Data anomaly detection method, device, equipment and storage medium
By updating the abnormal threshold and Gaussian mixed model clustering in real time, the problem that fixed thresholds are difficult to adapt to data characteristics changes is solved, and the accuracy and efficiency of data anomaly detection are improved.
Patent Information
- Application Number
- CN202010763541.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-07-31
- Publication Date
- 2024-11-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the data abnormality detection, it is difficult to adapt to data characteristics changes in fixed abnormality thresholds, resulting in misjudgment and misreport, affecting detection efficiency.
By updating the abnormal threshold in real time, based on the abnormal probability and data characteristics of the operating state information of the first N moments, the abnormal threshold at the N+1th moment is dynamically determined, and data clustering is used for data clustering to determine the abnormal probability threshold.
It improves the accuracy of abnormal detection, reduces false alarms and missed alarms, improves functional maintenance efficiency, and can adapt to changes in data characteristics.
Smart Images

Figure CN111913850B_ABST
Abstract
Description
Technical Field
[0001] Embodiments of the present disclosure relate to data processing technology, and more particularly to a data anomaly detection method, apparatus, device, and storage medium. Background Art
[0002] With the advent of the big data era, data analysis and processing have become basic work in many industries. By analyzing and processing large amounts of data, a lot of effective information can be obtained.
[0003] In terms of software and hardware functional maintenance, it is difficult for operation and maintenance personnel to directly obtain the operating status of the maintenance object. Generally, the operating status is indirectly analyzed and judged through data that can represent the operating status. The judgment logic is generally to set an abnormality threshold for possible abnormalities. When the detection data is greater than or equal to this abnormality threshold, it is considered that the operation is abnormal.
[0004] However, the setting standards for abnormal thresholds in different industries and even in different business scenarios of the same industry are not the same. Even in the same business scenario, the data characteristics of different time periods may be different, and the corresponding abnormal thresholds must be different. Therefore, the method of directly setting abnormal thresholds often causes misjudgment, resulting in low function maintenance efficiency. Summary of the invention
[0005] The embodiments of the present disclosure provide a data anomaly detection method, apparatus, device and storage medium, which adapt to changes in data characteristics by updating anomaly thresholds in real time, thereby reducing false positives and false negatives of abnormal data and improving anomaly detection efficiency and function maintenance efficiency.
[0006] In a first aspect, an embodiment of the present disclosure provides a data anomaly detection method, comprising:
[0007] Obtain the abnormal probability corresponding to the running status information at N moments;
[0008] According to the abnormal probabilities corresponding to the running status information at the N moments, an abnormal threshold corresponding to the N+1th moment is determined, and the abnormal threshold is used to determine whether the running status information at the N+1th moment is abnormal.
[0009] In a second aspect, an embodiment of the present disclosure provides a data anomaly detection device, comprising:
[0010] An acquisition module is used to obtain the abnormal probability corresponding to the running status information at N moments;
[0011] The processing module is used to determine the abnormal threshold corresponding to the N+1th moment according to the abnormal probability corresponding to the running status information at the N moments, and the abnormal threshold is used to determine whether the running status information at the N+1th moment is abnormal.
[0012] In a third aspect, an embodiment of the present disclosure provides a data anomaly detection device, including:
[0013] A memory for storing program instructions;
[0014] The processor is used to call and execute the program instructions in the memory to perform the method described in the first aspect.
[0015] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium, wherein the computer storage medium stores a computer program, and when the computer program is executed by a processor, the method described in the first aspect is implemented.
[0016] In a fifth aspect, an embodiment of the present disclosure provides a program product, comprising a computer program, wherein the computer program is stored in a readable storage medium, and a processor of a data anomaly detection device can read the computer program from the readable storage medium, and the processor executes the computer program so that the data anomaly detection device implements the method described in the first aspect.
[0017] The embodiments of the present disclosure provide a data anomaly detection method, device, equipment and storage medium. The method includes: obtaining the anomaly probability corresponding to the operating status information at N moments; determining the anomaly threshold corresponding to the N+1th moment according to the anomaly probability corresponding to the operating status information at the N moments, and the anomaly threshold is used to determine whether the operating status information at the N+1th moment is abnormal. The method of the embodiment of the present disclosure determines the anomaly threshold at the N+1th moment according to the anomaly probability of the operating status information at the first N moments, and the determined anomaly threshold is more accurate based on the continuity of the data and the stability of the data characteristics. And the anomaly threshold at each moment is determined in real time based on the operating status information at the first N moments, which can adapt to the changes in data characteristics and improve the accuracy. A more accurate anomaly threshold can reduce the underreporting and false alarm of abnormal data, and improve the efficiency of anomaly detection and function maintenance. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0019] Figure 1 A scene graph provided for an embodiment of the present disclosure;
[0020] Figure 2 A flowchart of a data anomaly detection method provided in an embodiment of the present disclosure;
[0021] Figure 3 A flowchart of a data anomaly detection method provided in an embodiment of the present disclosure;
[0022] Figure 4 A flowchart of a data anomaly detection method provided in an embodiment of the present disclosure;
[0023] Figure 5 A schematic diagram of the structure of a data anomaly detection device provided in an embodiment of the present disclosure;
[0024] Figure 6 A schematic diagram of the structure of a data anomaly detection device provided in an embodiment of the present disclosure;
[0025] Figure 7 A schematic diagram of the structure of a data anomaly detection device provided in an embodiment of the present disclosure;
[0026] Figure 8 A schematic diagram of the structure of a data anomaly detection device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION
[0027] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all of the embodiments. Based on the embodiments in the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the embodiments of the present disclosure.
