CN112766342A - Abnormity detection method for electrical equipment - Google Patents
Abnormity detection method for electrical equipment Download PDFInfo
- Publication number
- CN112766342A CN112766342A CN202110034065.8A CN202110034065A CN112766342A CN 112766342 A CN112766342 A CN 112766342A CN 202110034065 A CN202110034065 A CN 202110034065A CN 112766342 A CN112766342 A CN 112766342A
- Authority
- CN
- China
- Prior art keywords
- multivariate time
- data set
- detection model
- anomaly
- time series
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 94
- 238000000034 method Methods 0.000 claims abstract description 76
- 230000002159 abnormal effect Effects 0.000 claims abstract description 34
- 230000005856 abnormality Effects 0.000 claims abstract description 29
- 238000012549 training Methods 0.000 claims abstract description 24
- 238000012545 processing Methods 0.000 claims abstract description 8
- 238000009826 distribution Methods 0.000 claims description 25
- 238000003860 storage Methods 0.000 claims description 13
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 5
- 230000008878 coupling Effects 0.000 claims description 3
- 238000010168 coupling process Methods 0.000 claims description 3
- 238000005859 coupling reaction Methods 0.000 claims description 3
- 238000004891 communication Methods 0.000 description 14
- 238000010586 diagram Methods 0.000 description 14
- 230000008569 process Effects 0.000 description 13
- 238000004364 calculation method Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000007476 Maximum Likelihood Methods 0.000 description 4
- 230000001133 acceleration Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 4
- 125000004122 cyclic group Chemical group 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 230000002093 peripheral effect Effects 0.000 description 3
- 238000003070 Statistical process control Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000005315 distribution function Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 206010000117 Abnormal behaviour Diseases 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000002547 anomalous effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000001816 cooling Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 230000002542 deteriorative effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000007667 floating Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000011541 reaction mixture Substances 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000007858 starting material Substances 0.000 description 1
- 230000036962 time dependent Effects 0.000 description 1
- 230000007723 transport mechanism Effects 0.000 description 1
- 238000004804 winding Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Human Resources & Organizations (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Probability & Statistics with Applications (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses an anomaly detection method of electrical equipment, which comprises the following steps: acquiring index data of the electrical equipment; constructing a multivariate time sequence by using the index data; inputting the multivariate time sequence into an anomaly detection model for processing to generate an anomaly score sequence containing a plurality of anomaly scores; determining a threshold for scoring anomalies based on the anomaly score sequence; determining whether the electrical device is abnormal based on the abnormality score and the threshold. The invention also discloses a method for training and generating the corresponding abnormal detection model of the electrical equipment and corresponding computing equipment.
Description
Technical Field
The invention relates to the technical field of equipment monitoring, in particular to an abnormity detection method for electrical equipment.
Background
With the development of social economy and the important position of industry in the economic industry of China, the safety of industrial systems is more and more emphasized by people, and electrical equipment required by factories is more and more advanced and multifunctional. The electrical equipment of the unit is an important component of the whole factory, and the normal operation and equipment safety of the factory are guaranteed to a great extent. In recent years, high-parameter and high-capacity units are continuously emerged along with the rapid development of the industry in China, the reasons for abnormity in the production process are more and more complex, and the safety and the economy of a factory are seriously influenced.
The abnormal detection of the equipment refers to making a reasonable plan, analyzing the working condition of the equipment by measuring parameters such as electrical quantity, mechanical quantity, thermal quantity and the like according to the plan and combining historical data and a working environment, and objectively evaluating the running state of the equipment. The loss caused by the failure of the electrical equipment is far beyond the electrical equipment. Therefore, the abnormity detection work of the electrical equipment is well done, a foundation is laid for reliable operation, maintenance arrangement and later-stage fault analysis of the equipment, and the method has important significance for reliable economic production of enterprises.
Since the electrical equipment comprises a large variety of equipment and components, the resulting anomalies are accompanied by different manifestations: voltage quality (overvoltage or undervoltage, voltage unbalance), harmonic distortion (usually generated by electronic commutation systems, which cause standard, negative-sequence and positive-sequence rotating magnetic fields in the machine), power factor, stator winding turn-to-turn short circuit faults, rotor bar breakage, stator/rotor eccentricity, air gap unevenness, coupler, bearing, etc. faults, which can directly or indirectly affect the normal operation of the unit equipment. Therefore, the anomaly detection of the equipment, especially in this scenario, is of particular importance.
Based on this, a new abnormality detection scheme for electrical equipment is required.
Disclosure of Invention
To this end, the present invention provides an anomaly detection scheme for electrical equipment in an attempt to solve or at least alleviate at least one of the problems presented above.
According to one aspect of the present invention, there is provided a method of training a generation of an anomaly detection model of an electrical device, comprising the steps of: acquiring index data of the electrical equipment; constructing a multivariate time sequence data set by using the index data; preprocessing the multivariate time sequence data set to obtain a preprocessed multivariate time sequence data set; and training an abnormality detection model by using the preprocessed multivariate time sequence dataset.
Optionally, in the method according to the invention, the anomaly detection model is a variational bayes-based generative model.
Optionally, in the method according to the invention, the anomaly detection model is generated by a coupling of an encoding component and a decoding component.
Optionally, in the method according to the present invention, the step of training an anomaly detection model using the preprocessed multivariate time series dataset comprises: extracting characteristic information from the processed multivariate time series data set through an encoding component; reconstructing a multi-element time sequence data set by the characteristic information through a decoding component; and calculating a reconstruction error and adjusting network parameters of the coding assembly and the decoding assembly based on the processed multivariate time sequence data set and the reconstructed multivariate time sequence data set until the reconstruction error is minimum, finishing training and generating an abnormal detection model.
Optionally, in the method according to the invention, the encoding component and the decoding component each comprise a long-short term memory artificial neural network.
Alternatively, in the method according to the present invention, the reconstruction error is obtained by calculating a first index for the reconstructed multivariate time-series data and a second index for the distribution of the feature information, wherein the first index is a conditional probability composed of the reconstructed multivariate time-series data set and the feature information; the second index is a KL divergence used to measure the distribution of the characteristic information.
Optionally, in the method according to the present invention, the step of preprocessing the multivariate time series dataset to obtain a preprocessed multivariate time series dataset includes: identifying error signals from the multivariate time sequence data set by combining a preset threshold value and a time-frequency domain image; and screening out error signals from the multivariate time sequence data set to obtain a preprocessed multivariate time sequence data set.
