CN115859202A - Abnormal detection method and device under non-stationary time sequence data flow field scene - Google Patents
Abnormal detection method and device under non-stationary time sequence data flow field scene Download PDFInfo
- Publication number
- CN115859202A CN115859202A CN202211484703.7A CN202211484703A CN115859202A CN 115859202 A CN115859202 A CN 115859202A CN 202211484703 A CN202211484703 A CN 202211484703A CN 115859202 A CN115859202 A CN 115859202A
- Authority
- CN
- China
- Prior art keywords
- data
- time
- gate
- flow field
- expression
- 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.)
- Granted
Links
Images
Landscapes
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The invention discloses an anomaly detection method and device under the scene of a non-stationary time sequence data flow field. The model optimized by the method improves the modeling precision and ensures the global optimization of the solution scheme on the premise of ensuring the rationality and the basic accuracy of the method. The invention can effectively distinguish the non-stationarity of the data and the data abnormality, and greatly reduce the false detection rate and the omission rate of abnormality detection under the scene of a non-stationary time sequence data flow field.
Description
Technical Field
The invention relates to the field of data mining, in particular to the field of time sequence data abnormity detection, and specifically relates to an abnormity detection method and device under the scene of a non-stable time sequence data flow field.
Background
The time-series data are data that change with time and are recorded chronologically by a uniform index. Anomaly detection is a critical part of the task of time series data analysis. According to the change rule of data, time series data can be divided into stationary sequences and non-stationary sequences:
1) A smooth sequence, i.e. a sequence with substantially no trend changes. The observed values of such sequences fluctuate substantially within a fixed interval. Although the fluctuation degrees of different time periods are different, the overall trend is unchanged, and the data value does not exceed the extreme value of the interval;
2) Non-stationary sequences, i.e., sequences containing significant trending, seasonal, or periodic characteristics. It may contain only one time series component or a combination of several time series components.
The non-stationary characteristic of time series data is also referred to in the data mining art as data concept drift. In a real-world scenario, many time-series data are accompanied by conceptual drift, i.e., the data stream is non-stationary and may change regularly over time. For example, in a bridge data acquisition scenario, the ambient temperature may change significantly over time, which in turn may cause changes in the properties of the bridge structure, such that the data acquired by some sensors is extremely unstable. The data are not outlier data generated by bridge structure abnormity or sensor abnormity, but normal change caused by environment, but the existing model can not discriminate the change of the bridge data and the abnormal data. Therefore, for data containing a concept drift phenomenon, the data value of the current time period may be abnormal even if it occurs in the previous time period; also, it may be normal even if the data value of the current time period is significantly different from the previous data value.
In recent years, due to the influence of deep learning, a new framework based on a variational self-encoder is generated by an anomaly detection task, the framework takes a reconstruction error as a standard, the theoretical rationality and the accuracy of detection are ensured, and the application scene is wide. However, the currently existing models for time series anomaly detection tasks cannot distinguish conceptual drift of data from data anomalies. The common time series model can confuse abnormal data with normal data of different modes, and a new data mode generated by concept drift is regarded as abnormal data, so that the performance of the time series abnormal detection model is usually reduced remarkably. Therefore, it is necessary to improve the existing time series data modeling method to make it sense and adapt to the non-stationarity of the data.
Disclosure of Invention
The invention provides an anomaly detection method and device based on an improved long-time and short-time memory network under the scene, which effectively distinguish the non-stationarity of time sequence data from data anomalies and reduce the false detection rate and the missing detection rate of anomaly detection under the scene of a non-stationary time sequence data flow field.
