CN114255373B - Sequence anomaly detection method, device, electronic equipment and readable medium - Google Patents
Sequence anomaly detection method, device, electronic equipment and readable medium Download PDFInfo
- Publication number
- CN114255373B CN114255373B CN202111615986.XA CN202111615986A CN114255373B CN 114255373 B CN114255373 B CN 114255373B CN 202111615986 A CN202111615986 A CN 202111615986A CN 114255373 B CN114255373 B CN 114255373B
- Authority
- CN
- China
- Prior art keywords
- sequence
- image
- detected
- local area
- region
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 60
- 238000000034 method Methods 0.000 claims abstract description 61
- 230000002159 abnormal effect Effects 0.000 claims abstract description 57
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 25
- 238000010845 search algorithm Methods 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims description 33
- 238000012545 processing Methods 0.000 claims description 21
- 230000008569 process Effects 0.000 claims description 19
- 238000001914 filtration Methods 0.000 claims description 17
- 238000004422 calculation algorithm Methods 0.000 claims description 14
- 230000000737 periodic effect Effects 0.000 claims description 14
- 238000005070 sampling Methods 0.000 claims description 11
- 238000003709 image segmentation Methods 0.000 claims description 9
- 238000013507 mapping Methods 0.000 claims description 8
- 230000005856 abnormality Effects 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 12
- 238000004364 calculation method Methods 0.000 description 3
- 238000002372 labelling Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000013179 statistical model Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 239000004278 EU approved seasoning Substances 0.000 description 1
- 238000012300 Sequence Analysis Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000001747 exhibiting effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 235000011194 food seasoning agent Nutrition 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000003534 oscillatory effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000001932 seasonal effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06N3/047—Probabilistic or stochastic 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
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
The present disclosure provides a sequence anomaly detection method, apparatus, electronic device, and readable medium, wherein the anomaly sequence detection method includes: converting the sequence to be detected into an image to obtain a sequence image to be detected; acquiring a first local area of an image of a sequence to be detected based on a convolutional neural network; based on a selective search algorithm, a second local area of the image of the sequence to be detected is obtained; inputting the first local area and the second local area into a pre-trained classification network VGGnet to obtain a classification result of the sequence to be detected, wherein the classification result is a normal sequence or an abnormal sequence. The method and the device convert the sequence to be detected into the picture, complete the abnormal detection of the sequence to be detected by adopting a weak supervision learning mode and an attention mechanism and positioning and identifying, and utilize the visually significant characteristics of the abnormal sequence and the normal sequence without considering the statistical difference of the abnormal sequence and the normal sequence, thereby detecting the abnormal sequence more accurately.
Description
Technical Field
The present disclosure relates to the field of data analysis and computer vision technologies, and in particular, to a sequence anomaly detection method, apparatus, electronic device, and readable medium.
Background
A non-stationary sequence is a time sequence that contains trends, seasonings, or periodicity. The abnormal detection of the non-stationary sequence is to find abnormal fluctuation which does not accord with the trend and periodicity, for example, in the operation and maintenance field, the analysis of key performance indicators (Key Performance Indicator, KPI) timely finds system abnormality, the analysis of network traffic timely finds network fault and the like, and the method has great significance for guaranteeing the reliability and stability of the system.
In the prior art, when abnormal detection is carried out on an unstable sequence, the abnormal detection is usually based on a statistical model, the problems of insensitivity of abnormal data, large limitation of a threshold value or a limit value for judging the abnormal data, small scene coverage of the abnormal data and the like often occur, a general algorithm is difficult to find to solve most of the abnormal detection problems, the detection accuracy is not high, and the abnormal data is easy to "submerge" in massive normal data.
Based on this, how to improve the applicability of the non-stationary sequence anomaly detection method and the accuracy of anomaly detection becomes a technical problem to be solved.
It should be noted that the information disclosed in the above background section is only for enhancing understanding of the background of the present disclosure and thus may include information that does not constitute prior art known to those of ordinary skill in the art.
Disclosure of Invention
The disclosure provides a sequence anomaly detection method, a device, an electronic device and a storage medium, which at least overcome the problems of small application range and low accuracy of the sequence anomaly detection method in the related art to a certain extent.
Other features and advantages of the present disclosure will be apparent from the following detailed description, or may be learned in part by the practice of the disclosure.
According to one aspect of the present disclosure, there is provided a sequence anomaly detection method including: converting the sequence to be detected into an image to obtain a sequence image to be detected; acquiring a first local area of the sequence image to be detected based on a convolutional neural network; acquiring a second local area of the sequence image to be detected based on a selective search algorithm; inputting the first local area and the second local area into a pre-trained classification network VGGnet to obtain a classification result of the sequence to be detected, wherein the classification result is a normal sequence or an abnormal sequence.
In one embodiment of the present disclosure, before the converting the sequence to be measured into the image, the method further includes: the filtering processing of the sliding average value is carried out on the sequence to be detected, and the method specifically comprises the following steps: taking the continuous N sampling values as a queue, wherein the length of the queue is fixed to be N; when new data is acquired, the acquired new data is put at the tail of the team, and the data at the head of the team is discarded; and carrying out arithmetic average operation on N data in the queue to obtain the sequence to be detected.
In one embodiment of the present disclosure, the test sequence is a periodic time sequence.
In one embodiment of the present disclosure, the acquiring, based on the convolutional neural network, the first local region of the image of the sequence to be measured includes: generating a score graph of the sequence image to be detected by utilizing feature mapping in a convolutional neural network, wherein the score graph consists of a first convolutional layer, a second convolutional layer and a spatial softmax layer; the first convolution layer utilizes 64 3 x 3 kernels and the second convolution layer utilizes 1 3 x 3 kernels to generate a single channel confidence map; the spatial softmax layer acts on the confidence map to convert confidence scores into region probabilities; and taking the region with the region probability exceeding the preset region probability as a first local region of the sequence image to be detected, and acquiring the first local region of the sequence image to be detected.