[0028] With the advent of the big data era, data analysis and processing have become basic work in many industries, and anomaly detection of real-time data has become an important part of data analysis. Currently, there are many methods for anomaly detection of real-time data, but generally an anomaly probability is calculated for a certain real-time data, an anomaly probability threshold is set, and the real-time data is judged to be abnormal by comparing the anomaly probability with the anomaly probability threshold.
[0029] In related technologies, the anomaly probability threshold is often set to a fixed value. However, the anomaly probability threshold setting standards in different business scenarios are not the same. Even in the same business scenario, the data characteristics of different time periods are not the same, and the corresponding anomaly probability thresholds must be different. Therefore, a fixed threshold is difficult to meet the needs of adapting to data changes, which will cause incorrect judgment of data anomalies and affect the detection results.
[0030] Therefore, the embodiment of the present disclosure proposes a data anomaly detection method based on a dynamic threshold, which first determines the size of the anomaly threshold dynamically in real time based on the data characteristics, and then uses it to determine whether the data is abnormal. Since the determination of the threshold size is based on the real-time data characteristics, the abnormality of the data can be indicated more accurately, thereby reducing the false detection rate.
[0031] Figure 1 A scene graph is provided for an embodiment of the present disclosure. Figure 1 As shown, the solution of the embodiment of the present disclosure can be used to detect the operating status of the target device 101. The server 102 is connected to the target device 101 and can perform data transmission. The server 102 obtains data of the target device 101 that can mark its operating status information, and determines whether the operating status of the target device 101 is normal by detecting abnormalities in the data.
[0032] The operating status of the target device 101 may be a hardware status or a built-in software status.
[0033] In the abnormality detection process, the method of determining the abnormality threshold and the method of judging whether there is an abnormality or not can be specifically referred to the following embodiments.
[0034] Figure 2 The flowchart of a data anomaly detection method provided by an embodiment of the present disclosure. The execution subject of the method can be a computer or a server, etc. Figure 2 As shown, the method of this embodiment may include:
[0035] S201. Obtain abnormal probabilities corresponding to operating status information at N moments.
[0036] The N moments may be N consecutive moments. The "consecutive moments" described in the embodiments of the present disclosure are only used to indicate the continuity of the time when the running status information is generated and obtained. Since time is continuous, there must be intervals between the moments when the running status information is obtained.
[0037] In order to make the acquisition of the operating status information as continuous as possible, the acquisition time of the operating status information can be determined according to the generation time of the operating status information. For example, in the process of detecting abnormalities in the operating status of the ambient temperature sensor, the ambient temperature data detected and fed back by the ambient temperature sensor can be used as a representation of the operating status information. The ambient temperature sensor detects the ambient temperature at a certain frequency during operation, so the acquisition of the operating status information can also be performed at this frequency. In this way, the operating status information acquired at N moments is the most continuous data in time.
[0038] In addition, the running status information may be one data or multiple data.
[0039] The "abnormal probability" mentioned in the embodiments of the present disclosure refers to the probability that the operating status information is abnormal data. Generally, the operating status information may not be a definite value, but a range of values. That is, when the operating status information is within a certain range, it can be considered as normal data. And this range of values may not be fixed. Therefore, judging one or some operating status information as "abnormal" is not absolute, but has a certain probability, that is, the abnormal probability. The "abnormal probability" here is only used to indicate the possibility that each data is abnormal data or normal data, and is not limited to how likely it is to be abnormal data. Its specific parameters are determined by the actual abnormality detection method.
[0040] The operating status information at N moments in this embodiment is data that has been detected to determine whether it is abnormal, and the abnormal probability corresponding to each data has been calculated and can be directly obtained in the subsequent process to determine the abnormal threshold.
[0041] Regarding the determination method of abnormal probability, the abnormal probability can be the probability that a certain operating status information is abnormal relative to the data under the same historical situation, or it can be the probability that a certain operating status information is abnormal relative to the predicted data obtained based on the operating status information at the previous moment. In fact, there are many specific methods for determining abnormal probability, and one is listed here, but its specific implementation method is not limited.
[0042] In some embodiments, the method of obtaining the abnormal probability corresponding to the running status information at a certain moment may specifically include: calculating the estimated value of the running status information at the moment according to the running status information at N moments before the moment; obtaining the true value of the running status information at the moment; determining the abnormal probability of the running status information at the moment according to the estimated value of the running status information at the moment and the true value of the running status information at the moment. The calculation of the abnormal probability corresponding to the running status information at N moments before the moment may also follow this method.
[0043] The operation status information at N moments can be obtained by a sliding window. Specifically, the operation status information generated at historical moments can form a time series, and N continuous data in the same sliding window are obtained, which is the operation status information at N moments.
[0044] The sliding window slides with a preset step length, the preset step length is K data, and K is an integer greater than or equal to 1. Preferably, the sliding speed of the sliding window can be determined according to the generation speed of the running status information, so that the running status information generated at the current moment is always located at the next data position of the window.
[0045] S202: Determine an abnormal threshold corresponding to the N+1th moment according to the abnormal probabilities corresponding to the running status information at N moments.
[0046] The abnormal threshold is used to determine whether the running status information at the N+1th moment is abnormal.
[0047] The abnormal threshold is a value used to determine whether the operating status information is normal or not. It is essentially a threshold of the abnormal probability.
[0048] Based on the operating status information of the previous N moments, the abnormal threshold at the N+1th moment can be determined. Compared with the method of manually setting the threshold, it is more accurate with relevant data as the basis.
