abnormal working condition data by the decoder of the encoder network, carrying out characteristic correction on the data sample, correcting the data sample point with overlarge deviation from the main characteristic sample in the characteristic space, and obtaining final reconstruction data.
Freezing the network weight parameters of the generated antagonistic network generator network in the generation of the antagonistic network, training the identifier network, freezing the network weight parameters of the generated antagonistic network identifier, and training the generator; after alternate training, enabling the value of the loss function of the generated countermeasure network to reach a preset threshold range, and obtaining the trained generated countermeasure model for data enhancement of abnormal working condition data; generating a plurality of abnormal working condition data sets with labels by using the generation countermeasure network, wherein the abnormal working condition data sets have the same potential characteristic distribution as the abnormal working condition data. The characteristic covariance matrix judgment characteristic deviation information of the input reconstruction data and the abnormal working condition data of different types of labels is obtained, and the characteristic covariance matrix judgment characteristic deviation information specifically comprises the following steps:
；
Wherein,representing characteristic deviation information>Representing the sum of the elements on the diagonal of the matrix, +.>Characteristic covariance matrix representing abnormal working condition data, < ->Representing the covariance matrix of the reconstructed data.
According to the embodiment of the invention, the local anomaly detection model is trained based on the anomaly condition data set by utilizing federal learning, the error of the global anomaly detection model is obtained according to the error of the local anomaly detection model, and the anomaly condition identification model is obtained through model selection, specifically:
constructing a local anomaly detection model by generating an countermeasure network, training the local anomaly detection model based on the anomaly condition data set according to federal learning, and acquiring the deviation degree of input data and reconstruction data in the local anomaly detection model;
presetting a deviation degree threshold, judging that input data is abnormal data when the deviation degree is smaller than the deviation degree threshold, acquiring errors of a local detection model, and acquiring global abnormal detection model errors by aggregation average according to the errors of the local detection model;
screening local detection models smaller than the global abnormal detection model error according to the local detection model error, marking, obtaining scoring information of the marked local detection models according to the error, and selecting a preset number of local abnormal detection models to aggregate by using the scoring information;
And uploading model parameters of the aggregation model to compress, updating parameters of the global anomaly detection model, obtaining optimal model parameters after iteration, and outputting the anomaly condition identification model.
It should be noted that, the scoring information is obtained according to the ratio of the error of the local detection model to the aggregate error of all the local detection models, and the local anomaly detection model, the detection error of the global anomaly detection model, and the scoring information of the local anomaly detection model are updated every time each iteration.
The third aspect of the present invention also provides a computer readable storage medium, where the computer readable storage medium includes a federal learning-based intelligent abnormal heating condition identification method program, where the federal learning-based intelligent abnormal heating condition identification method program, when executed by a processor, implements the steps of the federal learning-based intelligent abnormal heating condition identification method according to any one of the above.
In the several embodiments provided in this application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above described device embodiments are only illustrative, e.g. the division of the units is only one logical function division, and there may be other divisions in practice, such as: multiple units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. In addition, the various components shown or discussed may be coupled or directly coupled or communicatively coupled to each other via some interface, whether indirectly coupled or communicatively coupled to devices or units, whether electrically, mechanically, or otherwise.
The units described above as separate components may or may not be physically separate, and components shown as units may or may not be physical units; can be located in one place or distributed to a plurality of network units; some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in each embodiment of the present invention may be integrated in one processing unit, or each unit may be separately used as one unit, or two or more units may be integrated in one unit; the integrated units may be implemented in hardware or in hardware plus software functional units.
Those of ordinary skill in the art will appreciate that: all or part of the steps for implementing the above method embodiments may be implemented by hardware related to program instructions, and the foregoing program may be stored in a computer readable storage medium, where the program, when executed, performs steps including the above method embodiments; and the aforementioned storage medium includes: a mobile storage device, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk or an optical disk, or the like, which can store program codes.
Alternatively, the above-described integrated units of the present invention may be stored in a computer-readable storage medium if implemented in the form of software functional modules and sold or used as separate products. Based on such understanding, the technical solutions of the embodiments of the present invention may be embodied in essence or a part contributing to the prior art in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a removable storage device, ROM, RAM, magnetic or optical disk, or other medium capable of storing program code.