In one embodiment of the present disclosure, after the converting the sequence to be measured into images, the method further comprises: generating an image pyramid of the image converted by the sequence to be detected based on an image algorithm; and selecting the image with the highest resolution ratio from the image golden sub-tower as the image of the sequence to be detected.
In one embodiment of the disclosure, the acquiring, based on the selective search algorithm, the second local area of the image of the sequence to be detected includes: performing image segmentation on the images of the sequence to be detected to obtain a region set of the images of the sequence to be detected; calculating the similarity between adjacent areas in the area set, and combining two areas with highest similarity between the adjacent areas; repeating the previous steps until the images are combined into a complete image of the sequence to be detected and stopping; and acquiring a second local area of the images of the sequence to be detected from the images generated in the merging process.
In one embodiment of the present disclosure, the method further comprises: acquiring a normal sequence and an abnormal sequence with random sequence length; converting the normal sequence and the abnormal sequence with the random sequence length into images to obtain a sample training set; and taking the sample training set as input, taking a classification result of the image in the sample training set as output, and training the classification network VGGnet to obtain the classification network VGGnet.
According to another aspect of the present disclosure, there is provided a sequence abnormality detection apparatus including: the sequence conversion image module is used for converting the preprocessed sequence to be detected into an image to obtain a sequence image to be detected; the first region acquisition module is used for acquiring a first local region of the sequence image to be detected based on a convolutional neural network; the second region acquisition module is used for acquiring a second local region of the sequence image to be detected based on a selective search algorithm; and the classification result obtaining module is used for inputting the first local area and the second local area into a pre-trained classification network VGGnet to obtain the classification result of the sequence to be detected, wherein the classification result is a normal sequence or an abnormal sequence.
In one embodiment of the disclosure, the apparatus further includes a moving average filtering module, where the moving average filtering module is configured to use N consecutive sampling values as a queue, and a length of the queue is fixed to N; when new data is acquired, the acquired new data is put at the tail of the team, and the data at the head of the team is discarded; and carrying out arithmetic average operation on N data in the queue to obtain the sequence to be detected.
In one embodiment of the disclosure, the first region obtaining module is specifically configured to generate a score map of the image to be measured by using feature mapping in a convolutional neural network, where the score map is composed of a first convolutional layer, a second convolutional layer, and a spatial softmax layer; the first convolution layer utilizes 64 3 x 3 kernels and the second convolution layer utilizes 1 3 x 3 kernels to generate a single channel confidence map; the spatial softmax layer acts on the confidence map to convert confidence scores into region probabilities; and taking the region with the region probability exceeding the preset region probability as a first local region of the sequence image to be detected, and acquiring the first local region of the sequence image to be detected.
In one embodiment of the disclosure, the apparatus further includes an image algorithm processing module, configured to generate an image pyramid of the image converted by the sequence to be tested based on an image algorithm; and selecting the image with the highest resolution ratio from the image golden sub-tower as the image of the sequence to be detected.
In an embodiment of the disclosure, the second region obtaining module is specifically configured to perform image segmentation on the sequence image to be detected to obtain a region set of the sequence image to be detected; calculating the similarity between adjacent areas in the area set, and combining two areas with highest similarity between the adjacent areas; repeating the previous steps until the images are combined into a complete image of the sequence to be detected and stopping; and acquiring a second local area of the images of the sequence to be detected from the images generated in the merging process.
In one embodiment of the disclosure, the apparatus further includes a classification network training module, where the classification network training module is configured to obtain a normal sequence and an abnormal sequence of a random sequence length; converting the normal sequence and the abnormal sequence with the random sequence length into images to obtain a sample training set; and taking the sample training set as input, taking a classification result of the image in the sample training set as output, and training the classification network VGGnet to obtain the classification network VGGnet.
According to still another aspect of the present disclosure, there is provided an electronic apparatus including: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the sequence anomaly detection method described above via execution of the executable instructions.
According to still another aspect of the present disclosure, there is provided a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the sequence anomaly detection method described above.
The embodiment of the disclosure provides a sequence anomaly detection method, a device, an electronic device and a readable medium, wherein the anomaly sequence detection method comprises the following steps: converting the sequence to be detected into an image to obtain a sequence image to be detected; acquiring a first local area of an image of a sequence to be detected based on a convolutional neural network; based on a selective search algorithm, a second local area of the image of the sequence to be detected is obtained; inputting the first local area and the second local area into a pre-trained classification network VGGnet to obtain a classification result of the sequence to be detected, wherein the classification result is a normal sequence or an abnormal sequence. The method and the device complete the abnormal detection of the sequence to be detected by converting the sequence to be detected into pictures and adopting a weak supervision learning mode and an attention mechanism and by positioning and identifying, and utilize the visually significant characteristics of the abnormal sequence and the normal sequence without considering the statistical difference of the abnormal sequence and the normal sequence. Meanwhile, the abnormal sequence is detected more accurately without depending on limiting factors such as a threshold value.
Further, after the sequence to be detected is converted into the image, an image pyramid of the converted image is constructed based on an image algorithm, and the image with the highest resolution is obtained from the image pyramid to serve as the image of the sequence to be detected, so that the visual characteristics of the sequence to be detected are more obvious.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the principles of the disclosure. It will be apparent to those of ordinary skill in the art that the drawings in the following description are merely examples of the disclosure and that other drawings may be derived from them without undue effort.