[0049] The data anomaly detection method of this embodiment includes: obtaining the anomaly probability corresponding to the operating status information at N moments; determining the anomaly threshold corresponding to the N+1th moment according to the anomaly probability corresponding to the operating status information at N moments, and the anomaly threshold is used to determine whether the operating status information at the N+1th moment is abnormal. The method of this embodiment determines the anomaly threshold at the N+1th moment according to the anomaly probability of the operating status information at the first N moments, and the determined anomaly threshold is more accurate based on the continuity of the data and the stability of the data characteristics. And the anomaly threshold at each moment is determined in real time based on the operating status information at the first N moments, which can adapt to the changes in data characteristics and improve the accuracy. Moreover, this anomaly detection method, due to its dependence on historical data characteristics, can effectively identify periodic sudden increases and decreases while improving detection efficiency, thereby reducing misjudgments.
[0050] In some embodiments, the above method may also include: determining the abnormal probability corresponding to the operating status information at the N+1th moment; determining whether the operating status information at the N+1th moment is abnormal based on the abnormal probability and abnormal threshold corresponding to the operating status information at the N+1th moment.
[0051] According to the determined abnormal threshold at the N+1th moment, the abnormality of the running status information at the N+1th moment can be judged. Because the accuracy of the abnormal threshold is improved, the correctness of the judgment of the abnormality of the running status information at the N+1th moment will also be improved, thereby reducing the underreporting and false alarm of abnormal data, and improving the efficiency of abnormal detection and function maintenance.
[0052] The method for determining the abnormal probability corresponding to the operating status information at the N+1th moment can refer to the method listed in the above S101. Specifically, it can include: calculating the estimated value of the operating status information at the N+1th moment according to the operating status information at N moments and the data characteristics of the operating status information at N moments; obtaining the true value of the operating status information at the N+1th moment; determining the abnormal probability corresponding to the operating status information at the N+1th moment according to the estimated value of the operating status information at the N+1th moment and the true value of the operating status information at the N+1th moment.
[0053] As continuous data, its data features should be correlated. Therefore, based on the operating status information and data features of the first N moments, the estimated value of the operating status information at the N+1th moment can be obtained. The mathematical expression of the difference between the true value and the estimated value of the operating status information at the N+1th moment is the abnormal probability corresponding to the operating status information at the N+1th moment.
[0054] In a specific implementation, the above-mentioned determination of the abnormal threshold corresponding to the N+1th moment based on the abnormal probability corresponding to the operating status information at N moments may include: determining the data characteristics of the operating status information at N moments; clustering the operating status information at N moments into M categories of data based on the data characteristics of the operating status information at N moments, each category of data including at least one data in the operating status information at N moments, M being an integer greater than 1; determining the center point data of each category of data; determining the abnormal threshold corresponding to the N+1th moment based on the abnormal probability of the center point data of each category of data in the M categories of data.
[0055] It is understandable that, in the running status information at N moments, all may be detected as normal data, some may be detected as abnormal data, or even all may be detected as abnormal data. This has a certain relationship with the original data value of each running status information and its corresponding abnormal probability, and even the corresponding abnormal threshold. However, not every abnormal probability of data can be used as a reference for the abnormal threshold at the next moment. The process of selecting the abnormal probability that can affect the abnormal threshold at the next moment is performed in a clustering manner in this embodiment. The running status information at N moments is clustered into M categories. There are differences between abnormal data and normal data, and they will definitely be clustered into different classes. The clustering process will distinguish abnormal data from normal data, which is convenient for subsequent processing.
[0056] The cluster center is the point in a cluster that is most representative of the data characteristics. Therefore, the center point of each type of data can be determined as the representative data of a cluster.
[0057] The specific clustering method may be a Gaussian Mixture Model (GMM). Specifically, the above-mentioned clustering of the operating status information at N moments into M types of data based on the data features of the operating status information at N moments includes: inputting the data features of the operating status information at N moments into a pre-trained Gaussian mixture model; clustering the operating status information at N moments into M types of data.
[0058] The Gaussian mixture model can be trained using historical operating status information data and its data features. These data features include at least one of the following: data year-on-year value, data month-on-month value, data statistics, and data Fourier transform value. Data year-on-year value refers to the value of the previous period; data month-on-month value refers to the value in the previous time window; data statistics include data distribution characteristics, etc.
[0059] In some possible implementations, the running status information at the N+1th moment may be data in the current sliding window or data of the next data bit of the sliding window. When the running status information at the N+1th moment is data in the current sliding window, the N data acquired in the current window include the running status information at the N+1th moment, because its data properties are unknown and it may not participate in the clustering process; when the running status information at the N+1th moment is data of the next data bit of the sliding window, the N data acquired in the current window do not include the running status information at the N+1th moment, and the N data are data that have been detected for abnormality.
[0060] Optionally, based on the abnormal probability of the center point data of each type of data in the M types of data, determining the abnormal threshold corresponding to the N+1th moment includes: determining the maximum abnormal probability from the abnormal probability of the center point data of each type of data in the M types of data; and determining the abnormal threshold corresponding to the N+1th moment based on the maximum abnormal probability.
[0061] There are M center point data in M categories of data. The abnormal probabilities corresponding to the M center point data are obtained, and the abnormal probability threshold can be determined according to the M abnormal probabilities.
[0062] In some embodiments, determining the abnormal threshold corresponding to the N+1th moment according to the maximum abnormal probability may include: determining the maximum abnormal probability as the abnormal threshold corresponding to the N+1th moment.
[0063] In some cases, when there is no abnormal data in the sliding window, according to the method of the above embodiment, the abnormal probabilities corresponding to the M cluster centers may be too low, so the abnormal threshold determined will also be too small. Then, normal data can easily be judged as abnormal data because the abnormal probability is greater than the abnormal threshold, resulting in misjudgment.