The foregoing is merely illustrative of the present invention, and the present invention is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (4)
1. The intelligent heat supply abnormal condition identification method based on federal learning is characterized by comprising the following steps of:
acquiring historical heat supply data of a heat supply pipe network of a target area, preprocessing the historical heat supply data, and processing the preprocessed historical heat supply data by using a clustering algorithm to acquire abnormal working condition data in the historical heat supply data;
acquiring potential characteristic distribution of the abnormal working condition data by using a self-encoder network, generating reconstruction data according to the potential characteristic distribution, performing countermeasure training on the reconstruction data to judge deviation between the reconstruction data and the abnormal working condition data, and outputting an abnormal working condition data set according to the deviation;
training a local abnormal detection model based on the abnormal working condition data set by utilizing federal learning, acquiring an error of a global abnormal detection model according to the error of the local abnormal detection model, and acquiring an abnormal working condition identification model through model selection;
extracting multidimensional heat supply data of a current time stamp heat supply pipe network, inputting the multidimensional heat supply data as a model of the abnormal condition identification model, identifying a heat supply abnormal condition of the current time stamp, and positioning the heat supply abnormal condition;
Processing the preprocessed historical heat supply data by using a clustering algorithm to acquire abnormal working condition data in the historical heat supply data, wherein the abnormal working condition data specifically comprises:
extracting primary side water supply temperature and pressure data and secondary side water supply temperature and pressure data according to the preprocessed historical heat supply data, and dividing the extracted data by taking a heat supply station in a target area as a unit to generate a training set and a testing set;
clustering the training samples in the training set, presetting a parameter initial value, randomly selecting heat supply station data, clustering by using the parameter initial value, and judging whether the profile coefficient meets a preset standard;
when the data of the current heating station is satisfied, outputting a clustering result of the data of the current heating station, and when the data of the current heating station is not satisfied, resetting a parameter initial value, and clustering again until the profile coefficient satisfies a preset standard;
after traversing all the heat station data, obtaining a clustering result of each heat station, comparing the data quantity of different clusters according to the clustering result, and obtaining an abnormal working condition data set through the comparison result;
acquiring abnormal characteristics of abnormal heat supply working conditions by using a big data method, calculating Manhattan distance between the abnormal characteristics and the abnormal working condition data set, matching the abnormal characteristics with the minimum abnormal working condition data set, and setting an abnormal working condition type label for the abnormal working condition data set;
Training a local abnormal detection model based on the abnormal working condition data set by utilizing federal learning, acquiring an error of a global abnormal detection model according to the error of the local abnormal detection model, and acquiring an abnormal working condition identification model through model selection, wherein the method specifically comprises the following steps:
constructing a local anomaly detection model by generating an countermeasure network, training the local anomaly detection model based on the anomaly condition data set according to federal learning, and acquiring the deviation degree of input data and reconstruction data in the local anomaly detection model;
presetting a deviation degree threshold, judging that input data is abnormal data when the deviation degree is smaller than the deviation degree threshold, acquiring errors of a local detection model, and acquiring global abnormal detection model errors by aggregation average according to the errors of the local detection model;
screening local detection models smaller than the global abnormal detection model error according to the local detection model error, marking, obtaining scoring information of the marked local detection models according to the error, and selecting a preset number of local abnormal detection models to aggregate by using the scoring information;
uploading model parameters of the aggregation model to compress, updating parameters of the global anomaly detection model, obtaining optimal model parameters after iteration, and outputting an anomaly condition identification model;
The multidimensional heat supply data is used as the model input of the abnormal condition identification model, the heat supply abnormal condition of the current timestamp is identified, and the heat supply abnormal condition is positioned, specifically:
acquiring multidimensional heat supply data of different heat supply stations with current time stamps in a target area to construct a multidimensional heat supply sequence, importing the multidimensional heat supply sequence into an abnormal working condition identification model, and judging whether the deviation degree of reconstructed data of the multidimensional heat supply data and original multidimensional heat supply data is smaller than a preset deviation degree threshold value;