Optionally, in the method according to the invention, the indicator data comprises at least one or more of the following data: current signal, voltage signal, peak factor, voltage deviation factor, three-phase average voltage.
Optionally, the method according to the invention further comprises the steps of: and updating parameters of the abnormity detection model periodically.
According to another aspect of the present invention, there is provided an abnormality detection method of an electric device, including the steps of: acquiring index data of the electrical equipment; constructing a multivariate time sequence by using the index data; inputting the multivariate time sequence into an anomaly detection model for processing to generate an anomaly score sequence containing a plurality of anomaly scores; determining a threshold for scoring anomalies based on the anomaly score sequence; determining whether the electrical equipment is abnormal based on the abnormality score and a threshold, wherein the abnormality detection model is obtained by executing the method as described above.
Optionally, in the method according to the present invention, the step of determining whether the electrical device is abnormal based on the abnormality score and the threshold value includes: if the abnormal value is larger than the threshold value, determining that the electrical equipment is normal in the corresponding time period; and if the abnormal score is not larger than the threshold value, determining that the electrical equipment is abnormal in the corresponding time period.
Optionally, in the method according to the present invention, the step of determining a threshold value for scoring anomaly based on the anomaly score sequence comprises: fitting generalized pareto distribution by using an extreme value theory; estimating parameters in generalized pareto distribution by using maximum likelihood estimation; based on the estimated parameters, a threshold is determined.
According to yet another aspect of the present invention, there is provided a computing device comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the methods described above.
According to a further aspect of the invention there is provided a computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a computing device, cause the computing device to perform any of the methods described above.
In summary, according to the scheme of the present invention, when an anomaly detection model is constructed and trained, the input of the model is a multivariate time series (i.e. multiple index data), and the input is defined as the multivariate time series, which can greatly improve the anomaly detection efficiency and overcome the defect that different indexes need to be individually detected when a statistical process control method (EWMA) is used in the past.
In addition, when the abnormality detection model is trained, index data of a large number of electrical devices of different device types are acquired, and therefore a multi-element time sequence data set is formed and used as a training sample, and the trained abnormality detection model has good generalization. And the acquired multivariate time sequence data set is preprocessed, error signals in the multivariate time sequence data set are deleted, and then the multivariate time sequence data set is input into the model for training, so that the model can well learn the data characteristics of different equipment types in a normal state, and further can detect any data characteristics (such as abnormal points with slow deterioration) which are not in accordance with the change of the normal state.
Drawings
To the accomplishment of the foregoing and related ends, certain illustrative aspects are described herein in connection with the following description and the annexed drawings, which are indicative of various ways in which the principles disclosed herein may be practiced, and all aspects and equivalents thereof are intended to be within the scope of the claimed subject matter. The above and other objects, features and advantages of the present disclosure will become more apparent from the following detailed description read in conjunction with the accompanying drawings. Throughout this disclosure, like reference numerals generally refer to like parts or elements.
Fig. 1 shows a schematic view of an anomaly detection system 100 of an electrical device according to an embodiment of the invention;
FIG. 2 shows a schematic diagram of a computing device 200 according to an embodiment of the invention;
FIG. 3 illustrates a flow diagram of a method 300 of training an anomaly detection model for generating an electrical device, according to one embodiment of the present invention;
FIG. 4 shows a schematic of a multivariate time series graph according to an embodiment of the invention;
FIGS. 5A and 5B illustratively show a schematic diagram of an encoding component and a decoding component, respectively, according to one embodiment of the present invention;
fig. 6A and 6B are schematic diagrams illustrating unit structures of an encoding component and a decoding component, respectively, according to an embodiment of the present invention;
FIG. 7 shows a flow diagram of an anomaly detection method 700 for an electrical device, in accordance with one embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
Currently, the anomaly detection method commonly used in the industry mainly utilizes various control charts to monitor the device anomalies, wherein the most common and representative control chart is the EWMA control chart. Although the EWMA control chart can detect almost any magnitude of deviation in the abnormality detection process, several disadvantages remain in the abnormality detection scenario of the electrical device. First, since the principle of the EWMA is an exponential weighted moving average, the exponential weighted moving average of the EWMA recognizes an early slow changing trend as a normal trend for a slowly changing sequence, and thus the EWMA cannot recognize its abnormal behavior well for a slowly deteriorating electrical device. Secondly, the EWMA control chart has weak generalization performance in the abnormal detection scene of the electrical equipment, and when the EWMA is used for realizing the abnormal detection, different detection schemes need to be designed for different types of electrical equipment. For example, in the case of an inverter device such as an inverter cooling fan, the current waveform of the device may appear as up-and-down jumps due to frequent changes in the inverter frequency, and if the sequence is monitored directly using the EWMA control chart, a large number of false alarms may be generated due to a large number of jumps. Therefore, to apply such data to the EWMA control chart, the sequence needs to be divided into upper and lower parts for monitoring, which makes the anomaly detection process complicated. Thirdly, the EWMA control chart can only detect a single index sequence, and can not detect all index sequences in a unified way, so that only one index generates abnormity when a plurality of associated indexes are detected, and an expert is required to comprehensively judge whether the equipment is really abnormal or not by combining a plurality of indexes.
Aiming at the defects of the EWMA control chart in the scene of electrical equipment anomaly detection, the embodiment of the invention provides an anomaly detection scheme. Fig. 1 shows a schematic view of an abnormality detection system 100 of an electrical apparatus according to an embodiment of the present invention. As shown in fig. 1, the anomaly detection system 100 includes an acquisition device 110 and a computing device 200.
The collection device 110 is connected to the electrical device to collect relevant index data of the electrical device, such as a current signal, a voltage signal, and the like, but not limited thereto.
In some embodiments, the acquisition device 110 may be one or more sensors, such as a three-phase current sensor and a three-phase voltage sensor arranged in an electrical cabinet, that respectively acquire three-phase current and three-phase voltage signals when the motor is running. Wherein the motor current is sensed with current sensor (CT) clamps around a power cord located in the motor control center or any convenient cable, and the voltage is sampled using voltage clamps and leads connected to the output of the starter or other convenient point.