The technical scheme adopted by the invention for solving the technical problem is as follows: an abnormality detection method under a non-stationary time series data flow field scene comprises the following steps:
(1) Data preprocessing: acquiring data from m sensors for monitoring the health of a bridge structure in real time, integrating and averaging the sensor data according to the sampling frequency of the sensors, transversely splicing the m sensor data, and combining the m sensor data into n-dimensional time sequence data;
(2) Inputting the obtained n-dimensional time sequence data into an encoder part of a variational self-encoder, wherein the encoder is structurally characterized by improving a long-time memory network of a hidden vector and learning the time sequence and the non-stationarity of the data; the implicit vector acquisition mode is as follows: mapping the hidden state by two different custom matrixes respectively to obtain the expression of a mean value and a variance, wherein the variance part is obtained by processing through a softplus function, all the custom matrixes are obtained by random initialization through an Xavier algorithm, and the weight is continuously adjusted in the training process; obtaining the expression of Gaussian distribution of data at the current moment through the mean value and the variance, calculating KL distance with the Gaussian distribution at the previous moment to obtain the quantization difference of the data distribution of the two parts, and splicing the quantization result with a hidden state to obtain a hidden vector; the expression is as follows:
z t ＝[h t ,D(N(W μ ·h t ,softplus(W σ ·h t ))||N(W μ ·h t-1 ,softplus(W σ ·h t-1 )))]
wherein h is t Is an output value at the present time, W μ 、W σ Is a custom matrix of mean and variance, respectively, h t-1 Softplus is the output value at the last moment, and is the activation function;
(3) To conceal the output of the encoder by the vector z t Decoding as input to a decoder; the structure of the decoder is consistent with that of the encoder, but a layer of fully-connected neural network is added at the end of the decoder, and the decoding result is mapped back to a data space;
(4) And performing reconstruction error calculation on the output result of the decoder and the original input, wherein the error quantization result is normalized, and the abnormal data which is larger than the user-defined threshold value is obtained, and the normal data is obtained.
Further, in the step (1), the data of each minute is averaged with the time unit of one minute as the minimum, and the average value of the sampled data of each minute is used as the data record of the current time, so as to process the missing value of the sensor sample.
Further, in the step (2), a forgetting gate f in the long-time memory network t The expression of (c) is as follows:
f t ＝sigmoid(W f ·[z t-1 ,x t ]+b f )
wherein z is t-1 For the implicit vector, x, obtained in the preceding period t As data of the current time, W f Custom matrix for forgetting to gate, b f Sigmoid is an activation function for the offset of a forgetting gate.
Further, in step (2), an input gate i in the long-short-term memory network t The obtaining steps are consistent with those of a forgetting gate, but the self-defined matrix and the offset are different; the expression is as follows:
i t ＝sigmoid(W i ·[z t-1 ,x t ]+b i )
wherein, W i For custom matrices of input gates, b i Is the offset of the input gate.
Further, in step (2), the intermediate state C 'of the cells in the long-short time memory network' t The obtaining steps are consistent with those of a forgetting gate and an input gate, but the self-defined matrix, the offset and the activation function are different, and the expression is as follows:
C′ t ＝tanh(W c ·[z t-1 ,x t ]+b C )
wherein, W C Is a custom matrix of cell intermediate states, b C Tan h is the activation function for the offset of the cell mesostate.
Further, in step (2), the current state C of the nodes in the network is memorized for a long time t The information quantity of the two-part structure is determined by a forgetting gate and an input gate, and the expression is as follows:
C t ＝f t *C t-1 +i t *C′ t 。
further, in step (2), the output gate o in the long-short time memory network t The obtaining steps are consistent with those of a forgetting gate and an input gate, but the self-defined matrix and the offset are different; the expression is as follows:
o t ＝sigmoid(W o ·[z t-1 ,x t ]+b o )
wherein, W o Is a custom matrix of cell intermediate states, b o Is the offset of the cell intermediate state.
Further, in step (2), the hidden state h of the current node in the network is memorized for a long time t The current node state is obtained after being processed by an activation function tanh, and the information quantity of the current node state is obtained by an output gateAnd controlling, wherein the expression is as follows:
h t ＝o t *tanh(C t )。
further, in the step (4), the reconstruction error adopts the euclidean distance, and the expression is as follows:
wherein x is i Is the original input value, x' i Is the decoder output value.