FIG. 1 illustrates a flow chart of a method of sequence anomaly detection in an embodiment of the present disclosure;
FIG. 2 illustrates a flow chart of another method of sequence anomaly detection in an embodiment of the present disclosure;
FIG. 3 illustrates a flow chart of another method of sequence anomaly detection in an embodiment of the present disclosure;
FIG. 4 illustrates a flow chart of another method of sequence anomaly detection in an embodiment of the present disclosure;
FIG. 5 illustrates a flow chart of another method of sequence anomaly detection in an embodiment of the present disclosure;
FIG. 6 illustrates a flow chart of another method of sequence anomaly detection in an embodiment of the present disclosure;
FIG. 7 illustrates a flow chart of yet another sequence anomaly detection method in an embodiment of the present disclosure;
FIG. 8 illustrates a schematic diagram of a moving average filtering process in an embodiment of the present disclosure;
FIG. 9 illustrates a schematic diagram of a convolution operator acquiring keypoints in an embodiment of the disclosure;
FIG. 10 illustrates a schematic diagram of a selective search algorithm acquiring a local region in an embodiment of the present disclosure;
FIG. 11 illustrates a schematic diagram of a collection of samples in an embodiment of the present disclosure;
fig. 12 is a schematic diagram showing a configuration of a sequence abnormality detection apparatus in an embodiment of the present disclosure; and
fig. 13 shows a block diagram of an electronic device in an embodiment of the disclosure.
Detailed Description
Example embodiments will now be described more fully with reference to the accompanying drawings. However, the exemplary embodiments may be embodied in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Furthermore, the drawings are merely schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus a repetitive description thereof will be omitted. Some of the block diagrams shown in the figures are functional entities and do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software or in one or more hardware modules or integrated circuits or in different networks and/or processor devices and/or microcontroller devices.
As mentioned in the background art, in the prior art, when detecting the anomalies of the unsteady sequence, the problems that the anomalies are insensitive, the threshold value or limit value of the anomalies is limited and the scene coverage of the anomalies is small often occur on the basis of a statistical model, so that a general algorithm is difficult to find to solve most of the anomalies, the detection accuracy is not high, and the anomalies are easy to "submerge" in massive normal data.
In addition, the conventional method for detecting abnormal non-stationary sequences generally requires converting to stationary sequences by using a method such as difference or decomposing to residual sequences by using STL (Seasonal and Trend decomposition using Loess, time sequence decomposition algorithm), and then performing abnormal sequence analysis. While such a process may tend to suffer from several problems: the transformation process may lose the characteristic of the difference between the outlier and the normal value; the abnormal value is insensitive, and if the abnormal data is not different from the normal data in the statistical level, the abnormal data is difficult to detect; the abnormal judgment is usually to compare the monitoring index with a threshold, wherein the threshold is different for different systems and different scenes and even different times, and a reasonable threshold is difficult to set, so that abnormal situations can be covered as much as possible.
Based on the above, the application provides an abnormal sequence detection method, which is characterized in that a sequence to be detected is converted into a picture, a weak supervision learning mode and an attention mechanism are adopted, the abnormal detection of the sequence to be detected is completed through positioning and recognition, the visually significant characteristics of the abnormal sequence and the normal sequence are utilized, the statistical difference between the abnormal sequence and the normal sequence is not required to be considered, and the abnormal sequence is detected more accurately. Meanwhile, the method does not depend on limiting factors such as a threshold value, and the like, so that the applicability of the detection method is improved.
For ease of understanding, the following first explains the several terms involved in this disclosure as follows:
weak supervised learning, which is a branch in the field of machine learning, uses limited, noisy, or inaccurately annotated data to train model parameters, as compared to traditional supervised learning. Weak supervised learning can be classified into the following three categories according to the labeling degree of data: incomplete supervision, inaccurate supervision, and inaccurate supervision. If part of the data in the sample has labeling information, the rest part does not have effective labeling, and the monitoring is not completely performed. When the data has only coarse-grained labels, it is called unsure-supervision. Inaccurate supervision means that the sample has labels, but is not accurate.
Image pyramids are a kind of multi-scale expression in images, are mainly used for image segmentation, are effective and conceptually simple structures for explaining images with multiple resolutions, are initially used for machine vision and image compression, and are a series of image sets which are arranged in pyramid shapes, gradually reduce in resolution and are derived from the same original image. It is obtained by downsampling a step down and does not stop sampling until a certain termination condition is reached. The bottom of the pyramid is a high resolution representation of the image to be processed, while the top is an approximation of low resolution, the higher the level, the smaller the image and the lower the resolution.
The present exemplary embodiment will be described in detail below with reference to the accompanying drawings and examples.
First, in the embodiment of the present disclosure, a sequence anomaly detection method is provided, and the method may be executed by any electronic device having a computing processing capability.
Fig. 1 shows a flowchart of a sequence anomaly detection method in an embodiment of the disclosure, and as shown in fig. 1, the sequence anomaly detection method provided in the embodiment of the disclosure includes the following steps:
s102, converting a sequence to be detected into an image to obtain a sequence image to be detected;
In this step, the sequence to be measured may be a non-stationary time sequence, where, in order to improve the generalization capability of the model, the model has a stretching invariance, and when the sequence to be measured is converted into an image, a random function randomly generates a time sequence conversion length, and then the sequence to be measured is converted into an image, so as to obtain an image of the sequence to be measured.
S104, acquiring a first local area of an image of a sequence to be detected based on a convolutional neural network;
in the step, a convolutional neural network (Convolutional Neural Networks, CNN) is a feedforward neural network with a depth structure comprising convolutional calculation, is one of representative algorithms of deep learning, and is used for inputting images of a sequence to be tested into the convolutional neural network trained in advance, and outputting a first local area of the images of the sequence to be tested by the convolutional neural network so as to obtain the first local area of the sequence to be tested. S106, acquiring a second local area of the image of the sequence to be detected based on a selective search algorithm;
in the step, a selective search algorithm is used for providing candidate areas for images, the processing process of the selective search algorithm comprises target detection and target recognition, a sequence image to be detected is processed by the selective search algorithm, target features in the sequence image to be detected are detected first, and the positions of the target features of the sequence image to be detected in the image are recognized so as to obtain a second local area with the target features in the sequence image to be detected.