In other embodiments, the collecting device 110 includes a collecting station in addition to the sensors disposed in the electrical device, and after the sensors transmit the collected index data into the collecting station, the collecting station calculates other index data such as a peak factor, a voltage deviation factor, a three-phase average voltage, and the like by means of edge calculation and the like. The embodiment of the present invention is not limited with respect to the process of acquiring the index data of the device, and any known or future known manner of calculating the index data of the electrical device may be combined with the embodiment of the present invention.
The collection device 110 transmits the collected index data to the computing device 200, and the computing device 200 processes the index data to finally obtain a detection result about whether the electrical device is abnormal. Computing device 200 may be implemented as a server, e.g., an application server, a Web server, etc.; but may also be implemented as a desktop computer, a notebook computer, a processor chip, a tablet computer, etc., but is not limited thereto.
In addition, the computing device 200 may generate training samples using various index data of various types of electrical devices acquired in advance. And training by utilizing the training sample to generate an abnormal detection model so as to learn the data characteristics of different equipment types under different health conditions.
In this way, in the process of detecting a certain electrical device, the detection result of the electrical device can be output by inputting a plurality of index data collected by the collection device 110 to the abnormality detection model at the same time.
In addition, the operation state of the electrical device may change over time due to aging of the electrical device, changes in the operating environment, and the like, and therefore, the computing device 200 may also perform fine tuning on the abnormality detection model periodically to update the model parameters.
According to an embodiment of the present invention, the computing device 200 in the anomaly detection system 100 may be implemented by a computing device as described below. FIG. 2 shows a schematic diagram of a computing device 200, according to one embodiment of the invention.
As shown in FIG. 2, in a basic configuration 202, a computing device 200 typically includes a system memory 206 and one or more processors 204. A memory bus 208 may be used for communication between the processor 204 and the system memory 206.
Depending on the desired configuration, the processor 204 may be any type of processing, including but not limited to: a microprocessor (μ P), a microcontroller (μ C), a Digital Signal Processor (DSP), or any combination thereof. The processor 204 may include one or more levels of cache, such as a level one cache 210 and a level two cache 212, a processor core 214, and registers 216. Example processor cores 214 may include Arithmetic Logic Units (ALUs), Floating Point Units (FPUs), digital signal processing cores (DSP cores), or any combination thereof. The example memory controller 218 may be used with the processor 204, or in some implementations the memory controller 218 may be an internal part of the processor 204.
Depending on the desired configuration, system memory 206 may be any type of memory, including but not limited to: volatile memory (such as RAM), non-volatile memory (such as ROM, flash memory, etc.), or any combination thereof. System memory 206 may include an operating system 220, one or more applications 222, and program data 224. In some implementations, the application 222 can be arranged to execute instructions with the program data 224 by the one or more processors 204 on an operating system.
The computing device 100 also includes a storage device 132, the storage device 132 including removable storage 136 and non-removable storage 138, the removable storage 136 and the non-removable storage 138 each connected to the storage interface bus 134.
Computing device 200 may also include an interface bus 240 that facilitates communication from various interface devices (e.g., output devices 242, peripheral interfaces 244, and communication devices 246) to the basic configuration 202 via the bus/interface controller 230. The example output device 242 includes a graphics processing unit 248 and an audio processing unit 250. They may be configured to facilitate communication with various external devices, such as a display or speakers, via one or more a/V ports 252. Example peripheral interfaces 244 can include a serial interface controller 254 and a parallel interface controller 256, which can be configured to facilitate communications with external devices such as input devices (e.g., keyboard, mouse, pen, voice input device, touch input device) or other peripherals (e.g., printer, scanner, etc.) via one or more I/O ports 258. An example communication device 246 may include a network controller 260, which may be arranged to facilitate communications with one or more other computing devices 262 over a network communication link via one or more communication ports 264.
A network communication link may be one example of a communication medium. Communication media may typically be embodied by computer readable instructions, data structures, program modules, and may include any information delivery media, such as carrier waves or other transport mechanisms, in a modulated data signal. A "modulated data signal" may be a signal that has one or more of its data set or its changes made in such a manner as to encode information in the signal. By way of non-limiting example, communication media may include wired media such as a wired network or private-wired network, and various wireless media such as acoustic, Radio Frequency (RF), microwave, Infrared (IR), or other wireless media. The term computer readable media as used herein may include both storage media and communication media.
Computing device 200 may be implemented as a server, such as a file server, database server, application server, WEB server, and the like, or as a personal computer including desktop and notebook computer configurations. Of course, computing device 200 may also be implemented as part of a small-sized portable (or mobile) electronic device.
In an embodiment according to the present invention, the computing device 200 is configured to execute a method of training a generation of an abnormality detection model of an electrical device according to the present invention, and an abnormality detection method of an electrical device. Application 222 of computing device 200 includes a plurality of program instructions for performing these methods.
FIG. 3 illustrates a flow diagram of a method 300 of training an anomaly detection model for generating an electrical device, according to one embodiment of the present invention. It should be appreciated that the contents of the method 300 are complementary to the contents of the anomaly detection system 100 described above.
As shown in fig. 3, the method 300 begins at step S310.
In step S310, index data of the electrical device is acquired.
According to an embodiment of the invention, the metric data is transmitted by the acquisition device 110 to the computing device 120. The index data may be data obtained directly via sensors, such as current, voltage; or may be data obtained by edge calculation by the acquisition device, such as a peak factor and a voltage deviation factor, which is not limited by the embodiment of the present invention.
In some embodiments, the metric data includes at least one or more of the following: current signal, voltage signal, peak factor, voltage deviation factor, three-phase average voltage. Further, the metric data may also include one or more of the following: the acceleration total value, the high acceleration total value, the harmonic total energy, the harmonic number, the 1-3 frequency multiplication total energy, the 4-n frequency multiplication total energy and the like.
Subsequently, in step S320, a multivariate time series data set is constructed using the index data.
In some embodiments, for a certain electrical device, the corresponding index data is acquired, and then, the acquired index data is used to generate a multivariate time sequence. When the number of electrical devices increases, the generated multivariate time series constitutes a multivariate time series data set.
In the process of training the model, in order to increase the generalization capability of the model, according to the embodiment of the invention, index data of a large number of electrical devices of different device types are acquired, and a multivariate time series data set is formed to be used as a training sample.
FIG. 4 shows a schematic diagram of a multivariate time series graph according to one embodiment of the invention. As shown in fig. 4, this is a multivariate time series chart consisting of 6 index data. The abscissa represents time, and the ordinate represents a value corresponding to each index. The six indexes respectively represent: the acceleration total value, the high acceleration total value, the harmonic total energy, the harmonic number, the frequency doubling total energy of 1-3 and the frequency doubling total energy of 4-n.