when the heat supply time stamp is smaller than the preset time stamp, proving that the multidimensional heat supply sequence of the current time stamp is abnormal working condition data, importing the multidimensional heat supply sequence into a full-connection layer, calculating the probability of the category of the abnormal heat supply working condition, and taking the category label with the maximum probability as the category of the abnormal heat supply working condition;
extracting related heat supply equipment according to the category of abnormal heat supply working conditions corresponding to the multidimensional heat supply sequence, performing coarse positioning according to the heat supply equipment, extracting working condition characteristics of the related heat supply equipment in a normal state, and obtaining deviation of the multidimensional heat supply sequence and the working condition characteristics;
acquiring weight information of the related heating equipment according to the deviation, and positioning the current timestamp abnormal heating condition according to the weight information;
Performing countermeasure training on the reconstruction data to judge deviation between the reconstruction data and abnormal working condition data, and outputting the abnormal working condition data according to the deviation, wherein the method specifically comprises the following steps:
introducing a generated countermeasure network, taking the self-encoder network as a generator network, acquiring reconstruction data with the same potential characteristic distribution as the abnormal working condition data by utilizing the generator, and inputting the reconstruction data into a discriminator network of the generated countermeasure network;
acquiring characteristic covariance matrixes of the input reconstruction data and abnormal working condition data of different types of labels, judging the characteristic covariance matrixes by a discriminator network, and reading deviation information of the reconstruction data and the abnormal working condition data, wherein the formula of the deviation information is as follows:
；
wherein,representing characteristic deviation information>Representing the sum of the elements on the diagonal of the matrix, +.>Characteristic covariance matrix representing abnormal working condition data, < ->A covariance matrix representing the reconstructed data;
alternately training the generator network and the discriminator network according to the deviation information until the discriminator network cannot distinguish the reconstruction network and the abnormal working condition data, and generating a large number of abnormal working condition data sets which are provided with labels and have the same potential characteristic distribution as the abnormal working condition data through the trained generation countermeasure network;
Acquiring working condition characteristics of heat supply equipment in a heat supply pipe network of a target area in a normal state, judging the operation characteristics of the heat supply equipment in a current preset step length, comparing and analyzing the operation characteristics with the working condition characteristics in the normal state, and generating deviation rate information; judging whether the deviation rate information is larger than a deviation rate threshold value, and if so, generating heat supply equipment aging early warning information; acquiring initial weight of heating equipment in a target heating pipe network according to historical heating abnormal conditions; and constructing a heat supply equipment aging evaluation model by using a deep learning method, inputting the deviation of the operation characteristics and the working condition characteristics in a normal state into the heat supply equipment aging evaluation model to estimate the aging degree of the heat supply equipment, and generating early warning information according to the aging degree to replace the heat supply equipment in time.
2. The intelligent heat supply abnormal condition identification method based on federal learning according to claim 1, wherein the abnormal condition data is obtained by utilizing a self-encoder network to obtain potential characteristic distribution, and reconstruction data is generated according to the potential characteristic distribution, specifically:
acquiring the labeled abnormal working condition data in the abnormal working condition data set, carrying out normalization processing, carrying out coding learning on the normalized data by using an encoder, and extracting the average mean value and the average variance of the abnormal working condition data to acquire potential characteristics corresponding to the abnormal working condition data;
Sampling the potential characteristics to obtain potential characteristic distribution, storing the potential characteristic distribution into a hidden layer to serve as hidden layer characteristics, obtaining characteristic reconstruction errors through layer-by-layer training, and carrying out back propagation by utilizing the reconstruction errors to optimize network parameters of a self-encoder;
and according to the data sample obtained after the reconstruction of the abnormal working condition data from the encoder network, carrying out characteristic correction on the data sample, correcting the data sample points with overlarge deviation from the main characteristic sample in the characteristic space, and obtaining final reconstruction data.