S108, inputting the first local area and the second local area into a pre-trained classification network VGGnet to obtain a classification result of the sequence to be detected, wherein the classification result is a normal sequence or an abnormal sequence.
In the step, the first local area and the second local area are respectively input into a classification network VGGnet to obtain a plurality of classification results of the images of the sequence to be detected, the plurality of classification results are combined, and weights are set to obtain the classification results of the images of the sequence to be detected, namely the classification results of the sequence to be detected.
The sequence abnormality detection method provided in the embodiment of the disclosure comprises the following steps: converting the sequence to be detected into an image to obtain a sequence image to be detected; acquiring a first local area of an image of a sequence to be detected based on a convolutional neural network; based on a selective search algorithm, a second local area of the image of the sequence to be detected is obtained; inputting the first local area and the second local area into a pre-trained classification network VGGnet to obtain a classification result of the sequence to be detected, wherein the classification result is a normal sequence or an abnormal sequence. The method and the device complete the abnormal detection of the sequence to be detected by converting the sequence to be detected into pictures and adopting a weak supervision learning mode and an attention mechanism and by positioning and identifying, and utilize the visually significant characteristics of the abnormal sequence and the normal sequence without considering the statistical difference of the abnormal sequence and the normal sequence. Meanwhile, the abnormal sequence is detected more accurately without depending on limiting factors such as a threshold value and a limit value.
In an embodiment of the present disclosure, before converting the sequence to be detected into the image, a sliding average filtering process may be further performed on the sequence to be detected, referring to a flowchart of another sequence anomaly detection method shown in fig. 2, a sliding average filtering process may be performed on the sequence to be detected according to steps disclosed in fig. 2, and may specifically include:
s202, using continuous N sampling values as a queue, wherein the length of the queue is fixed to be N;
in the step, the length of the queue is set to be N, N sampling values which are continuously collected are used as a queue, wherein N can be different values, and the method can be suitable for detecting and identifying the abnormality of the periodic sequence with any starting point and any length.
S204, when new data are acquired, the acquired new data are put at the tail of the team, and the data at the head of the team are discarded;
in the step, according to the first-in first-out principle, every time a new sampling value is acquired, the acquired sampling value is placed at the tail of the queue, and the head data of the queue are discarded.
S206, carrying out arithmetic average operation on N data in the queue to obtain a sequence to be tested.
In the step, every time sampling is carried out, arithmetic average operation is carried out on N data of the queue, and a new average value is obtained through calculation, so that the data processing speed is increased, and the sequence to be detected is obtained.
According to the method, the device and the system, the sequence to be detected is subjected to the sliding average value filtering treatment so as to effectively inhibit periodic interference, and the sequence after the sliding average value filtering has higher smoothness.
In one embodiment of the present disclosure, the sequence to be measured may be a periodic time sequence. Wherein the periodic time sequence is a sequence exhibiting a wave or oscillatory type of variation around a long-term trend.
In one embodiment of the present disclosure, the step of acquiring the first local area may refer to the step disclosed in fig. 3, and the step of acquiring the first local area may include:
s302, generating a score graph of a sequence image to be detected by utilizing feature mapping in a convolutional neural network, wherein the score graph consists of a first convolutional layer, a second convolutional layer and a spatial softmax layer;
in the step, the input layer of the convolutional neural network is composed of 32×32 sensing points and is used for receiving the sequence images to be tested, the convolutional neural network further comprises a feature extraction layer and a feature mapping layer, wherein the feature mapping layer can be used for generating a score map of the sequence images to be tested, and the score map is composed of a first convolutional layer, a second convolutional layer and a spatial softmax layer.
S304, the first convolution layer uses 64 3×3 kernels and the second convolution layer uses 1 3×3 kernels to generate a single-channel confidence map;
in this step, a single-channel confidence map is generated by using the first convolution layer and the second convolution layer included in the score map, and specifically, the first convolution layer may use 64 kernels of 3×3 and the second convolution layer may use 1 kernel of 3×3 to generate a single-channel confidence map by using a combined action.
S306, the space softmax layer acts on the confidence map to convert the confidence score into region probability;
in this step, the spatial softmax layer acts on the confidence map generated by the first convolution layer and the second convolution layer, and the confidence score included in the confidence map may be converted into a region probability, where the region probability is used to indicate whether the region corresponding to the probability is a target region.
S308, taking the region with the region probability exceeding the preset region probability as the first local region of the sequence image to be detected, and acquiring the first local region of the sequence image to be detected.
In the step, the region probability of all the regions in the confidence map is compared with the preset region probability, and when the region probability of a certain region is larger than the preset region probability, the region is used as a target region, namely a first local region, and the region is output, so that the first local region of the region image to be detected is obtained.
In one implementation of the present disclosure, after converting the sequence to be detected into the image, an image pyramid may be further constructed on the converted image, and an image with the highest resolution is selected from the image pyramid as the sequence image to be detected, and referring to a flowchart of another sequence anomaly detection method disclosed in fig. 4, the step of obtaining the sequence image to be detected from the image pyramid may include:
s402, generating an image pyramid of an image converted by a sequence to be detected based on an image algorithm;
s404, selecting the image with the highest resolution as the image of the sequence to be detected in the image golden sub-tower.
According to the method, the image pyramid of the image converted by the sequence to be detected is constructed based on an image algorithm, and the image with high resolution is selected from the image pyramid to serve as the image of the sequence to be detected, so that the accuracy of subsequent abnormality detection and identification is improved.