It should be understood that, in the acquisition stage of the index data, the multivariate time series is usually represented by a waveform diagram, such as a multivariate time series diagram, and needs to be converted into an N-dimensional array when the model training is subsequently performed by using the multivariate time series. The six-dimensional time sequence diagram shown in fig. 4 can be converted into a six-dimensional array, which is not described herein again.
Subsequently, in step S330, the multivariate time series data set is preprocessed to obtain a preprocessed multivariate time series data set.
The sensor is damaged, the installation is incorrect, the operating environment of the equipment is changed, and the like, which can cause error signals in the collected data, and the error signals can not correctly describe the operating condition of the equipment and often bring false alarm results for abnormal detection. Therefore, after the multi-element time sequence data set is obtained, preprocessing is performed on the multi-element time sequence data set, namely, the error signals are removed, and the preprocessed multi-element time sequence data set is obtained; and then training the model by utilizing the preprocessed multivariate time series dataset.
According to one embodiment, the preprocessed multivariate time series dataset is obtained by identifying and screening out false signals. There are generally two ways to determine false signals: firstly, based on the indicators such as skewness, kurtosis, and total value (these indicator data can also be obtained in step S210, which is not described herein), different determination rules or preset threshold values are set for different types of error signals; secondly, whether a section of signal is a false signal is judged through a deep learning method based on the time-frequency domain image. In one embodiment according to the present invention, the above two ways are combined, that is, combining a preset threshold value and a time-domain image, and a false signal is identified from the multivariate time-series data set. Then, the identified error signal is filtered from the multivariate time series data set, and a preprocessed multivariate time series data set is obtained.
According to another embodiment, in addition to identifying and screening out the error signal, the start-stop signal needs to be determined, so as to obtain the multivariate time series (i.e. index data of the electrical equipment in operation) of the electrical equipment in the start-up state as the preprocessed multivariate time series data set. In one embodiment, the threshold is determined based on a clustering algorithm (e.g., K-means) in combination with expert-defined auxiliary indicators and a threshold correction strategy, and the start-up signal and the stop-down signal can be identified according to the threshold. And determining the start-up state of the electrical equipment based on the start-up signal and the stop signal. And finally, screening the multivariate time series in the startup state from the multivariate time series data set to serve as the preprocessed multivariate time series data set.
Subsequently, in step S340, an abnormality detection model is trained using the preprocessed multivariate time series dataset.
According to an embodiment of the invention, the anomaly detection model is a variational bayes-based generative model. In one embodiment, the anomaly detection model is generated by coupling an encoding component and a decoding component. Wherein, the coding component comprises a long-short term memory artificial neural network (LSTM) and a full connection layer, and the decoding component also comprises a long-short term memory artificial neural network (LSTM) and a full connection layer.
Fig. 5A and 5B exemplarily show schematic diagrams of an encoding component and a decoding component, respectively.
The processed multivariate time sequence data set xtInputting coding assembly, processing by LSTM and full connection layer, extracting characteristic information, namely hidden variable zt. The decoding component obtaining the implicit variable ztThen, the feature information z is processedtAs input to the decoding component, a multi-component time series data set, denoted x, is reconstructed, also via the LSTM and the full connection layert'. In the whole training process, a multi-element time sequence data set x is processedtAnd a reconstructed multivariate time series data set xtAnd', calculating a reconstruction error, adjusting network parameters of the coding component and the decoding component until the reconstruction error is minimum, finishing training and generating an abnormal detection model.
In a preferred embodiment, the reconstruction errorsDifference (i.e., loss function) is computed for the reconstructed multivariate time series dataset xt' and a second index for the distribution of the characteristic information. Wherein the first index is a time-series data set x composed of reconstructed multiple elementst' and feature information ztThe second index is used for measuring the characteristic information ztKL divergence of the distribution of. The reconstruction error can be expressed as the following equation (1) (where, the sequence length of the L input model is expressed):
wherein the first part log (p)θ(xt-T:t'|zt-T:t) Is expressed by the feature information z (i.e., the first index)tReconstruction xt' the meaning of this term is indicated in the feature information ztMaximize the reconstruction probability p on the premise ofθ(xt-T:t'|zt-T:t) I.e. letting the input and output be maximally similar, second part
(i.e., the second indicator) represents the KL divergence, and is a measure of the degree of similarity between the two distributions. Since the anomaly detection model is improved based on VAE, an assumption premise exists in the VAE model that the extracted feature information z is usedtFollowing a general normal distribution, the meaning of the second component is expressed, letting the model-based input xtGenerated ztDistribution of (2)
Approximating the assumed distribution
Further, fig. 6A and 6B are schematic diagrams illustrating the unit structures of an encoding component and a decoding component, respectively, according to an embodiment of the present invention. One unit represents one of the encoding components (or decoding components)xtAfter encoding, a corresponding z is generatedt(or by z)tDecoding produces xt') network element structure.
The following provides a brief description of the principles and methods to which the two network components apply.
First, the encoding and decoding components use the complex time dependencies between x-space acquisition multivariate observations of the LSTM, where the characteristics of the LSTM network itself, the hidden variable e in the LSTM network, are mainly exploitedtCan capture xtComplex time information of a long time before.
Second, the observations (i.e., the input observations in x-space) are mapped to random variables (i.e., z-space) using the commonly used representation learning variational algorithm VAE.
Third, to explicitly model the temporal dependencies between random variables in hidden space, the network applies a random variable join technique (linear gaussian state space model joins random variables and concatenation of random variables with LSTM hidden variables).