3. Intelligent heat supply abnormal condition recognition system based on federal learning, which is characterized by comprising: the intelligent abnormal heating condition recognition system comprises a memory and a processor, wherein the memory comprises an intelligent abnormal heating condition recognition method program based on federal learning, and the intelligent abnormal heating condition recognition method program based on federal learning realizes the following steps when being executed by the processor:
acquiring historical heat supply data of a heat supply pipe network of a target area, preprocessing the historical heat supply data, and processing the preprocessed historical heat supply data by using a clustering algorithm to acquire abnormal working condition data in the historical heat supply data;
Acquiring potential characteristic distribution of the abnormal working condition data by using a self-encoder network, generating reconstruction data according to the potential characteristic distribution, performing countermeasure training on the reconstruction data to judge deviation between the reconstruction data and the abnormal working condition data, and outputting an abnormal working condition data set according to the deviation;
training a local abnormal detection model based on the abnormal working condition data set by utilizing federal learning, acquiring an error of a global abnormal detection model according to the error of the local abnormal detection model, and acquiring an abnormal working condition identification model through model selection;
extracting multidimensional heat supply data of a current time stamp heat supply pipe network, inputting the multidimensional heat supply data as a model of the abnormal condition identification model, identifying a heat supply abnormal condition of the current time stamp, and positioning the heat supply abnormal condition;
processing the preprocessed historical heat supply data by using a clustering algorithm to acquire abnormal working condition data in the historical heat supply data, wherein the abnormal working condition data specifically comprises:
extracting primary side water supply temperature and pressure data and secondary side water supply temperature and pressure data according to the preprocessed historical heat supply data, and dividing the extracted data by taking a heat supply station in a target area as a unit to generate a training set and a testing set;
Clustering the training samples in the training set, presetting a parameter initial value, randomly selecting heat supply station data, clustering by using the parameter initial value, and judging whether the profile coefficient meets a preset standard;
when the data of the current heating station is satisfied, outputting a clustering result of the data of the current heating station, and when the data of the current heating station is not satisfied, resetting a parameter initial value, and clustering again until the profile coefficient satisfies a preset standard;
after traversing all the heat station data, obtaining a clustering result of each heat station, comparing the data quantity of different clusters according to the clustering result, and obtaining an abnormal working condition data set through the comparison result;
acquiring abnormal characteristics of abnormal heat supply working conditions by using a big data method, calculating Manhattan distance between the abnormal characteristics and the abnormal working condition data set, matching the abnormal characteristics with the minimum abnormal working condition data set, and setting an abnormal working condition type label for the abnormal working condition data set;
training a local abnormal detection model based on the abnormal working condition data set by utilizing federal learning, acquiring an error of a global abnormal detection model according to the error of the local abnormal detection model, and acquiring an abnormal working condition identification model through model selection, wherein the method specifically comprises the following steps:
Constructing a local anomaly detection model by generating an countermeasure network, training the local anomaly detection model based on the anomaly condition data set according to federal learning, and acquiring the deviation degree of input data and reconstruction data in the local anomaly detection model;
presetting a deviation degree threshold, judging that input data is abnormal data when the deviation degree is smaller than the deviation degree threshold, acquiring errors of a local detection model, and acquiring global abnormal detection model errors by aggregation average according to the errors of the local detection model;
screening local detection models smaller than the global abnormal detection model error according to the local detection model error, marking, obtaining scoring information of the marked local detection models according to the error, and selecting a preset number of local abnormal detection models to aggregate by using the scoring information;
uploading model parameters of the aggregation model to compress, updating parameters of the global anomaly detection model, obtaining optimal model parameters after iteration, and outputting an anomaly condition identification model;
the multidimensional heat supply data is used as the model input of the abnormal condition identification model, the heat supply abnormal condition of the current timestamp is identified, and the heat supply abnormal condition is positioned, specifically:
Acquiring multidimensional heat supply data of different heat supply stations with current time stamps in a target area to construct a multidimensional heat supply sequence, importing the multidimensional heat supply sequence into an abnormal working condition identification model, and judging whether the deviation degree of reconstructed data of the multidimensional heat supply data and original multidimensional heat supply data is smaller than a preset deviation degree threshold value;