[0064] Similarly, in other cases, when there are many abnormal data in the sliding window, according to the method of the above embodiment, the abnormal probabilities corresponding to the M cluster centers may be too high, so the abnormal threshold determined will also be too large. Then, the abnormal data may be easily judged as normal data because the abnormal probability is less than the abnormal threshold, resulting in missed judgment of abnormal data, thereby affecting the detection efficiency.
[0065] In order to avoid the above two situations, the size of the abnormal probability threshold can be further limited.
[0066] In some embodiments, if the maximum abnormal probability is greater than a first preset value, the first preset value is determined as the abnormal threshold corresponding to the N+1th moment.
[0067] In other embodiments, if the maximum abnormal probability is less than the second preset value, the second preset value is determined as the abnormal threshold corresponding to the N+1th moment.
[0068] In other embodiments, the first preset value is the upper limit value of the abnormal probability threshold. If the maximum abnormal probability is greater than the first preset threshold, the first preset value is directly determined as the abnormal probability threshold; correspondingly, the second preset value is the lower limit value of the abnormal probability threshold. If the maximum abnormal probability is less than the first preset threshold, the second preset value is directly determined as the abnormal probability threshold; only when the maximum abnormal probability is between the first preset value and the second preset value, the maximum abnormal probability is determined as the abnormal probability threshold.
[0069] While dynamically adjusting the size of the anomaly probability threshold in real time according to data characteristics, constraining the upper and lower limits of the anomaly probability threshold can further reduce false positives and false negatives of abnormal data and improve the efficiency of anomaly detection.
[0070] In some embodiments, whether the running status information at the N+1th moment is abnormal is determined based on the abnormal probability and abnormal threshold corresponding to the running status information at the N+1th moment, including: if the abnormal probability corresponding to the running status information at the N+1th moment is greater than the abnormal threshold, then the N+1th data is determined to be abnormal data; if the abnormal probability corresponding to the running status information at the N+1th moment is less than or equal to the abnormal threshold, then the N+1th data is determined to be normal data.
[0071] In some embodiments, the above method further includes: if it is determined that the operating status information at the N+1th moment is abnormal, outputting an alarm message.
[0072] When the operating status information at the N+1th moment is determined to be abnormal data, an alarm message is output for corresponding processing. Especially in some scenarios with high data requirements, such as the abnormal detection process of temperature sensor data, when the data is abnormal, it means that the sensor is faulty or the temperature of the device or environment to be detected has changed dramatically. These may be more important situations that need to be handled in time, and the output of alarm information is particularly important at this time.
[0073] Figure 3 The flowchart of a data anomaly detection method provided by an embodiment of the present disclosure. The execution subject of the method can be a computer or a server, etc. Figure 3 As shown, the method of this embodiment may include:
[0074] S301. Obtain operation status information at N moments.
[0075] S302: Cluster the running status information at N moments into M types of data, each type of data including at least one of the running status information at the N moments.
[0076] S303: Determine the center point data of each type of data.
[0077] S304: Obtain the abnormal probability of the center point data of each type of data.
[0078] S305 , determining an abnormal threshold value according to the abnormal probability of the central point data of each type of data in the M types of data.
[0079] S306: Obtain the abnormal probability of the running status information at the N+1th moment.
[0080] S307, determine whether the abnormal probability of the running status information at the N+1th moment is greater than the abnormal threshold, if so, execute S308, otherwise execute S309.
[0081] S308: Determine that the running status information at the N+1th moment is abnormal data.
[0082] S309: Determine that the running status information at the N+1th moment is normal data.
[0083] The specific implementation of the method of this embodiment can refer to the description of the above embodiments. By clustering N data, determining the cluster center point, and then determining the abnormal probability threshold according to the abnormal probability of the data at each cluster center point, the size of the abnormal probability threshold can be dynamically adjusted in real time according to the data characteristics, thereby reducing the false positives and false negatives of abnormal data and improving the efficiency of abnormal detection.
[0084] Figure 4The flowchart of a data anomaly detection method provided by an embodiment of the present disclosure. The execution subject of the method can be a computer or a server, etc. Figure 4 As shown, the method of this embodiment may include:
[0085] S401. Obtain operation status information at N moments.
[0086] S402: Cluster the running status information at N moments into M types of data, each type of data including at least one of the running status information at N moments.
[0087] S403: Determine the center point data of each type of data.
[0088] S404: Obtain the abnormal probability of the center point data of each type of data.
[0089] S405 , determining the maximum abnormal probability from the abnormal probabilities of the central point data of each type of data in the M types of data.
[0090] S406: Determine whether the maximum abnormal probability is greater than a first preset value, if so, execute S407, otherwise execute S408.
[0091] S407: Determine the first preset value as the abnormal probability threshold.
[0092] S408. Determine whether the maximum abnormal probability is less than a second preset value. If so, execute S409; otherwise, execute S410.
[0093] S409: Determine the second preset value as the abnormal probability threshold.
[0094] S410: Determine the maximum abnormal probability as the abnormal probability threshold.
[0095] S411. Determine whether the running status information at the N+1th moment is abnormal data according to the abnormal probability threshold.
[0096] The specific implementation of the method of this embodiment can refer to the description of the above embodiments. The size of the abnormal probability threshold can be adjusted dynamically in real time according to the data characteristics. At the same time, the upper and lower limits of the abnormal probability threshold are constrained, which can further reduce the false positives and false negatives of abnormal data and improve the efficiency of abnormal detection.