In one implementation of the present disclosure, the step of acquiring the second local area of the image of the sequence to be detected based on the selective search algorithm may refer to a flowchart of another sequence anomaly detection method disclosed in fig. 5, and the step may include:
s502, carrying out image segmentation on the images to be sequenced to obtain a region set of the images to be sequenced;
In the step, image segmentation is carried out on the images to be sequenced according to a selective search algorithm, wherein the image segmentation can be equal segmentation or unequal segmentation, and the segmented image areas of the sequences to be sequenced are arranged to obtain an area set of the images of the sequences to be sequenced.
S504, calculating the similarity between adjacent areas in the area set, and combining two areas with the highest similarity between the adjacent areas;
in the step, the similarity of all adjacent two areas in the area set of the sequence image to be detected is calculated, and the adjacent two areas with the highest similarity are combined to generate a new area.
S506, repeatedly executing the previous steps until the images are combined into a complete image of the sequence to be detected and then stopping;
in the step, the steps of adjacent region similarity calculation and highest similarity region combination are repeatedly executed to obtain new regions generated in the combination process, and the combination is stopped after the images are combined into complete images of the images to be tested in sequence, so that all the new regions generated in the combination process can be obtained.
S508, obtaining a second local area of the images of the sequence to be detected from the images generated in the merging process.
In this step, a second partial region of the image of the sequence to be measured is acquired from the new region image generated in the merging process.
In one embodiment of the present disclosure, training the classification network VGGnet may be accomplished with reference to the steps disclosed in fig. 6 to obtain a pre-trained classification network VGGnet, such as the flowchart of another sequence anomaly detection method shown in fig. 6, which may include:
s602, acquiring a normal sequence and an abnormal sequence with random sequence lengths;
s604, converting a normal sequence and an abnormal sequence with random sequence length into images to obtain a sample training set;
s606, taking the sample training set as input, taking the classification result of the images in the sample training set as output, and training the classification network VGGnet to obtain the classification network VGGnet.
According to the method and the device, the normal sequence and the abnormal sequence with the random sequence length are obtained and converted into the image to serve as a sample training set, so that the classification network VGGnet obtained through training can conduct anomaly detection and identification on the periodic sequence with any starting point and any length, and the anomaly detection capability of a training model is improved.
In one embodiment of the present disclosure, the present disclosure further provides a further sequence anomaly detection method, referring to a flowchart of the further sequence anomaly detection method disclosed in fig. 7, which may include the steps of:
Step 1, acquiring a periodic time sequence;
step 2, carrying out moving average filtering processing on the obtained periodic time sequence;
step 3, converting the processed periodic time sequence into an image;
step 4, positioning key points of the image by using a convolution operator, and performing fine-grained feature learning on the image by using a Selective Search;
step 5, performing joint learning and cascade classification on the two types of features obtained in the step 4 by using a classification network VGGnet;
and 6, carrying out abnormal recognition to obtain a recognition result.
Referring to a schematic diagram of a moving average filtering process shown in fig. 8, it can be clearly seen that, after the moving average filtering process is performed on the image in step 2, the visual characteristics of the processed image are more obvious, the periodic interference of the image before processing is removed, and the processed image has higher smoothness.
Referring to a schematic diagram of obtaining a keypoint by using a convolution operator shown in fig. 9, it can be seen that step 4 locates the keypoint of the image by using the convolution operator, and the convolution obtains the position of the keypoint.
Referring to a schematic diagram of a local area obtained by a Selective Search algorithm shown in fig. 10, it can be seen that, in step 4, fine-grained feature learning is performed on an image by using a Selective Search method, and the position of a fine-grained recognition object is extracted by using a Selective Search method.
Referring to a schematic diagram of a sample collection set shown in fig. 11, after a moving average filtering process is performed on a periodic time sequence, a periodic time sequence with random length is obtained and converted into an image, the obtained image is added into the sample collection, and the sample collection is used as an input in a training process of a classification network VGGnet, so that the classification network VGGnet performs anomaly detection and identification on the periodic sequence with any starting point and any length.
The sequence anomaly detection method of the present disclosure is the same as the technical problem solved by the sequence anomaly detection method, and the technical effects achieved are the same, and are not described in detail herein.
Based on the same inventive concept, a sequence anomaly detection device is also provided in the embodiments of the present disclosure, such as the following embodiments. Since the principle of solving the problem of the embodiment of the device is similar to that of the embodiment of the method, the implementation of the embodiment of the device can be referred to the implementation of the embodiment of the method, and the repetition is omitted.
Fig. 12 is a schematic diagram of a sequence anomaly detection apparatus according to an embodiment of the disclosure, as shown in fig. 12, the apparatus includes:
the sequence conversion image module 1210 is configured to convert the preprocessed sequence to be detected into an image, so as to obtain a sequence image to be detected;
A first region acquiring module 1220, configured to acquire a first local region of the image of the sequence to be detected based on the convolutional neural network;
a second region acquiring module 1230, configured to acquire a second local region of the image of the sequence to be detected based on a selective search algorithm; and
the classification result obtaining module 1240 is configured to input the first local area and the second local area to a pre-trained classification network VGGnet to obtain a classification result of the sequence to be tested, where the classification result is a normal sequence or an abnormal sequence.
In one embodiment of the disclosure, the apparatus further includes a moving average filtering module, where the moving average filtering module is configured to use N consecutive sampling values as a queue, and a length of the queue is fixed to N; when new data is acquired, the acquired new data is put at the tail of the team, and the data at the head of the team is discarded; and carrying out arithmetic average operation on N data in the queue to obtain a sequence to be tested.