when the heat supply time stamp is smaller than the preset time stamp, proving that the multidimensional heat supply sequence of the current time stamp is abnormal working condition data, importing the multidimensional heat supply sequence into a full-connection layer, calculating the probability of the category of the abnormal heat supply working condition, and taking the category label with the maximum probability as the category of the abnormal heat supply working condition;
extracting related heat supply equipment according to the category of abnormal heat supply working conditions corresponding to the multidimensional heat supply sequence, performing coarse positioning according to the heat supply equipment, extracting working condition characteristics of the related heat supply equipment in a normal state, and obtaining deviation of the multidimensional heat supply sequence and the working condition characteristics;
acquiring weight information of the related heating equipment according to the deviation, and positioning the current timestamp abnormal heating condition according to the weight information;
performing countermeasure training on the reconstruction data to judge deviation between the reconstruction data and abnormal working condition data, and outputting the abnormal working condition data according to the deviation, wherein the method specifically comprises the following steps:
Introducing a generated countermeasure network, taking the self-encoder network as a generator network, acquiring reconstruction data with the same potential characteristic distribution as the abnormal working condition data by utilizing the generator, and inputting the reconstruction data into a discriminator network of the generated countermeasure network;
acquiring characteristic covariance matrixes of the input reconstruction data and abnormal working condition data of different types of labels, judging the characteristic covariance matrixes by a discriminator network, and reading deviation information of the reconstruction data and the abnormal working condition data, wherein the formula of the deviation information is as follows:
；
wherein,representing characteristic deviation information>Representing the sum of the elements on the diagonal of the matrix, +.>Characteristic covariance matrix representing abnormal working condition data, < ->A covariance matrix representing the reconstructed data;
alternately training the generator network and the discriminator network according to the deviation information until the discriminator network cannot distinguish the reconstruction network and the abnormal working condition data, and generating a large number of abnormal working condition data sets which are provided with labels and have the same potential characteristic distribution as the abnormal working condition data through the trained generation countermeasure network;
acquiring working condition characteristics of heat supply equipment in a heat supply pipe network of a target area in a normal state, judging the operation characteristics of the heat supply equipment in a current preset step length, comparing and analyzing the operation characteristics with the working condition characteristics in the normal state, and generating deviation rate information; judging whether the deviation rate information is larger than a deviation rate threshold value, and if so, generating heat supply equipment aging early warning information; acquiring initial weight of heating equipment in a target heating pipe network according to historical heating abnormal conditions; and constructing a heat supply equipment aging evaluation model by using a deep learning method, inputting the deviation of the operation characteristics and the working condition characteristics in a normal state into the heat supply equipment aging evaluation model to estimate the aging degree of the heat supply equipment, and generating early warning information according to the aging degree to replace the heat supply equipment in time.
4. The intelligent abnormal heating condition identification system based on federal learning according to claim 3, wherein the abnormal condition data is obtained by using a self-encoder network to obtain potential characteristic distribution, and reconstruction data is generated according to the potential characteristic distribution, specifically:
acquiring the labeled abnormal working condition data in the abnormal working condition data set, carrying out normalization processing, carrying out coding learning on the normalized data by using an encoder, and extracting the average mean value and the average variance of the abnormal working condition data to acquire potential characteristics corresponding to the abnormal working condition data;
sampling the potential characteristics to obtain potential characteristic distribution, storing the potential characteristic distribution into a hidden layer to serve as hidden layer characteristics, obtaining characteristic reconstruction errors through layer-by-layer training, and carrying out back propagation by utilizing the reconstruction errors to optimize network parameters of a self-encoder;
and according to the data sample obtained after the reconstruction of the abnormal working condition data from the encoder network, carrying out characteristic correction on the data sample, correcting the data sample points with overlarge deviation from the main characteristic sample in the characteristic space, and obtaining final reconstruction data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311515645.4A CN117235655B (en) | 2023-11-15 | 2023-11-15 | Intelligent heat supply abnormal condition identification method and system based on federal learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311515645.4A CN117235655B (en) | 2023-11-15 | 2023-11-15 | Intelligent heat supply abnormal condition identification method and system based on federal learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117235655A CN117235655A (en) | 2023-12-15 |