[0097] The running status information in the embodiments of the present disclosure may be the amount of interface call requests, or the interface call success rate, or sensor data, etc. However, it is not limited to such data, and the method provided by the embodiments of the present disclosure may be used in any scenario where data anomaly detection is required.
[0098] In a specific application process, the above data anomaly detection method can be applied to anomaly detection of relevant operation and maintenance indicator data in the business system, such as: detection of interface request volume data, and detection of interface call request success rate data.
[0099] Take the detection process of interface call request success rate data as an example. The interface call request success rate data is uploaded to the server once every 10 seconds, which is called a point cycle, and contains all interface call request success rate data information within 10 seconds.
[0100] The sliding window length is set to 2 minutes, and data of a total of 12 point periods of 2 minutes within the sliding window are obtained each time.
[0101] The Gaussian mixture model GMM is used to cluster these data. There are multiple models based on Gaussian probability density functions in the Gaussian mixture model, and each data may conform to a different Gaussian distribution. By calculating the possibility of each data existing in each Gaussian model and dividing each data into the Gaussian model with the highest possibility, 10 classes are obtained, and then the cluster center point data of the 10 clusters are determined.
[0102] At the same time, obtain the abnormal probabilities corresponding to the data of the 10 cluster center points, select the largest abnormal probability and compare it with the upper and lower limits of the abnormal probability threshold. If the maximum abnormal probability is between the upper and lower limits, the maximum abnormal probability is determined as the abnormal probability threshold at the next moment. If the maximum abnormal probability is greater than the upper limit, the upper limit is determined as the abnormal probability threshold at the next moment. If the maximum abnormal probability is less than the lower limit, the lower limit is determined as the abnormal probability threshold at the next moment.
[0103] The determined abnormal probability threshold is compared with the abnormal probability of the interface call request success rate data at the next moment, so as to determine whether the interface call request success rate data at the next moment is abnormal data. If it is determined that the interface call request success rate data at the next moment is abnormal data, an alarm message is output to indicate that the corresponding interface call request success rate is low at this time and the system is unstable.
[0104] After an anomaly detection is completed, the sliding window moves backward, the step size can be 1 point period, and the data of 12 point periods in the updated sliding window is obtained again, and the detection process is repeated, and the cycle is repeated.
[0105] Before performing anomaly detection, the GMM model needs to be trained using historical interface call request success rate data to enable it to cluster the interface call request success rate data. The model training process is not described here.
[0106] The calculation process of the abnormal probability can first estimate the value of the interface call request success rate data at the next moment based on the data in the sliding window, and then obtain the actual value of the interface call request success rate data at the next moment. By comparing the actual value with the estimated value, the abnormal probability of this interface call request success rate data can be determined for subsequent judgment. The abnormal probabilities corresponding to other data can be obtained using the same calculation method.
[0107] It is understandable that "abnormal probability" in the embodiments of the present disclosure is just a general term for indicators that characterize the degree of abnormality of data. For example, 100 is set as normal and 0 is set as abnormal. Values in the range of 0 to 1 can also characterize the degree of abnormality of data and can be called abnormal probability. The specific setting method may depend on the calculation method of abnormal probability in the abnormality detection process, but no matter how the value of abnormal probability is limited, as long as it has the ability to characterize the degree of abnormality of data, it belongs to the concept of abnormal probability in the embodiments of the present disclosure.
[0108] The specific calculation process of the abnormal probability is not limited here, and other calculation methods are not described in detail.
[0109] In another specific application process, the above-mentioned data anomaly detection method can also be applied to the anomaly detection of relevant data indicators in the hardware system, such as the detection of temperature sensor data in the vehicle system.
[0110] The temperature sensor detects temperature data in real time and uploads it to the main control system. The main control system uploads the temperature data to the server in chronological order for abnormality detection and analysis.
[0111] The sliding window length can be set to 50 data, and 50 temperature data in the sliding window are obtained each time.
[0112] The Gaussian mixture model (GMM) is used to cluster these data. By calculating the likelihood of each data in each Gaussian model and dividing each data into the Gaussian model with the highest likelihood, five clusters are obtained. Then the cluster center point data of the five clusters are determined.
[0113] At the same time, obtain the abnormal probabilities corresponding to the data of the five cluster center points, select the largest abnormal probability and compare it with the upper and lower limits of the abnormal probability threshold. If the maximum abnormal probability is greater than the upper limit, determine the upper limit as the abnormal probability threshold at the next moment.
[0114] The abnormal probability of the temperature data at the next moment is compared with that of the temperature data at the next moment to determine whether the temperature data at the next moment is abnormal data. If the temperature data at the next moment is determined to be abnormal data, an alarm message is output to indicate that the corresponding temperature sensor is faulty or the vehicle system temperature changes drastically and needs to be adjusted.
[0115] Figure 5 A schematic diagram of a data anomaly detection device provided by an embodiment of the present disclosure is shown in FIG. Figure 5 As shown, the data anomaly detection device 500 of this embodiment may include: an acquisition module 501 and a processing module 502.
[0116] The acquisition module 501 is used to acquire the abnormality probabilities corresponding to the running status information at N moments.
[0117] The processing module 502 is used to determine the abnormal threshold corresponding to the N+1th moment according to the abnormal probability corresponding to the running status information at N moments, and the abnormal threshold is used to determine whether the running status information at the N+1th moment is abnormal.
[0118] Optionally, the processing module 502 is further configured to:
[0119] Determine the abnormal probability corresponding to the operating status information at the N+1th moment;
[0120] Whether the running status information at the N+1th moment is abnormal is determined based on the abnormal probability and the abnormal threshold corresponding to the running status information at the N+1th moment.