For the coding component, at time t, the observed value x will be inputtAnd et-1(hidden variable calculated in LSTM at time t-1) is sent to LSTM network to generate next hidden variable et(as in equation (2)). Then etAnd zt-1Connecting into full connection layer to generate mean value
And standard deviation of
(as in equations (3) and (4)). Due to zt-1And etThe coding component takes advantage of the cyclic recursive nature of LSTM to make the z-space variable time dependent, which needs to be merged and input to the fully-connected layer together. Wherein the variables in the coding component are calculated as follows:
wherein r in the formula (2)t eAnd
respectively representing the reset gate and refresh gate mechanisms in the LSTM, rt eDeciding how to associate a new input with the previous memory (e)t-1) The combination of the components is carried out,
indicating that the update gate decides how much past information the sequence needs to be preserved. Wherein r ist eAnd
are all represented by xtInputting LSTM, multiplying by weight w, adding bias term b, and obtaining by a sigmoid activation function. The specific calculation mode is shown in formula (5) and formula (6):
since the idea of VAE is to make a look by assuming p (Z | X)k) The posterior probability distribution is a normal distribution, and a Z is randomly sampled from the normal distributionk(since additivity of normal distribution makes it easy to derive p (Z) subject to normal distribution), by using ZkReconstruction of Xk. Therefore, assuming a normal distribution, the mean μ and variance σ need to be determined2In conventional VAE, neural nets are usedThe method of the complex approximates the two parameters. However, in the range from μ and σ2To ZkThe process of (1) is a sampling process, which cannot be derived, so that the parameters cannot be optimized by using a gradient descent mode in the process of reverse retransmission, and a heavy parameter skill is adopted, namely Z is enabledkMu + epsilon sigma, where epsilon is normally distributed, ensures that p (z) is normally distributed and that the entire network can be optimally solved by back propagation. The above formula (3) and formula (4) are just to calculate Zkμ + ε · σ and ε · μ.
For decoding components, use is made of ztReconstruction xt' the network structure and manner is similar to the encoding components. The temporal dependency between z is achieved in the decoding component by connecting the hidden variables z using a linear gaussian state space modeling method (as in equation (7)). O isθAnd TθRespectively, observation and transfer matrix, vtAnd εtRespectively, the transfer and the observed noise. The calculation and data transmission of the decoding component are similar to those of the encoding component, and the calculation mode of each key node is shown in formula (8) to formula (12):
zt＝Oθ(Tθzt-1+vt)+εt (7)
wherein r in the formula (8)t dAnd
and r in formula (2)t eAnd
similarly, the reset gate and update gate mechanisms in the LSTM are shown separately. The two variables are calculated in the manner shown in formula (11) and formula (12)
Thus, the anomaly detection model according to the embodiment of the invention is trained. When the anomaly detection model is constructed and trained, the input of the model is a multivariate time sequence (namely a plurality of index data), and the input is defined as the multivariate time sequence, so that the anomaly detection efficiency can be greatly improved, and the defect that different indexes need to be subjected to anomaly detection separately when a statistical process control method (EWMA) is used in the past is overcome.
It should be noted that, only an example of the abnormality detection model according to an embodiment of the present invention is shown here. In a network architecture, the anomaly detection model may also be constructed using a generative countermeasure network instead of a VAE, which is not limited by the embodiments of the present invention. Meanwhile, in the encoder and decoder, GRU, RNN, Transformers, and the like may be used instead of LSTM, and the same expression can be achieved.
In addition, as can be seen from the network unit structures of the encoding component and the decoding component, the anomaly detection model well integrates the stochastic cyclic neural network method and the VAE. On one hand, the time dependence of the multivariate time series and the data characteristics in time can be well learned by combining the recurrent neural network. On the other hand, the characteristic information of the multi-element time series except the time and the trend can be well extracted by applying the VAE idea. By utilizing the two methods, the anomaly detection model can recognize the characteristics of time dependence, trend and the like, and can learn the image characteristics beyond the multivariate time series trend which cannot be recognized by a control chart such as EWMA (incremental variable moving object), so that the anomaly detection model can well recognize the anomaly of the slowly-degraded electrical equipment.
In addition, the anomaly detection model can effectively improve the generalization performance of the model. On one hand, the model is allowed to contain a large number of parameters by constructing the deep learning model, a large number of training samples are used for fitting the model, and the large-capacity data set means that the model can learn more data characteristics and patterns, so that normal signals can be better recognized, and abnormal signals can be judged. On the other hand, the combination with the VAE model also means that the model is able to learn data samples that carry some noise.
FIG. 7 shows a flow diagram of an anomaly detection method 700 for an electrical device, in accordance with one embodiment of the present invention. As shown in fig. 7, the method 700 begins at step S710.
In step S710, index data of the electrical device is acquired.
Subsequently, in step S720, a multivariate time series is constructed using the index data.
For the related description of the index data and the multivariate time series, reference may be made to the related contents above, and further description is omitted here.
Subsequently, in step S730, the multivariate time series is input to the anomaly detection model and processed to generate an anomaly score series including a plurality of anomaly scores. Wherein the anomaly detection model is obtained by performing the method 300.
As described above, the N-ary time sequence is converted into an N-dimensional array, and the N-dimensional array is input into the anomaly detection model, and after the anomaly detection model is processed, whether the observation sequence at time step t is abnormal is judged; then, for each time step in the N-ary time series, a value as to whether the observation series for each time step is abnormal is output.
Assuming that 4 index data are obtained, which are respectively denoted as a, b, c, and d, the corresponding multivariate time series can be expressed as the following array form:
accordingly, the observation sequence of time step t is: { at,bt,ct,dtIn which t ∈ [1, 2.,. n.)]. After the abnormity detection model is processed, outputting an abnormity score S corresponding to the time step tt。
Aiming at each time step in the multivariate time sequence input into the anomaly detection model, the finally output anomaly scores also form an anomaly score sequence, and the sequence is recorded as { S1,S2,...,Sn}。
Referring to the description of the anomaly detection model in the method 300, the sequence of the anomaly detection model is x when the multivariate time sequence is processedt-T:tI.e. order xtAnd T successive observations before it as input to reconstruct xt. Reuse of conditional log of probability (p)θ(xt|zt-T:t) To evaluate the level of reconstruction of the model. The evaluation index is also used as an abnormality score S representing whether the electrical equipment is abnormal or nottI.e. St＝log(pθ(xt|zt-T:t))。
According to embodiments of the present invention, a high anomaly score means that x is inputtCan be well reconstructed by the model, because the hidden variable z learned by the original model can be regarded as the sequence characteristic of the normal sequence, so that x can be regarded astSimilar to the normal sequence, i.e., an observation obeys the characteristic pattern of the normal sequence, then the observation can be reconstructed well at a high confidence level. Conversely, the smaller the anomaly score, the less likely the observation is to be reconstructed, and the greater the probability that the segment of the observation is considered anomalous.
Therefore, in the subsequent step S740, a threshold value regarding the score abnormality is adaptively determined based on the output abnormality score sequence.