On the other hand, the invention also provides an abnormality detection device under the non-stationary time series data flow field scene, which comprises a memory and one or more processors, wherein the memory is stored with executable codes.
The invention has the beneficial effects that: provided is an abnormality detection method in a non-stationary time series data flow field scene. A variational self-encoder is used as an external basic frame, and the non-stationarity of data and the time sequence of the data are modeled and optimized together by improving the acquisition mode of hidden vectors in a long-time and short-time memory network. The model optimized by the method improves the modeling precision and ensures the global optimization of the solution scheme on the premise of ensuring the rationality and the basic accuracy of the method.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.
FIG. 1 is a general schematic diagram of a non-stationary time series data flow anomaly detection method framework;
fig. 2 is a detailed diagram of an improved long term memory network.
Fig. 3 is a structural diagram of an abnormality detection apparatus in a non-stationary time-series data flow field scenario according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
The method selects bridge structure health detection as a task, and the data set is data acquired in real time in a bridge structure health monitoring system sensor. And extracting time sequence characteristics and non-stationary characteristics of the bridge structure data, and optimizing the long-time and short-time memory neural network model. And inputting real-time acquired data in a bridge structure health monitoring system sensor into the optimized model, outputting an error quantification result serving as bridge structure data, and completing the health monitoring of the bridge structure through a self-defined threshold.
As shown in fig. 1, the overall schematic diagram of the framework of the anomaly detection method in the non-stationary time sequence scene provided by the present invention is provided, the method encodes the processed data, extracts the time sequence characteristics and the non-stationary characteristics of the data, and maps the data to the hidden space. And obtaining an implicit vector corresponding to the current data through sampling of the implicit space, inputting the implicit vector into a decoder for decoding operation, and mapping the implicit vector back to the data space through key information contained in the implicit vector so as to achieve the purpose of reconstruction.
For normal data, because the trend of the change of the data can be learned by the model, the model can reconstruct the data well in theory, namely the reconstruction error is small; for abnormal data, since the abnormal data does not conform to the conventional change rule, the model cannot learn the data pattern thereof, and therefore the reconstruction error is large. By the above determination criteria, it is possible to accurately capture an abnormality in data.
The method proposed by the present invention needs to satisfy the following assumptions:
1) The data contains time sequence properties, namely each data point in the data has a front-back incidence relation;
2) The data contains non-stationary properties, i.e. the data changes or gradually changes over time, rather than within a fixed interval.
Aiming at a real bridge data set, the method provided by the invention comprises the following specific steps:
(1) Data preprocessing: the sensors for acquiring bridge data comprise a cable force acceleration sensor, a mechanical anemoscope, a structural strain sensor, an external temperature and humidity sensor, a structural temperature sensor and a water level sensor. The data of 6 sensors are processed respectively, the sampling frequency of each sensor is very high, generally 10Hz or 50Hz, which results in the data volume being much larger than the requirement of model training, and most data have almost no change in a period of time, and are redundant data for the training process, so the data volume needs to be reduced. In addition, the sampling frequency and the sampling time period of different sensors are inconsistent, and the sampling of each sensor has certain loss, but the lost data are completely inconsistent, so that great difficulty is caused to data alignment. Data of a certain month is selected as preprocessed data, and sampling records of all sensors in the month are guaranteed, so that the problem that sampling time periods of the sensors are inconsistent can be solved. However, even one month's worth of data has millions of records, and due to the missing problem, it is difficult to align the data records of all sensors according to the time stamp. Therefore, the data of each minute is averaged with one minute as the minimum time unit, and the data is recorded as the current time. The data of the sensor is generally not continuously lost for one minute, so that the problem of data loss can be solved, and redundant data is obviously reduced. After the steps are completed, normalization operation is carried out on the data of each sensor, and finally, a piece of multi-feature time sequence data X belongs to l X d through splicing.