In one embodiment of the disclosure, the first region obtaining module 1220 is specifically configured to generate a score map of an image of a sequence to be measured by using feature mapping in a convolutional neural network, where the score map is composed of a first convolutional layer, a second convolutional layer, and a spatial softmax layer; the first convolution layer uses 64 3 x 3 kernels and the second convolution layer uses 1 3 x 3 kernel to generate a single channel confidence map; the space softmax layer acts on the confidence map to convert the confidence score into region probability; and taking the region with the region probability exceeding the preset region probability as a first local region of the sequence image to be detected, and acquiring the first local region of the sequence image to be detected.
In one embodiment of the disclosure, the apparatus further includes an image algorithm processing module, configured to generate an image pyramid of the image converted by the sequence to be tested based on the image algorithm; and selecting the image with the highest resolution ratio from the image golden sub-towers as the image of the sequence to be detected.
In one embodiment of the disclosure, the second region obtaining module 1230 is specifically configured to perform image segmentation on the sequence image to be detected to obtain a region set of the sequence image to be detected; calculating the similarity between adjacent areas in the area set, and combining two areas with the highest similarity between the adjacent areas; repeating the previous steps until the images are combined into a complete image of the sequence to be detected and stopping; and acquiring a second local area of the images of the sequence to be detected from the images generated in the merging process.
In one embodiment of the disclosure, the apparatus further includes a classification network training module, where the classification network training module is configured to obtain a normal sequence and an abnormal sequence of a random sequence length; converting the normal sequence and the abnormal sequence with random sequence length into images to obtain a sample training set; and taking the sample training set as input, taking the classification result of the images in the sample training set as output, and training the classification network VGGnet to obtain the classification network VGGnet.
Those skilled in the art will appreciate that the various aspects of the present disclosure may be implemented as a system, method, or program product. Accordingly, various aspects of the disclosure may be embodied in the following forms, namely: an entirely hardware embodiment, an entirely software embodiment (including firmware, micro-code, etc.) or an embodiment combining hardware and software aspects may be referred to herein as a "circuit," module "or" system.
An electronic device 1300 according to such an embodiment of the present disclosure is described below with reference to fig. 13. The electronic device 1300 shown in fig. 13 is merely an example and should not be construed to limit the functionality and scope of use of embodiments of the present disclosure in any way.
As shown in fig. 13, the electronic device 1300 is embodied in the form of a general purpose computing device. The components of the electronic device 1300 may include, but are not limited to: the at least one processing unit 1310, the at least one memory unit 1320, and a bus 1330 connecting the different system components (including the memory unit 1320 and the processing unit 1310).
Wherein the storage unit stores program code that is executable by the processing unit 1310 such that the processing unit 1310 performs steps according to various exemplary embodiments of the present disclosure described in the above section of the "exemplary method" of the present specification. For example, the processing unit 1310 may perform the following steps of the method embodiment: converting the preprocessed sequence to be detected into an image to obtain a sequence image to be detected; acquiring a first local area of an image of a sequence to be detected based on a convolutional neural network; based on a selective search algorithm, a second local area of the image of the sequence to be detected is obtained; and inputting the first local area and the second local area into a pre-trained classification network VGGnet to obtain a classification result of the sequence to be detected, wherein the classification result is a normal sequence or an abnormal sequence.
In one embodiment of the present disclosure, the processing unit 1310 provided by the present disclosure is further configured to: before converting the sequence to be tested into an image, carrying out a sliding average value filtering treatment on the sequence to be tested, wherein the filtering treatment comprises the following steps: taking the continuous N sampling values as a queue, wherein the length of the queue is fixed to be N; when new data is acquired, the acquired new data is put at the tail of the team, and the data at the head of the team is discarded; and carrying out arithmetic average operation on N data in the queue to obtain a sequence to be tested.
In one embodiment of the present disclosure, the processing unit 1310 provided by the present disclosure is further configured to: generating a score graph of the sequence image to be detected by utilizing feature mapping in the convolutional neural network, wherein the score graph consists of a first convolutional layer, a second convolutional layer and a spatial softmax layer; the first convolution layer uses 64 3 x 3 kernels and the second convolution layer uses 13 x 3 kernel to generate a single channel confidence map; the space softmax layer acts on the confidence map to convert the confidence score into region probability; and taking the region with the region probability exceeding the preset region probability as a first local region of the sequence image to be detected, and acquiring the first local region of the sequence image to be detected.
In one embodiment of the present disclosure, the processing unit 1310 provided by the present disclosure is further configured to: generating an image pyramid of an image converted by a sequence to be detected based on an image algorithm; and selecting the image with the highest resolution ratio from the image golden sub-towers as the image of the sequence to be detected.
In one embodiment of the present disclosure, the processing unit 1310 provided by the present disclosure is further configured to: image segmentation is carried out on the images of the sequence to be detected, and a region set of the images of the sequence to be detected is obtained; calculating the similarity between adjacent areas in the area set, and combining two areas with the highest similarity between the adjacent areas; repeating the previous steps until the images are combined into a complete image of the sequence to be detected and stopping; and acquiring a second local area of the images of the sequence to be detected from the images generated in the merging process.
In one embodiment of the present disclosure, the processing unit 1310 provided by the present disclosure is further configured to: acquiring a normal sequence and an abnormal sequence with random sequence length; converting the normal sequence and the abnormal sequence with random sequence length into images to obtain a sample training set; and taking the sample training set as input, taking the classification result of the images in the sample training set as output, and training the classification network VGGnet to obtain the classification network VGGnet.
The storage unit 1320 may include readable media in the form of volatile storage units, such as Random Access Memory (RAM) 13201 and/or cache memory 13202, and may further include Read Only Memory (ROM) 13203.
The storage unit 1320 may also include a program/utility 13204 having a set (at least one) of program modules 13205, such program modules 13205 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of which may include an implementation of a network environment.
Bus 1330 may be a local bus representing one or more of several types of bus structures including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or using any of a variety of bus architectures.