According to an embodiment of the present invention, the process of determining the threshold for score anomalies is as follows. First, a generalized pareto distribution (GDP) is fitted using POT theorem in Extreme Value Theory (EVT). Then, using maximum likelihood estimation, parameters in the generalized pareto distribution are estimated. Finally, a threshold for point anomaly is determined based on the estimated parameters.
In one embodiment, the fitted distribution can be represented by equation (13):
in the formula (13), the reaction mixture is,
for cumulative probability distribution function, s means
S is an anomaly score sequence, and the parameters β and γ represent the shape parameter and scale parameter of the generalized pareto distribution, respectively. Then, a probability density function can be obtained by accumulating probability distribution functions, and parameters beta and gamma can be estimated by maximum likelihood estimation.
After the parameters are estimated, the threshold is calculated using the estimated parameters β and γ. In one embodiment, the threshold is determined using equation (14):
where th denotes an initialization threshold, which is typically a sufficiently high threshold, parameter, set empirically based on the resulting anomaly score
And
respectively representing the shape parameter and scale parameter of the estimated generalized pareto distribution, wherein q is an abnormal score sequence S obtained by observation<th expected probability, N 'denotes the observed number, N'thIndicates that S is satisfiedi< th.
Subsequently in step S750, it is determined whether the electrical device is abnormal based on the abnormality score and the threshold value.
According to the embodiment of the invention, if the abnormal score is greater than the threshold value, the electrical equipment is determined to be normal in the corresponding time period; and if the abnormal score is not larger than the threshold value, determining that the electrical equipment is abnormal in the corresponding time period.
According to the anomaly detection method, after the anomaly score is output by the anomaly detection model, whether the electrical equipment is abnormal or not is not judged directly according to the preset threshold value, but the threshold value aiming at the anomaly score detected this time can be determined in a self-adaptive manner based on a series of output anomaly scores (namely, an anomaly score sequence), so that the accuracy of the anomaly detection model on anomaly detection is effectively improved.
In summary, the anomaly detection scheme according to the present invention has the following advantages: firstly, on the basis of a random cyclic neural network, a variational self-encoder (VAE) is combined, the characteristics of normal voltage and current time domain signals can be well extracted by an unsupervised method, and on the basis, the data characteristics of different equipment types under normal conditions can be well learned through a model trained by a large amount of normal data (namely, a preprocessed multivariate time series data set) of different equipment, so that the abnormal point with slow degradation can be detected; secondly, after the anomaly score is output through the anomaly detection model, a threshold value related to anomaly detection is calculated in a self-adaptive mode through an anomaly score sequence, namely the threshold value is changed for each anomaly detection instead of a fixed value, so that an anomaly detection result is more accurate; thirdly, a plurality of index data are simultaneously input into the abnormality detection model, and finally, the abnormality score of the electrical equipment is output without calculating each index once.
In the description provided herein, numerous specific details are set forth. It is understood, however, that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been shown in detail in order not to obscure an understanding of this description.
Similarly, it should be appreciated that in the foregoing description of exemplary embodiments of the invention, various features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure and aiding in the understanding of one or more of the various inventive aspects. However, the disclosed method should not be interpreted as reflecting an intention that: that the invention as claimed requires more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive aspects lie in less than all features of a single foregoing disclosed embodiment. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment of this invention.
Those skilled in the art will appreciate that the modules or units or components of the devices in the examples disclosed herein may be arranged in a device as described in this embodiment or alternatively may be located in one or more devices different from the devices in this example. The modules in the foregoing examples may be combined into one module or may be further divided into multiple sub-modules.
Those skilled in the art will appreciate that the modules in the device in an embodiment may be adaptively changed and disposed in one or more devices different from the embodiment. The modules or units or components of the embodiments may be combined into one module or unit or component, and furthermore they may be divided into a plurality of sub-modules or sub-units or sub-components. All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and all of the processes or elements of any method or apparatus so disclosed, may be combined in any combination, except combinations where at least some of such features and/or processes or elements are mutually exclusive. Each feature disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise.
Furthermore, those skilled in the art will appreciate that while some embodiments described herein include some features included in other embodiments, rather than other features, combinations of features of different embodiments are meant to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments may be used in any combination.
The invention also discloses:
a8, the method as claimed in any of A1-7, wherein index data comprises at least one or more of: current signal, voltage signal, peak factor, voltage deviation factor, three-phase average voltage.
A9, the method of any one of A1-8, further comprising the steps of: and updating parameters of the abnormality detection model periodically.
B11, the method of B10, wherein the step of determining whether the electrical device is abnormal based on the abnormality score and the threshold value comprises: if the abnormal value is larger than the threshold value, determining that the electrical equipment is normal in the corresponding time period; and if the abnormal score is not larger than the threshold value, determining that the electrical equipment is abnormal in the corresponding time period.
B12, the method of B10 or 11, wherein the step of determining a threshold for scoring anomalies based on anomaly score sequences comprises: fitting generalized pareto distribution by using an extreme value theory; estimating parameters in generalized pareto distribution by using maximum likelihood estimation; based on the estimated parameters, a threshold is determined.
The various techniques described herein may be implemented in connection with hardware or software or, alternatively, with a combination of both. Thus, the methods and apparatus of the present invention, or certain aspects or portions thereof, may take the form of program code (i.e., instructions) embodied in tangible media, such as floppy diskettes, CD-ROMs, hard drives, or any other machine-readable storage medium, wherein, when the program is loaded into and executed by a machine, such as a computer, the machine becomes an apparatus for practicing the invention.
In the case of program code execution on programmable computers, the computing device will generally include a processor, a storage medium readable by the processor (including volatile and non-volatile memory and/or storage elements), at least one input device, and at least one output device. Wherein the memory is configured to store program code; the processor is configured to perform the method of the invention according to instructions in said program code stored in the memory.
Furthermore, some of the described embodiments are described herein as a method or combination of method elements that can be performed by a processor of a computer system or by other means of performing the described functions. A processor having the necessary instructions for carrying out the method or method elements thus forms a means for carrying out the method or method elements. Further, the elements of the apparatus embodiments described herein are examples of the following apparatus: the apparatus is used to implement the functions performed by the elements for the purpose of carrying out the invention.
As used herein, unless otherwise specified the use of the ordinal adjectives "first", "second", "third", etc., to describe a common object, merely indicate that different instances of like objects are being referred to, and are not intended to imply that the objects so described must be in a given sequence, either temporally, spatially, in ranking, or in any other manner.