(2) Inputting the preprocessed bridge data into an encoder part of a variational self-encoder shown in fig. 2, wherein the encoder is basically composed of an improved long-time and short-time memory network, and simultaneously learns the time sequence and the non-stationarity of the data: data point x for the current time t Linear transformation is carried out on the matrix by a plurality of different user-defined matrixes, different bias terms are added, and finally the matrix is processed by an activation function sigmoid or tanhTo obtain a forgetting door f t Input gate i t Output gate o t And cell intermediate state C 'of current time' t . The specific process is as follows:
(2.1) first calculate the forgetting gate f t . The implicit vector z obtained in the last period t-1 Data x from the current time t Combined with the custom matrix W f Multiplied by each other and added with an offset b f And processing the obtained result by an activation function sigmoid to obtain the forgetting gate. The formula is expressed as: f. of t ＝sigmoid(W f ·[z t-1 ,x t ]+b f )。z t-1 For the implicit vector, x, obtained in the preceding period t As data of the current time, W f Custom matrix for forgetting to gate, b f Sigmoid is an activation function for the offset of a forgetting gate. The acquisition mode of the hidden vector is described in the following steps;
(2.2) calculating an input gate. The acquisition step of the input gate is consistent with that of the forgetting gate, but the self-defined matrix and the offset are different. The formula is expressed as: i.e. i t ＝sigmoid(W i ·[z t-1 ,x t ]+b i )；W i For custom matrices of input gates, b i The offset of the input gate.
And (2.3) calculating the intermediate state and the current state of the cell. The acquisition steps of the intermediate state of the cell are the same as those of the two gate control units, but the self-defined matrix, the offset and the activation function are different, and the formula is expressed as follows: c' t ＝tanh(W C ·[z t-1 ,x t ]+b C )。W C Is a custom matrix of cell intermediate states, b C Tan h is the activation function, which is the offset of the cell intermediate state.
The weighting processing of a forgetting gate and an input gate is used for determining the information quantity of the cell state at the previous moment and the cell state at the current moment, which is input to the cell state at the current moment from the cell state at the current moment, and the cell state at the current moment, namely C, is obtained t ＝f t *C t-1 +i t *C′ t 。
And (2.4) calculating an output gate. The acquisition step of the input gate is identical to the gate control unit described above, butThe custom matrix and the offset are different. The formula is expressed as: o t ＝sigmoid(W o ·[z t-1 ,x t ]+b o )；W o A custom matrix for the cell intermediate state, b o Is the offset of the cell intermediate state.
And (2.5) calculating the hidden state and the hidden vector of the current node. After the cell state at the current moment is processed by an activation function tanh, the output gate determines the output information quantity to obtain the hidden state of the current node, namely h t ＝o t *tanh(C t )。
And obtaining the distribution expression of the hidden space from the hidden state at the current moment through two different linear transformations. Obtaining a non-stationary quantization result of the data at the current time according to the difference between the implicit spatial distribution of the previous time and the implicit spatial distribution of the data at the current time, and splicing the result and the hidden state of the current node to obtain an implicit vector z of the data at the current time t . The above steps can be expressed as: z is a radical of t ＝[h t ,D(N(W μ ·h t ,softplus(W σ ·h t ))||N(W μ ·h t-1 ,softplus(W σ ·h t-1 )))]. Wherein h is t Is an output value at the present time, W μ 、W σ Is a custom matrix of mean and variance, respectively, h t-1 Softplus is the output value at the last moment, and is the activation function;
(3) The hidden vector z obtained from the output of the encoder t Inputting the data into a decoder, dividing the decoder into two parts, wherein the internal structure of the first part is consistent with that of the encoder, and obtaining an intermediate output z t ′ . In addition, the decoder needs to map the implicit spatial data back to the original data space, i.e. first perform a linear transformation W on the intermediate output z · t ′ And adding an offset term b z And then the reconstructed output x can be obtained through the sigmoid processing of the activation function t ′ 。
(4) Calculating Euclidean distance of the reconstructed output from the original input, i.e.
x i For the original input value, x i ′ Is the decoder output value. In the model training process, the reconstruction error is used as the reconstruction error of the bridge structure data at the current moment and is optimized as one item in a loss function; and in the using process of the model, the result is an abnormal score, and whether the bridge structure data at the current moment is abnormal or not is judged according to the relation between the score and the user-defined threshold value.