The electronic device 1300 may also communicate with one or more external devices 1340 (e.g., keyboard, pointing device, bluetooth device, etc.), one or more devices that enable a user to interact with the electronic device 1300, and/or any device (e.g., router, modem, etc.) that enables the electronic device 1300 to communicate with one or more other computing devices. Such communication may occur through an input/output (I/O) interface 1350. Also, the electronic device 1300 may communicate with one or more networks such as a Local Area Network (LAN), a Wide Area Network (WAN) and/or a public network, for example, the Internet, through a network adapter 1360. As shown, the network adapter 1360 communicates with other modules of the electronic device 1300 over the bus 1330. It should be appreciated that although not shown, other hardware and/or software modules may be used in connection with electronic device 1300, including, but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, data backup storage systems, and the like.
From the above description of embodiments, those skilled in the art will readily appreciate that the example embodiments described herein may be implemented in software, or may be implemented in software in combination with the necessary hardware. Thus, the technical solution according to the embodiments of the present disclosure may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (may be a CD-ROM, a U-disk, a mobile hard disk, etc.) or on a network, including several instructions to cause a computing device (may be a personal computer, a server, a terminal device, or a network device, etc.) to perform the method according to the embodiments of the present disclosure.
In an exemplary embodiment of the present disclosure, a computer-readable storage medium, which may be a readable signal medium or a readable storage medium, is also provided. On which a program product is stored which enables the implementation of the method described above of the present disclosure. In some possible implementations, various aspects of the disclosure may also be implemented in the form of a program product comprising program code for causing a terminal device to carry out the steps according to the various exemplary embodiments of the disclosure as described in the "exemplary methods" section of this specification, when the program product is run on the terminal device.
More specific examples of the computer readable storage medium in the present disclosure may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
In this disclosure, a computer readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, with readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A readable signal medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Alternatively, the program code embodied on a computer readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
In particular implementations, the program code for carrying out operations of the present disclosure may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device, partly on a remote computing device, or entirely on the remote computing device or server. In the case of remote computing devices, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., connected via the Internet using an Internet service provider).
It should be noted that although in the above detailed description several modules or units of a device for action execution are mentioned, such a division is not mandatory. Indeed, the features and functionality of two or more modules or units described above may be embodied in one module or unit in accordance with embodiments of the present disclosure. Conversely, the features and functions of one module or unit described above may be further divided into a plurality of modules or units to be embodied.
Furthermore, although the steps of the methods in the present disclosure are depicted in a particular order in the drawings, this does not require or imply that the steps must be performed in that particular order or that all illustrated steps be performed in order to achieve desirable results. Additionally or alternatively, certain steps may be omitted, multiple steps combined into one step to perform, and/or one step decomposed into multiple steps to perform, etc.
From the description of the above embodiments, those skilled in the art will readily appreciate that the example embodiments described herein may be implemented in software, or may be implemented in software in combination with the necessary hardware. Thus, the technical solution according to the embodiments of the present disclosure may be embodied in the form of a software product, which may be stored in a non-volatile storage medium (may be a CD-ROM, a U-disk, a mobile hard disk, etc.) or on a network, including several instructions to cause a computing device (may be a personal computer, a server, a mobile terminal, or a network device, etc.) to perform the method according to the embodiments of the present disclosure.
Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This disclosure is intended to cover any adaptations, uses, or adaptations of the disclosure following the general principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
Claims (10)
1. A sequence anomaly detection method, comprising:
converting the sequence to be detected into an image to obtain a sequence image to be detected;
based on a convolutional neural network, a first local area of the sequence image to be detected is obtained by positioning key points of the sequence image to be detected by using a convolutional operator;
based on a selective search algorithm, a second local area of the sequence image to be detected is obtained, and the second local area is obtained by carrying out fine-grained feature learning on the sequence image to be detected by utilizing the selective search algorithm;
inputting the first local area and the second local area into a pre-trained classification network VGGnet to obtain a classification result of the sequence to be detected, wherein the classification result is a normal sequence or an abnormal sequence.
2. The method of sequence anomaly detection of claim 1, wherein prior to the converting the sequence to be detected into an image, the method further comprises: the filtering processing of the sliding average value is carried out on the sequence to be detected, and the method specifically comprises the following steps:
taking the continuous N sampling values as a queue, wherein the length of the queue is fixed to be N;
when new data is acquired, the acquired new data is put at the tail of the team, and the data at the head of the team is discarded;
And carrying out arithmetic average operation on N data in the queue to obtain the sequence to be detected.
3. The method according to claim 1, wherein the sequence to be detected is a periodic time sequence.
4. The method for detecting sequence anomalies according to claim 1, wherein the acquiring a first local area of the sequence image to be detected based on a convolutional neural network includes:
generating a score graph of the sequence image to be detected by utilizing feature mapping in a convolutional neural network, wherein the score graph consists of a first convolutional layer, a second convolutional layer and a spatial softmax layer;
the first convolution layer utilizes 64 3 x 3 kernels and the second convolution layer utilizes 1 3 x 3 kernels to generate a single channel confidence map;
the spatial softmax layer acts on the confidence map to convert confidence scores into region probabilities;
and taking the region with the region probability exceeding the preset region probability as a first local region of the sequence image to be detected, and acquiring the first local region of the sequence image to be detected.
5. The sequence anomaly detection method of claim 1, wherein after the converting the sequence to be detected into an image, the method further comprises:
Generating an image pyramid of the image converted by the sequence to be detected based on an image algorithm;
and selecting the image with the highest resolution ratio from the image golden sub-tower as the image of the sequence to be detected.
6. The method for detecting sequence anomalies according to claim 1, wherein the acquiring, based on a selective search algorithm, a second local area of the sequence image to be detected includes:
performing image segmentation on the images of the sequence to be detected to obtain a region set of the images of the sequence to be detected;
calculating the similarity between adjacent areas in the area set, and combining two areas with highest similarity between the adjacent areas;
repeating the previous steps until the images are combined into a complete image of the sequence to be detected and stopping;
and acquiring a second local area of the images of the sequence to be detected from the images generated in the merging process.