[0121] Optionally, when the processing module 502 determines the abnormality threshold corresponding to the N+1th moment according to the abnormality probability corresponding to the running status information at N moments, it is specifically configured to:
[0122] Determine data characteristics of operation status information at N moments;
[0123] According to the data features of the running status information at N moments, the running status information at N moments is clustered into M types of data, each type of data includes at least one data of the running status information at N moments, and M is an integer greater than 1;
[0124] Determine the center point data of each type of data;
[0125] According to the abnormal probability of the central point data of each type of data in the M types of data, the abnormal threshold corresponding to the N+1th moment is determined.
[0126] Optionally, when the processing module 502 clusters the running status information at N moments into M types of data according to the data features of the running status information at N moments, it is specifically configured to:
[0127] Input the data features of the operating status information at N moments into the pre-trained Gaussian mixture model;
[0128] Cluster the running status information at N moments into M types of data.
[0129] Optionally, when the processing module 502 determines the abnormality threshold corresponding to the N+1th moment according to the abnormality probability of the central point data of each type of data in the M types of data, it is specifically used to:
[0130] Determine the maximum abnormal probability from the abnormal probability of the center point data of each type of data in the M types of data;
[0131] According to the maximum abnormal probability, determine the abnormal threshold corresponding to the N+1th moment.
[0132] Optionally, when the processing module 502 determines the abnormality threshold corresponding to the N+1th moment according to the maximum abnormality probability, it is specifically configured to:
[0133] The maximum abnormal probability is determined as the abnormal threshold corresponding to the N+1th moment;
[0134] or,
[0135] If the maximum abnormal probability is greater than the first preset value, the first preset value is determined as the abnormal threshold corresponding to the N+1th moment;
[0136] or,
[0137] If the maximum abnormal probability is less than the second preset value, the second preset value is determined as the abnormal threshold corresponding to the N+1th moment.
[0138] Optionally, when the processing module 502 determines whether the running status information at the N+1th moment is abnormal according to the abnormal probability and the abnormal threshold corresponding to the running status information at the N+1th moment, it is specifically configured to:
[0139] If the abnormal probability corresponding to the running status information at the N+1th moment is greater than the abnormal threshold, the N+1th data is determined to be abnormal data;
[0140] If the abnormal probability corresponding to the running status information at the N+1th moment is less than or equal to the abnormal threshold, the N+1th data is determined to be normal data.
[0141] Optionally, when determining the abnormal probability corresponding to the running status information at the N+1th moment, the processing module 502 is specifically configured to:
[0142] Calculate an estimated value of the running status information at the N+1th moment according to the running status information at the Nth moment and the data features of the running status information at the Nth moment;
[0143] Get the true value of the running status information at the N+1th moment;
[0144] The abnormal probability corresponding to the running status information at the N+1th moment is determined according to the estimated value of the running status information at the N+1th moment and the true value of the running status information at the N+1th moment.
[0145] Optionally, the device further comprises:
[0146] The output module 503 is used to output alarm information when it is determined that the operating status information at the N+1th moment is abnormal.
[0147] Optionally, the acquisition module 501 is specifically used to: obtain the operating status information of N moments in the same sliding window, and the data is sorted by time; wherein the sliding window slides a preset step size at each moment, the preset step size is K data, and K is an integer greater than or equal to 1.
[0148] Optionally, the data features include at least one of the following: a year-on-year data value, a quarter-on-quarter data value, a data statistic, and a data Fourier transform value.
[0149] Optionally, the running status information at the N+1th moment is undetected data generated at the current moment or undetected data generated at the next moment.
[0150] Optionally, the operation status information includes: the amount of interface call requests, or the interface call success rate, or sensor data.
[0151] The device of this embodiment can be used to execute the method of any of the above embodiments. Its implementation principle and technical effects are similar and will not be described in detail here.
[0152] Figure 6 A schematic diagram of a data anomaly detection device provided by an embodiment of the present disclosure is shown in FIG. Figure 6 As shown, the data anomaly detection device 600 of this embodiment may include: a memory 601 and a processor 602.
[0153] The memory 601 is used to store program instructions.
[0154] Processor 602 is used to call and execute program instructions in memory 601 to perform: obtaining the abnormal probability corresponding to the operating status information at N moments; determining the abnormal threshold corresponding to the N+1th moment based on the abnormal probability corresponding to the operating status information at N moments, and the abnormal threshold is used to determine whether the operating status information at the N+1th moment is abnormal.
[0155] Optionally, the processor 602 is further configured to:
[0156] Determine the abnormal probability corresponding to the operating status information at the N+1th moment;
[0157] Whether the running status information at the N+1th moment is abnormal is determined based on the abnormal probability and the abnormal threshold corresponding to the running status information at the N+1th moment.
[0158] Optionally, when the processor 602 determines the abnormality threshold corresponding to the N+1th moment according to the abnormality probability corresponding to the running status information at N moments, it is specifically configured to:
[0159] Determine data characteristics of operation status information at N moments;
[0160] According to the data features of the running status information at N moments, the running status information at N moments is clustered into M types of data, each type of data includes at least one data of the running status information at N moments, and M is an integer greater than 1;
[0161] Determine the center point data of each type of data;
[0162] According to the abnormal probability of the central point data of each type of data in the M types of data, the abnormal threshold corresponding to the N+1th moment is determined.
[0163] Optionally, when the processor 602 clusters the running status information at N moments into M types of data according to the data features of the running status information at N moments, it is specifically configured to:
[0164] Input the data features of the operating status information at N moments into the pre-trained Gaussian mixture model;
[0165] Cluster the running status information at N moments into M types of data.