Corresponding to the embodiment of the abnormal detection method under the non-stationary time sequence data flow field scene, the invention also provides an embodiment of an abnormal detection device under the non-stationary time sequence data flow field scene.
Referring to fig. 3, the apparatus for detecting an abnormality in a non-stationary time series data flow field scene provided in an embodiment of the present invention includes a memory and one or more processors, where the memory stores executable codes, and when the processors execute the executable codes, the apparatus is configured to implement the method for detecting an abnormality in a non-stationary time series data flow field scene in the foregoing embodiment.
The embodiment of the abnormality detection device in the non-stationary time series data flow field scene can be applied to any equipment with data processing capability, and the any equipment with data processing capability can be equipment or devices such as computers. The device embodiments may be implemented by software, or by hardware, or by a combination of hardware and software. The software implementation is taken as an example, and as a logical device, the device is formed by reading corresponding computer program instructions in the nonvolatile memory into the memory for running through the processor of any device with data processing capability. In terms of hardware, as shown in fig. 3, the present invention is a hardware structure diagram of any device with data processing capability where the anomaly detection apparatus in the non-stationary time series data flow field is located, except for the processor, the memory, the network interface, and the nonvolatile memory shown in fig. 3, in the embodiment, any device with data processing capability where the apparatus is located may also include other hardware according to the actual function of the any device with data processing capability, which is not described again.
The implementation process of the functions and actions of each unit in the above device is specifically described in the implementation process of the corresponding step in the above method, and is not described herein again.
For the device embodiment, since it basically corresponds to the method embodiment, reference may be made to the partial description of the method embodiment for relevant points. The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of the invention. One of ordinary skill in the art can understand and implement it without inventive effort.
An embodiment of the present invention further provides a computer-readable storage medium, on which a program is stored, where the program, when executed by a processor, implements the abnormal detection method in the non-stationary time-series data flow field scenario in the foregoing embodiment.
The computer readable storage medium may be an internal storage unit, such as a hard disk or a memory, of any data processing capability device described in any of the foregoing embodiments. The computer readable storage medium may also be any external storage device of a device with data processing capabilities, such as a plug-in hard disk, a Smart Media Card (SMC), an SD Card, a Flash memory Card (Flash Card), etc. provided on the device. Further, the computer readable storage medium may include both an internal storage unit and an external storage device of any data processing capable device. The computer-readable storage medium is used for storing the computer program and other programs and data required by the arbitrary data processing-capable device, and may also be used for temporarily storing data that has been output or is to be output.
The above-described embodiments are intended to illustrate rather than to limit the invention, and any modifications and variations of the present invention are within the spirit of the invention and the scope of the appended claims.