While the invention has been described with respect to a limited number of embodiments, those skilled in the art, having benefit of this description, will appreciate that other embodiments can be devised which do not depart from the scope of the invention as described herein. Furthermore, it should be noted that the language used in the specification has been principally selected for readability and instructional purposes, and may not have been selected to delineate or circumscribe the inventive subject matter. Accordingly, many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the appended claims. The present invention has been disclosed in an illustrative rather than a restrictive sense, and the scope of the present invention is defined by the appended claims.
Claims (10)
1. A method of training a generation of an anomaly detection model of an electrical device, the method comprising the steps of:
acquiring index data of the electrical equipment;
constructing a multivariate time sequence data set by using the index data;
preprocessing the multivariate time sequence data set to obtain a preprocessed multivariate time sequence data set;
and training an abnormality detection model by using the preprocessed multivariate time sequence dataset.
2. The method of claim 1, wherein the anomaly detection model is a variational bayes-based generative model.
3. The method of claim 1 or 2, wherein the anomaly detection model is generated by a coupling of an encoding component and a decoding component.
4. The method of claim 3, wherein the step of training an anomaly detection model using the preprocessed multivariate time series dataset comprises:
extracting characteristic information from the processed multivariate time series data set through the coding component;
the characteristic information is used for reconstructing a multivariate time sequence data set through the decoding component;
and calculating a reconstruction error and adjusting network parameters of the coding assembly and the decoding assembly based on the processed multivariate time series data set and the reconstructed multivariate time series data set until the reconstruction error is minimum, finishing training and generating an abnormal detection model.
5. The method of claim 4, wherein the encoding component and the decoding component each comprise a long-short term memory artificial neural network.
6. The method of claim 4 or 5,
the reconstruction error is obtained by calculating a first index for the reconstructed multivariate time series data set and a second index for the characteristic information distribution, wherein,
the first index is a conditional probability composed of the reconstructed multivariate time series data set and the characteristic information; the second indicator is a KL divergence used to measure the distribution of the characteristic information.
7. The method of any one of claims 1-6, wherein the preprocessing the multivariate time series data set to obtain a preprocessed multivariate time series data set comprises:
identifying error signals from the multivariate time sequence data set by combining a preset threshold value and a time-frequency domain image;
and screening the error signals from the multivariate time sequence data set to obtain a preprocessed multivariate time sequence data set.
8. An abnormality detection method for an electrical device, comprising the steps of:
acquiring index data of the electrical equipment;
constructing a multivariate time sequence by using the index data;
inputting the multivariate time sequence into an anomaly detection model for processing to generate an anomaly score sequence comprising a plurality of anomaly scores;
determining a threshold for scoring anomalies based on the anomaly score sequence;
determining whether the electrical device is abnormal based on the abnormality score and the threshold value,
wherein the anomaly detection model is obtained by performing the method of any one of claims 1-7.
9. A computing device, comprising:
one or more processors;
a memory; and
one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs comprising instructions for performing any of the methods of claims 1-8.
10. A computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a computing device, cause the computing device to perform any of the methods of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110034065.8A CN112766342A (en) | 2021-01-12 | 2021-01-12 | Abnormity detection method for electrical equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110034065.8A CN112766342A (en) | 2021-01-12 | 2021-01-12 | Abnormity detection method for electrical equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112766342A true CN112766342A (en) | 2021-05-07 |
Family
ID=75701549
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110034065.8A Pending CN112766342A (en) | 2021-01-12 | 2021-01-12 | Abnormity detection method for electrical equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112766342A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113221759A (en) * | 2021-05-17 | 2021-08-06 | 北京文安智能技术股份有限公司 | Road scattering identification method and device based on anomaly detection model |
| CN113255835A (en) * | 2021-06-28 | 2021-08-13 | 国能大渡河大数据服务有限公司 | Hydropower station pump equipment anomaly detection method |
| CN113312413A (en) * | 2021-07-29 | 2021-08-27 | 国网浙江省电力有限公司营销服务中心 | Enterprise carbon emission abnormity monitoring method based on electric power big data |
| CN113435258A (en) * | 2021-06-06 | 2021-09-24 | 西安电子科技大学 | Rotor system abnormity intelligent detection method and system, computer equipment and terminal |
| CN113656461A (en) * | 2021-08-17 | 2021-11-16 | 北京百度网讯科技有限公司 | Data processing method, device, electronic equipment and medium |
| CN113780238A (en) * | 2021-09-27 | 2021-12-10 | 京东科技信息技术有限公司 | Multi-index time sequence signal abnormity detection method and device and electronic equipment |
| CN114676593A (en) * | 2022-04-18 | 2022-06-28 | 无锡物联网创新中心有限公司 | Abnormity detection method of textile equipment and related device |
| CN115001997A (en) * | 2022-04-11 | 2022-09-02 | 北京邮电大学 | Extreme value theory-based smart city network equipment performance abnormity threshold evaluation method |
| CN115412455A (en) * | 2022-07-28 | 2022-11-29 | 南京航空航天大学 | Server multi-performance index abnormity detection method and device based on time sequence |
| CN115456097A (en) * | 2022-09-22 | 2022-12-09 | 国网四川省电力公司自贡供电公司 | Power utilization detection method and detection terminal suitable for high-power-supply low-count special transformer users |
| CN116146515A (en) * | 2022-11-15 | 2023-05-23 | 安徽智质工程技术有限公司 | Abnormality detection method for centrifugal fan equipment of cement plant |
| CN117349478A (en) * | 2023-10-08 | 2024-01-05 | 国网江苏省电力有限公司经济技术研究院 | Resource data reconstruction integration system based on digital transformation enterprise |