[0166] Optionally, when the processor 602 determines the abnormality threshold corresponding to the N+1th moment according to the abnormality probability of the central point data of each type of data in the M types of data, it is specifically configured to:
[0167] Determine the maximum abnormal probability from the abnormal probability of the center point data of each type of data in the M types of data;
[0168] According to the maximum abnormal probability, determine the abnormal threshold corresponding to the N+1th moment.
[0169] Optionally, when the processor 602 determines the abnormality threshold corresponding to the N+1th moment according to the maximum abnormality probability, it is specifically configured to:
[0170] The maximum abnormal probability is determined as the abnormal threshold corresponding to the N+1th moment;
[0171] or,
[0172] If the maximum abnormal probability is greater than the first preset value, the first preset value is determined as the abnormal threshold corresponding to the N+1th moment;
[0173] or,
[0174] If the maximum abnormal probability is less than the second preset value, the second preset value is determined as the abnormal threshold corresponding to the N+1th moment.
[0175] Optionally, when the processor 602 determines whether the running status information at the N+1th moment is abnormal according to the abnormal probability and the abnormal threshold corresponding to the running status information at the N+1th moment, it is specifically configured to:
[0176] If the abnormal probability corresponding to the running status information at the N+1th moment is greater than the abnormal threshold, the N+1th data is determined to be abnormal data;
[0177] If the abnormal probability corresponding to the running status information at the N+1th moment is less than or equal to the abnormal threshold, the N+1th data is determined to be normal data.
[0178] Optionally, when determining the abnormal probability corresponding to the running status information at the N+1th moment, the processor 602 is specifically configured to:
[0179] Calculate an estimated value of the running status information at the N+1th moment according to the running status information at the Nth moment and the data features of the running status information at the Nth moment;
[0180] Get the true value of the running status information at the N+1th moment;
[0181] The abnormal probability corresponding to the running status information at the N+1th moment is determined according to the estimated value of the running status information at the N+1th moment and the true value of the running status information at the N+1th moment.
[0182] Optionally, the device further comprises:
[0183] The output module is used to output alarm information when it is determined that the operating status information at the N+1th moment is abnormal.
[0184] Optionally, the processor 602 is specifically used to: obtain operating status information at N moments in the same sliding window, and sort the data by time; wherein the sliding window slides a preset step size at each moment, the preset step size is K data, and K is an integer greater than or equal to 1.
[0185] Optionally, the data features include at least one of the following: a year-on-year data value, a quarter-on-quarter data value, a data statistic, and a data Fourier transform value.
[0186] Optionally, the running status information at the N+1th moment is undetected data generated at the current moment or undetected data generated at the next moment.
[0187] Optionally, the operation status information includes: the amount of interface call requests, or the interface call success rate, or sensor data.
[0188] The data anomaly detection device of this embodiment can be used to execute the method of any of the above embodiments. Its implementation principle and technical effects are similar and will not be described in detail here.
[0189] Figure 7 A schematic diagram of the structure of a data anomaly detection device provided in an embodiment of the present disclosure. For example, the device can be provided as a computer. Figure 7 , the apparatus includes a processing component 701, which further includes one or more processors, and a memory resource represented by a memory 702 for storing instructions, such as an application, that can be executed by the processing component 701. The application stored in the memory 702 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 701 is configured to execute the instructions to perform the above method.
[0190] The device may also include a power supply component 703 configured to perform power management of the device, a wired or wireless network interface 704 configured to connect the device to a network, and an input / output (I / O) interface 705. The device may operate based on an operating system stored in the memory 702, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™ or the like.
[0191] Figure 8 A schematic diagram of the structure of a data anomaly detection device provided in an embodiment of the present disclosure, the device may be a mobile phone, a computer, a tablet device, etc.
[0192] The device 800 may include one or more of the following components: a processing component 802 , a memory 804 , a power component 806 , a multimedia component 808 , an audio component 810 , an input / output (I / O) interface 812 , a sensor component 814 , and a communication component 816 .
[0193] The processing component 802 generally controls the overall operation of the device 800, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the above-mentioned method. In addition, the processing component 802 may include one or more modules to facilitate the interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.
[0194] The memory 804 is configured to store various types of data to support operations on the device 800. Examples of such data include instructions for any application or method operating on the device 800, contact data, phone book data, messages, pictures, videos, etc. The memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.
[0195] The power supply component 806 provides power to the various components of the device 800. The power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the device 800.
[0196] The multimedia component 808 includes a screen that provides an output interface between the device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. When the device 800 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera may receive external multimedia data. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.
[0197] The audio component 810 is configured to output and / or input audio signals. For example, the audio component 810 includes a microphone (MIC), and when the device 800 is in an operating mode, such as a call mode, a recording mode, and a speech recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in the memory 804 or sent via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for outputting audio signals.
[0198] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: a home button, a volume button, a start button, and a lock button.
[0199] The sensor assembly 814 includes one or more sensors for providing various aspects of status assessment for the device 800. For example, the sensor assembly 814 can detect the open / closed state of the device 800, the relative positioning of components, such as the display and keypad of the device 800, the sensor assembly 814 can also detect the position change of the device 800 or a component of the device 800, the presence or absence of user contact with the device 800, the orientation or acceleration / deceleration of the device 800 and the temperature change of the device 800. The sensor assembly 814 can include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 814 can also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 can also include an accelerometer, a gyroscope sensor, a magnetic sensor, a pressure sensor or a temperature sensor.
[0200] The communication component 816 is configured to facilitate wired or wireless communication between the device 800 and other devices. The device 800 can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.