Claims (10)
1. An abnormality detection method under a non-stationary time series data flow field scene is characterized by comprising the following steps:
(1) Data preprocessing: acquiring data from m sensors for monitoring the health of a bridge structure in real time, integrating and averaging the sensor data according to the sampling frequency of the sensors, transversely splicing the m sensor data, and combining the m sensor data into n-dimensional time sequence data;
(2) Inputting the obtained n-dimensional time sequence data into an encoder part of a variational self-encoder, wherein the encoder is structurally characterized by improving a long-time memory network of a hidden vector and learning the time sequence and the non-stationarity of the data; the implicit vector acquisition mode is as follows: mapping the hidden state through two different custom matrixes respectively to obtain expression of a mean value and a variance, wherein the variance part is obtained by processing through a softplus function, all the custom matrixes are obtained through random initialization of an Xavier algorithm, and the weight is continuously adjusted in the training process; obtaining the expression of Gaussian distribution of data at the current moment through the mean value and the variance, calculating KL distance with the Gaussian distribution at the previous moment to obtain the quantization difference of the data distribution of the two parts, and splicing the quantization result with a hidden state to obtain a hidden vector; the expression is as follows:
z t ＝[h t ，D(N(W μ ·h t ，softplus(W σ ·h t ))||N(W μ ·h t-1 ，softplus(W σ ·h t-1 )))]
wherein h is t Is an output value at the present time, W μ 、W σ Is a custom matrix of mean and variance, respectively, h t-1 Softplus is the output value at the last moment, and is the activation function;
(3) To conceal the output of the encoder by the vector z t Decoding as input to a decoder; the structure of the decoder is consistent with that of the encoder, but a layer of fully-connected neural network is added at the end of the decoder, and the decoding result is mapped back to the numberAccording to the space;
(4) And performing reconstruction error calculation on the output result of the decoder and the original input, wherein the error quantization result is normalized, and the abnormal data which is larger than the user-defined threshold value is obtained, and the normal data is obtained.
2. The method as claimed in claim 1, wherein in step (1), the data of each minute is averaged with a minimum time unit of one minute, and the average of the sampled data per minute is used as the data record of the current time to process the missing value of the sensor sample.
3. The method for detecting the abnormality under the view of the non-stationary time-series data flow field according to claim 1, wherein in the step (2), the long-time and short-time memory network comprises a forgetting gate f t The expression of (a) is as follows:
f t ＝sigmoid(W f ·[z t-1 ，x t ]+b f )
wherein z is t-1 For the implicit vector, x, obtained in the preceding period t As data of the current time, W f Custom matrix for forgetting to gate, b f Sigmoid is an activation function for the offset of a forgetting gate.
4. The method as claimed in claim 3, wherein in step (2), the input gate i in the long-short time memory network is set to be a long-short time memory network t The obtaining steps are consistent with those of a forgetting gate, but the self-defined matrix and the offset are different; the expression is as follows:
i t ＝sigmoid(W i ·[z t-1 ，x t ]+b i )
wherein, W i A custom matrix for the input gate, b i Is the offset of the input gate.
5. A non-stationary time-series data flow field scene as claimed in claim 3 or 4The abnormality detection method is characterized in that in the step (2), the intermediate state C 'of the cells in the long and short term memory network' t The obtaining steps are consistent with those of a forgetting gate and an input gate, but the self-defined matrix, the offset and the activation function are different, and the expression is as follows:
C′ t ＝tanh(W C ·[z t-1 ，x t ]+b C )
wherein, W C Is a custom matrix of cell intermediate states, b C Tan h is the activation function, which is the offset of the cell intermediate state.
6. The method as claimed in claim 5, wherein in step (2), the current state C of the node in the network is stored in a long-and-short-term manner t The information quantity of the two-part structure is determined by a forgetting gate and an input gate, and the expression is as follows:
C t ＝f t *C t-1 +i t *C′ t 。
7. the method as claimed in claim 3 or 4, wherein in step (2), the output gate o in the long-short time memory network is used for detecting the abnormal condition in the non-stationary time-series data flow field scene t The obtaining steps are consistent with those of a forgetting gate and an input gate, but the self-defined matrix and the offset are different; the expression is as follows:
o t ＝sigmoid(W o ·[z t-1 ，x t ]+b o )
wherein, W o A custom matrix for the cell intermediate state, b o Is the offset of the cell intermediate state.
8. The method as claimed in claim 7, wherein in step (2), the hidden state h of the current node in the long-short term memory network is determined according to the length of the long-short term memory network t The current node state is obtained after being processed by an activation function tanh, the information quantity of the current node state is controlled by an output gate, and the current node state is obtained after being processed by the activation function tanhThe expression is as follows:
h t ＝o t *tanh(C t )。
9. the method for detecting the abnormality in the non-stationary time series data flow field scene as claimed in claim 1, wherein in the step (4), the reconstruction error adopts euclidean distance, and the expression is as follows:
wherein x is i Is the original input value, x' i The decoder outputs the value.