| CN117235655B true CN117235655B (en) | 2024-02-02 |
Family
ID=89086519
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311515645.4A Active CN117235655B (en) | 2023-11-15 | 2023-11-15 | Intelligent heat supply abnormal condition identification method and system based on federal learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117235655B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117556274B (en) * | 2024-01-11 | 2024-03-15 | 北京英沣特能源技术有限公司 | Temperature data anomaly analysis method for heat pipe backboard |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109948117A (en) * | 2019-03-13 | 2019-06-28 | 南京航空航天大学 | A kind of satellite method for detecting abnormality fighting network self-encoding encoder |
| CN112862347A (en) * | 2021-03-02 | 2021-05-28 | 同济大学 | Equipment abnormity monitoring method and system based on federal learning, storage medium and terminal |
| CN113962425A (en) * | 2021-08-02 | 2022-01-21 | 北京百度网讯科技有限公司 | Heating data generation method, device, equipment and computer storage medium |
| CN114462509A (en) * | 2022-01-12 | 2022-05-10 | 重庆邮电大学 | Distributed Internet of things equipment anomaly detection method |
| CN114820303A (en) * | 2022-03-24 | 2022-07-29 | 南京邮电大学 | Method, system and storage medium for reconstructing super-resolution face image from low-definition image |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20230308465A1 (en) * | 2023-04-12 | 2023-09-28 | Roobaea Alroobaea | System and method for dnn-based cyber-security using federated learning-based generative adversarial network |
-
2023
- 2023-11-15 CN CN202311515645.4A patent/CN117235655B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109948117A (en) * | 2019-03-13 | 2019-06-28 | 南京航空航天大学 | A kind of satellite method for detecting abnormality fighting network self-encoding encoder |
| CN112862347A (en) * | 2021-03-02 | 2021-05-28 | 同济大学 | Equipment abnormity monitoring method and system based on federal learning, storage medium and terminal |
| CN113962425A (en) * | 2021-08-02 | 2022-01-21 | 北京百度网讯科技有限公司 | Heating data generation method, device, equipment and computer storage medium |
| CN114462509A (en) * | 2022-01-12 | 2022-05-10 | 重庆邮电大学 | Distributed Internet of things equipment anomaly detection method |
| CN114820303A (en) * | 2022-03-24 | 2022-07-29 | 南京邮电大学 | Method, system and storage medium for reconstructing super-resolution face image from low-definition image |
Non-Patent Citations (1)
| Title |
|---|
| 基于数据驱动的供热系统异常工况诊断研究;钟健;中国优秀硕士学位论文全文数据库;26-66 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117235655A (en) | 2023-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111967502B (en) | Network intrusion detection method based on conditional variation self-encoder | |
| CN117235655B (en) | Intelligent heat supply abnormal condition identification method and system based on federal learning | |
| CN110458195B (en) | Abnormal data discrimination method based on multi-criterion fusion | |
| CN114494837B (en) | Intelligent density identification method and system for fishery resources | |
| CN110059845B (en) | Metering device clock error trend prediction method based on time sequence evolution gene model | |
| CN106708738B (en) | Software test defect prediction method and system | |
| CN108304567B (en) | Method and system for identifying working condition mode and classifying data of high-voltage transformer | |
| CN105354198A (en) | Data processing method and apparatus | |
| CN116881749B (en) | Pollution site construction monitoring method and system | |
| CN113270197A (en) | Health prediction method, system and storage medium based on artificial intelligence | |
| CN113822366A (en) | Service index abnormality detection method and device, electronic equipment and storage medium | |
| CN117172556B (en) | Construction risk early warning method and system for bridge engineering | |
| CN116384837B (en) | Information processing method, system and inspector terminal for engineering quality inspection | |
| CN115600513A (en) | Karst collapse monitoring early warning and prevention and control integrated informatization simulation studying and judging system | |
| CN117421692B (en) | Garbage illegal delivery identification method, device and equipment for garbage delivery station | |
| CN117368651B (en) | Comprehensive analysis system and method for faults of power distribution network | |
| CN114139931A (en) | Enterprise data evaluation method and device, computer equipment and storage medium | |
| CN116520068B (en) | Diagnostic method, device, equipment and storage medium for electric power data | |
| CN111506636A (en) | System and method for analyzing residential electricity consumption behavior based on autoregressive and neighbor algorithm | |
| CN116126807A (en) | Log analysis method and related device | |
| CN116502155A (en) | Safety supervision system for numerical control electric screw press | |
| CN114580472B (en) | Large-scale equipment fault prediction method with repeated cause and effect and attention in industrial internet | |
| CN116108376A (en) | Monitoring system and method for preventing electricity stealing, electronic equipment and medium | |
| CN115392710A (en) | Wind turbine generator operation decision method and system based on data filtering | |
| CN112014821B (en) | Unknown vehicle target identification method based on radar broadband characteristics |
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 |