| CN117688499A (en) * | 2024-01-31 | 2024-03-12 | 成方金融科技有限公司 | Multi-index anomaly detection method and device, electronic equipment and storage medium |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111178456A (en) * | 2020-01-15 | 2020-05-19 | 腾讯科技（深圳）有限公司 | Abnormal index detection method and device, computer equipment and storage medium |
-
2021
- 2021-01-12 CN CN202110034065.8A patent/CN112766342A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111178456A (en) * | 2020-01-15 | 2020-05-19 | 腾讯科技（深圳）有限公司 | Abnormal index detection method and device, computer equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| YA SU等: ""Robust Anomaly Detection for Multivariate Time Series through Stochastic Recurrent Neural Network"", 《SIGKDD EXPLORATIONS》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113221759A (en) * | 2021-05-17 | 2021-08-06 | 北京文安智能技术股份有限公司 | Road scattering identification method and device based on anomaly detection model |
| CN113435258A (en) * | 2021-06-06 | 2021-09-24 | 西安电子科技大学 | Rotor system abnormity intelligent detection method and system, computer equipment and terminal |
| CN113435258B (en) * | 2021-06-06 | 2024-04-23 | 西安电子科技大学 | Rotor system abnormality intelligent detection method, system, computer equipment and terminal |
| CN113255835A (en) * | 2021-06-28 | 2021-08-13 | 国能大渡河大数据服务有限公司 | Hydropower station pump equipment anomaly detection method |
| CN113312413A (en) * | 2021-07-29 | 2021-08-27 | 国网浙江省电力有限公司营销服务中心 | Enterprise carbon emission abnormity monitoring method based on electric power big data |
| CN113312413B (en) * | 2021-07-29 | 2021-10-29 | 国网浙江省电力有限公司营销服务中心 | Enterprise carbon emission abnormity monitoring method based on electric power big data |
| CN113656461A (en) * | 2021-08-17 | 2021-11-16 | 北京百度网讯科技有限公司 | Data processing method, device, electronic equipment and medium |
| CN113656461B (en) * | 2021-08-17 | 2024-03-19 | 北京百度网讯科技有限公司 | Data processing method, device, electronic equipment and medium |
| CN113780238A (en) * | 2021-09-27 | 2021-12-10 | 京东科技信息技术有限公司 | Multi-index time sequence signal abnormity detection method and device and electronic equipment |
| CN113780238B (en) * | 2021-09-27 | 2024-04-05 | 京东科技信息技术有限公司 | Abnormality detection method and device for multi-index time sequence signal and electronic equipment |
| CN115001997B (en) * | 2022-04-11 | 2024-02-09 | 北京邮电大学 | Extreme value theory-based smart city network equipment performance abnormal threshold evaluation method |
| CN115001997A (en) * | 2022-04-11 | 2022-09-02 | 北京邮电大学 | Extreme value theory-based smart city network equipment performance abnormity threshold evaluation method |
| CN114676593B (en) * | 2022-04-18 | 2023-05-23 | 无锡物联网创新中心有限公司 | Abnormality detection method and related device for textile equipment |
| CN114676593A (en) * | 2022-04-18 | 2022-06-28 | 无锡物联网创新中心有限公司 | Abnormity detection method of textile equipment and related device |
| CN115412455B (en) * | 2022-07-28 | 2023-12-19 | 南京航空航天大学 | Method and device for detecting abnormality of multiple performance indexes of server based on time sequence |
| CN115412455A (en) * | 2022-07-28 | 2022-11-29 | 南京航空航天大学 | Server multi-performance index abnormity detection method and device based on time sequence |
| CN115456097A (en) * | 2022-09-22 | 2022-12-09 | 国网四川省电力公司自贡供电公司 | Power utilization detection method and detection terminal suitable for high-power-supply low-count special transformer users |
| CN116146515A (en) * | 2022-11-15 | 2023-05-23 | 安徽智质工程技术有限公司 | Abnormality detection method for centrifugal fan equipment of cement plant |
| CN117349478A (en) * | 2023-10-08 | 2024-01-05 | 国网江苏省电力有限公司经济技术研究院 | Resource data reconstruction integration system based on digital transformation enterprise |
| CN117349478B (en) * | 2023-10-08 | 2024-05-24 | 国网江苏省电力有限公司经济技术研究院 | Resource data reconstruction integration system based on digital transformation enterprise |
| CN117688499A (en) * | 2024-01-31 | 2024-03-12 | 成方金融科技有限公司 | Multi-index anomaly detection method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112766342A (en) | Abnormity detection method for electrical equipment | |
| Wang et al. | Multilevel information fusion for induction motor fault diagnosis | |
| JP7167084B2 (en) | Anomaly detection system, anomaly detection method, anomaly detection program, and learned model generation method | |
| CN111459700B (en) | Equipment fault diagnosis method, diagnosis device, diagnosis equipment and storage medium | |
| CN111914873A (en) | Two-stage cloud server unsupervised anomaly prediction method | |
| CN112416643A (en) | Unsupervised anomaly detection method and unsupervised anomaly detection device | |
| CN116226646B (en) | Method, system, equipment and medium for predicting health state and residual life of bearing | |
| CN112416662A (en) | Multi-time series data anomaly detection method and device | |
| CN114037478A (en) | Advertisement abnormal flow detection method and system, electronic equipment and readable storage medium | |
| CN117131110B (en) | Method and system for monitoring dielectric loss of capacitive equipment based on correlation analysis | |
| CN117036732A (en) | Electromechanical equipment detection system, method and equipment based on fusion model | |
| CN116720983A (en) | Power supply equipment abnormality detection method and system based on big data analysis | |
| CN115186772B (en) | Method, device and equipment for detecting partial discharge of power equipment | |
| CN117074852A (en) | Power distribution network electric energy monitoring and early warning management method and system | |
| Xu et al. | Deep transfer learning remaining useful life prediction of different bearings | |
| CN114580472A (en) | Large-scale equipment fault prediction method with repeated cause and effect and attention in industrial internet | |
| Scheffel et al. | Data confidence applied to wind turbine power curves | |
| KR102212022B1 (en) | Method of automatically determining condition of hydro turbine in hydroelectric power plant and system for the same | |
| CN117723917B (en) | Monitoring application method based on optical fiber extrinsic Fabry-Perot type ultrasonic sensor | |
| CN116643170B (en) | Motor shafting vibration testing method and device and computer equipment | |
| CN117688496B (en) | Abnormality diagnosis method, system and equipment for satellite telemetry multidimensional time sequence data | |
| CN117407784B (en) | Sensor data abnormality-oriented intelligent fault diagnosis method and system for rotary machine | |
| Liu et al. | Bearing fault diagnosis analysis based on improved adaptive search algorithms and SVMs | |
| CN118226313A (en) | Storage battery remaining life assessment method and system based on data analysis | |
| CN117649015A (en) | Power grid time sequence data quality assessment method, system, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210507 |
|
| RJ01 | Rejection of invention patent application after publication |