10. An anomaly detection device in a non-stationary time series data flow field scene, comprising a memory and one or more processors, wherein the memory stores executable code, and wherein the processors, when executing the executable code, are configured to implement the steps of the anomaly detection method in the non-stationary time series data flow field scene as claimed in any one of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211484703.7A CN115859202B (en) | 2022-11-24 | 2022-11-24 | Abnormality detection method and device under non-stationary time sequence data stream scene |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211484703.7A CN115859202B (en) | 2022-11-24 | 2022-11-24 | Abnormality detection method and device under non-stationary time sequence data stream scene |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115859202A true CN115859202A (en) | 2023-03-28 |
| CN115859202B CN115859202B (en) | 2023-10-10 |
Family
ID=85666071
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211484703.7A Active CN115859202B (en) | 2022-11-24 | 2022-11-24 | Abnormality detection method and device under non-stationary time sequence data stream scene |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115859202B (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109583570A (en) * | 2018-11-30 | 2019-04-05 | 重庆大学 | The method for determining bridge health monitoring system abnormal data source based on deep learning |
| 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 |
| US20210019619A1 (en) * | 2019-07-17 | 2021-01-21 | Robert Bosch Gmbh | Machine learnable system with conditional normalizing flow |
| CN112488235A (en) * | 2020-12-11 | 2021-03-12 | 江苏省特种设备安全监督检验研究院 | Elevator time sequence data abnormity diagnosis method based on deep learning |
| US20210334656A1 (en) * | 2018-09-05 | 2021-10-28 | Sartorius Stedim Data Analytics Ab | Computer-implemented method, computer program product and system for anomaly detection and/or predictive maintenance |
| US20220084371A1 (en) * | 2020-09-11 | 2022-03-17 | Capital One Services, Llc | Systems and methods for unsupervised detection of anomalous customer interactions to secure and authenticate a customer session |
| US11294756B1 (en) * | 2019-09-19 | 2022-04-05 | Amazon Technologies, Inc. | Anomaly detection in a network |
| CN114626479A (en) * | 2022-03-23 | 2022-06-14 | 南京大学 | Time sequence anomaly detection method and device based on variational self-coding time sequence decomposition |
| CN114676778A (en) * | 2022-03-28 | 2022-06-28 | 北京科技大学 | Abnormal data detection method and device for cooperative and mutual feedback of shield equipment data |
| US20220217062A1 (en) * | 2020-08-10 | 2022-07-07 | Samsung Electronics Co., Ltd. | Apparatus and method for monitoring network |
| KR102419782B1 (en) * | 2021-09-29 | 2022-07-12 | 가온플랫폼 주식회사 | Industrial facility failure prediction modeling technique and alarm integrated system and method based on artificial intelligence |
| WO2022160902A1 (en) * | 2021-01-28 | 2022-08-04 | 广西大学 | Anomaly detection method for large-scale multivariate time series data in cloud environment |
-
2022
- 2022-11-24 CN CN202211484703.7A patent/CN115859202B/en active Active
Patent Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210334656A1 (en) * | 2018-09-05 | 2021-10-28 | Sartorius Stedim Data Analytics Ab | Computer-implemented method, computer program product and system for anomaly detection and/or predictive maintenance |
| CN109583570A (en) * | 2018-11-30 | 2019-04-05 | 重庆大学 | The method for determining bridge health monitoring system abnormal data source based on deep learning |
| 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 |
| US20210019619A1 (en) * | 2019-07-17 | 2021-01-21 | Robert Bosch Gmbh | Machine learnable system with conditional normalizing flow |
| US11294756B1 (en) * | 2019-09-19 | 2022-04-05 | Amazon Technologies, Inc. | Anomaly detection in a network |
| US20220217062A1 (en) * | 2020-08-10 | 2022-07-07 | Samsung Electronics Co., Ltd. | Apparatus and method for monitoring network |