7. The method for detecting a sequence abnormality according to claim 1, characterized in that the method further comprises:
acquiring a normal sequence and an abnormal sequence with random sequence length;
converting the normal sequence and the abnormal sequence with the random sequence length into images to obtain a sample training set;
And taking the sample training set as input, taking a classification result of the image in the sample training set as output, and training the classification network VGGnet to obtain the classification network VGGnet.
8. A sequence anomaly detection device, comprising:
the sequence conversion image module is used for converting the preprocessed sequence to be detected into an image to obtain a sequence image to be detected;
the first region acquisition module is used for acquiring a first local region of the sequence image to be detected based on a convolutional neural network, wherein the first local region is obtained by positioning key points of the sequence image to be detected by using a convolutional operator;
the second region acquisition module is used for acquiring a second local region of the sequence image to be detected based on a selective search algorithm, wherein the second local region is obtained by carrying out fine-grained feature learning on the sequence image to be detected by utilizing the selective search algorithm; and
the classification result obtaining module is used for inputting the first local area and the second local area into a pre-trained classification network VGGnet to obtain a classification result of the sequence to be detected, wherein the classification result is a normal sequence or an abnormal sequence.
9. An electronic device, comprising:
a processor; and
a memory for storing executable instructions of the processor;
wherein the processor is configured to perform the sequence anomaly detection method of any one of claims 1 to 7 via execution of the executable instructions.
10. A computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when executed by a processor, implements the sequence anomaly detection method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111615986.XA CN114255373B (en) | 2021-12-27 | 2021-12-27 | Sequence anomaly detection method, device, electronic equipment and readable medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111615986.XA CN114255373B (en) | 2021-12-27 | 2021-12-27 | Sequence anomaly detection method, device, electronic equipment and readable medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114255373A CN114255373A (en) | 2022-03-29 |
| CN114255373B true CN114255373B (en) | 2024-02-02 |
Family
ID=80795286
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111615986.XA Active CN114255373B (en) | 2021-12-27 | 2021-12-27 | Sequence anomaly detection method, device, electronic equipment and readable medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114255373B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110837769A (en) * | 2019-08-13 | 2020-02-25 | 广州三木智能科技有限公司 | Embedded far infrared pedestrian detection method based on image processing and deep learning |
| CN113743607A (en) * | 2021-09-15 | 2021-12-03 | 京东科技信息技术有限公司 | Training method of anomaly detection model, anomaly detection method and device |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110096933B (en) * | 2018-01-30 | 2023-07-18 | 华为技术有限公司 | Target detection method, device and system |
-
2021
- 2021-12-27 CN CN202111615986.XA patent/CN114255373B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110837769A (en) * | 2019-08-13 | 2020-02-25 | 广州三木智能科技有限公司 | Embedded far infrared pedestrian detection method based on image processing and deep learning |
| CN113743607A (en) * | 2021-09-15 | 2021-12-03 | 京东科技信息技术有限公司 | Training method of anomaly detection model, anomaly detection method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114255373A (en) | 2022-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111914873A (en) | Two-stage cloud server unsupervised anomaly prediction method | |
| WO2022116322A1 (en) | Method and apparatus for generating anomaly detection model, and anomaly event detection method and apparatus | |
| CN111060337B (en) | Running equipment real-time fault diagnosis method based on expert system | |
| EP3767551A1 (en) | Inspection system, image recognition system, recognition system, discriminator generation system, and learning data generation device | |
| Mechefske et al. | Fault detection and diagnosis in low speed rolling element bearings Part II: The use of nearest neighbour classification | |
| KR102300049B1 (en) | Fault diagnosis apparatus and method using frequency analysis based on machine-learning | |
| CN114170184A (en) | Product image anomaly detection method and device based on embedded feature vector | |
| CN111695620A (en) | Method and system for detecting and correcting abnormal data of time sequence of power system | |
| Azzalini et al. | A minimally supervised approach based on variational autoencoders for anomaly detection in autonomous robots | |
| CN112529678B (en) | Financial index time sequence anomaly detection method based on self-supervision discriminant network | |
| Arul et al. | Data anomaly detection for structural health monitoring of bridges using shapelet transform | |
| US9613271B2 (en) | Determining severity of a geomagnetic disturbance on a power grid using similarity measures | |
| CN117131110B (en) | Method and system for monitoring dielectric loss of capacitive equipment based on correlation analysis | |
| CN111008576A (en) | Pedestrian detection and model training and updating method, device and readable storage medium thereof | |
| CN115270860A (en) | Transformer abnormity diagnosis method, system and diagnosis equipment | |
| CN115471216A (en) | Data management method of intelligent laboratory management platform | |
| CN117036732B (en) | Electromechanical equipment detection system, method and equipment based on fusion model | |
| JPWO2012032889A1 (en) | Learning apparatus, learning system, learning method, and learning program for object identification | |
| CN114255373B (en) | Sequence anomaly detection method, device, electronic equipment and readable medium | |
| CN111881978A (en) | Ocean fixed-point buoy equipment fault diagnosis method and device and electronic equipment | |
| CN116167008A (en) | Abnormal positioning method for internet of things sensing cloud data center based on data enhancement | |
| US20230267779A1 (en) | Method and system for collecting and monitoring vehicle status information | |
| US20220187789A1 (en) | Equipment failure diagnosis apparatus, equipment failure diagnosis method, smart factory system and application agent | |
| CN113033397A (en) | Target tracking method, device, equipment, medium and program product | |
| CN114662541A (en) | Fault diagnosis model construction method and device and electronic equipment |
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 |