| US20220084371A1 (en) * | 2020-09-11 | 2022-03-17 | Capital One Services, Llc | Systems and methods for unsupervised detection of anomalous customer interactions to secure and authenticate a customer session |
| CN112488235A (en) * | 2020-12-11 | 2021-03-12 | 江苏省特种设备安全监督检验研究院 | Elevator time sequence data abnormity diagnosis method based on deep learning |
| WO2022160902A1 (en) * | 2021-01-28 | 2022-08-04 | 广西大学 | Anomaly detection method for large-scale multivariate time series data in cloud environment |
| KR102419782B1 (en) * | 2021-09-29 | 2022-07-12 | 가온플랫폼 주식회사 | Industrial facility failure prediction modeling technique and alarm integrated system and method based on artificial intelligence |
| CN114626479A (en) * | 2022-03-23 | 2022-06-14 | 南京大学 | Time sequence anomaly detection method and device based on variational self-coding time sequence decomposition |
| CN114676778A (en) * | 2022-03-28 | 2022-06-28 | 北京科技大学 | Abnormal data detection method and device for cooperative and mutual feedback of shield equipment data |
Non-Patent Citations (2)
| Title |
|---|
| 李晓菊等: "基于变分循环自动编码器的协同推荐方法", 计算机应用与软件, vol. 39, no. 09, pages 258 - 280 * |
| 杨英等: "VAE＿LSTM算法在时间序列预测模型中的研究", 湖南科技大学学报(自然科学版), vol. 35, no. 03, pages 93 - 101 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115859202B (en) | 2023-10-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110383299B (en) | Memory enhanced generation time model | |
| CN111352965B (en) | Training method of sequence mining model, and processing method and equipment of sequence data | |
| CN112699998B (en) | Time series prediction method and device, electronic equipment and readable storage medium | |
| CN112101554A (en) | Anomaly detection method and device, equipment and computer-readable storage medium | |
| CN116522265A (en) | Industrial Internet time sequence data anomaly detection method and device | |
| CN116381753B (en) | Neural network assisted navigation method of GNSS/INS integrated navigation system during GNSS interruption | |
| Zhu et al. | A coupled model for dam foundation seepage behavior monitoring and forecasting based on variational mode decomposition and improved temporal convolutional network | |
| US20230117088A1 (en) | Method and device for improving performance of data processing model, storage medium and electronic device | |
| CN118295029B (en) | Seismic data denoising method integrating self-attention and Mamba architecture | |
| CN115098789B (en) | Multidimensional interest fusion recommendation method and device based on neural network and related equipment | |
| Wang et al. | Local–global correlation fusion-based graph neural network for remaining useful life prediction | |
| Qi et al. | Fault detection and localization in distributed systems using recurrent convolutional neural networks | |
| CN114298299A (en) | Model training method, device, equipment and storage medium based on course learning | |
| CN110866672B (en) | Data processing method, device, terminal and medium | |
| CN112949947A (en) | Power failure early warning method and system based on convolution long-term and short-term memory network | |
| CN116992380A (en) | Satellite multidimensional telemetry sequence anomaly detection model construction method and device, anomaly detection method and device | |
| CN116882538B (en) | Training method and related device for marine environment prediction model | |
| CN115859202B (en) | Abnormality detection method and device under non-stationary time sequence data stream scene | |
| CN116628612A (en) | Unsupervised anomaly detection method, device, medium and equipment | |
| CN117807470A (en) | Multiple calibration sheet classification method for unsupervised time series anomaly detection | |
| CN116070076A (en) | High-precision high-robustness radar missing signal completion system | |
| CN115562216A (en) | Fault diagnosis method and system for non-stationary industrial process | |
| CN116992937A (en) | Neural network model restoration method and related equipment | |
| CN113222972A (en) | Image anomaly detection method based on variational self-encoder algorithm | |
| CN118362809A (en) | Power system fault detection method, device, equipment, medium and product |
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 |