[0201] In an exemplary embodiment, the apparatus 800 may be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to perform the above method.
[0202] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, and the instructions can be executed by the processor 820 of the device 800 to perform the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.
[0203] A non-temporary computer-readable storage medium, when the instructions in the storage medium are executed by the processor of a terminal device, enables the terminal device to execute the above-mentioned terminal device split-screen processing method.
[0204] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.
[0205] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The embodiments of the present disclosure are intended to cover any variations, uses, or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The specification and embodiments are to be regarded as exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.
[0206] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.
Claims
1. A data anomaly detection method, characterized in that: include: Obtain the abnormal probability corresponding to the running status information at N moments; Determining data characteristics of the operating status information at the N moments; Clustering the running status information at the N moments into M categories of data according to data features of the running status information at the N moments, each category of data including at least one data of the running status information at the N moments, where M is an integer greater than 1; Determine the center point data of each type of data; According to the abnormal probability of the central point data of each type of data in the M types of data, the abnormal threshold corresponding to the N+1th moment is determined, and the abnormal threshold is used to determine whether the running status information at the N+1th moment is abnormal.
2. The method according to claim 1, characterized in that Also includes: Determine the abnormal probability corresponding to the operating status information at the N+1th moment; Whether the running status information at the N+1th moment is abnormal is determined according to the abnormal probability corresponding to the running status information at the N+1th moment and the abnormal threshold.
3. The method according to claim 1, characterized in that The clustering of the running status information at the N moments into M types of data according to the data features of the running status information at the N moments includes: Inputting the data features of the operating status information at the N moments into a pre-trained Gaussian mixture model; The running status information at the N moments is clustered into M types of data.
4. The method according to claim 1, characterized in that: The determining, according to the abnormal probability of the central point data of each type of data in the M types of data, the abnormal threshold corresponding to the N+1th moment includes: Determine the maximum abnormal probability from the abnormal probability of the center point data of each type of data in the M types of data; According to the maximum abnormal probability, the abnormal threshold corresponding to the N+1th moment is determined.
5. The method according to claim 4, characterized in that The determining, according to the maximum abnormal probability, an abnormal threshold corresponding to the N+1th moment includes: Determine the maximum abnormal probability as the abnormal threshold corresponding to the N+1th moment; or, If the maximum abnormal probability is greater than a first preset value, the first preset value is determined as the abnormal threshold corresponding to the N+1th moment; or, If the maximum abnormal probability is less than a second preset value, the second preset value is determined as the abnormal threshold corresponding to the N+1th moment.
6. The method according to claim 2, characterized in that The determining whether the running status information at the N+1th moment is abnormal according to the abnormal probability corresponding to the running status information at the N+1th moment and the abnormal threshold value includes: If the abnormal probability corresponding to the running status information at the N+1th moment is greater than the abnormal threshold, determining that the N+1th data is abnormal data; If the abnormal probability corresponding to the running status information at the N+1th moment is less than or equal to the abnormal threshold, it is determined that the N+1th data is normal data.
7. The method according to claim 1, characterized in that The determining of the abnormal probability corresponding to the running status information at the N+1th moment includes: Calculate an estimated value of the running status information at the N+1th moment according to the running status information at the N moments and data features of the running status information at the N moments; Obtaining the true value of the running status information at the N+1th moment; The abnormal probability corresponding to the running status information at the N+1th moment is determined according to the estimated value of the running status information at the N+1th moment and the true value of the running status information at the N+1th moment.
8. The method according to claim 2, characterized in that: Also includes: If it is determined that the operating status information at the N+1th moment is abnormal, an alarm message is output.
9. A data anomaly detection device, characterized in that: include: An acquisition module is used to obtain the abnormal probability corresponding to the running status information at N moments; Processing modules for: Determining data characteristics of the operating status information at the N moments; Clustering the running status information at the N moments into M categories of data according to data features of the running status information at the N moments, each category of data including at least one data of the running status information at the N moments, where M is an integer greater than 1; Determine the center point data of each type of data; According to the abnormal probability of the central point data of each type of data in the M types of data, the abnormal threshold corresponding to the N+1th moment is determined, and the abnormal threshold is used to determine whether the running status information at the N+1th moment is abnormal.
10. The device according to claim 9, characterized in that The processing module is also used for: Determine the abnormal probability corresponding to the operating status information at the N+1th moment; Whether the running status information at the N+1th moment is abnormal is determined according to the abnormal probability corresponding to the running status information at the N+1th moment and the abnormal threshold.
11. The device according to claim 9, characterized in that When the processing module clusters the operation status information at the N moments into M types of data according to the data features of the operation status information at the N moments, it is specifically used to: Inputting the data features of the operating status information at the N moments into a pre-trained Gaussian mixture model; The running status information at the N moments is clustered into M types of data.
12. The device according to claim 9, characterized in that When the processing module determines the abnormality threshold corresponding to the N+1th moment according to the abnormality probability of the central point data of each type of data in the M types of data, it is specifically used to: Determine the maximum abnormal probability from the abnormal probability of the center point data of each type of data in the M types of data; According to the maximum abnormal probability, the abnormal threshold corresponding to the N+1th moment is determined.
13. The device according to claim 12, characterized in that When the processing module determines the abnormal threshold corresponding to the N+1th moment according to the maximum abnormal probability, it is specifically used to: Determine the maximum abnormal probability as the abnormal threshold corresponding to the N+1th moment; or, If the maximum abnormal probability is greater than a first preset value, the first preset value is determined as the abnormal threshold corresponding to the N+1th moment; or, If the maximum abnormal probability is less than a second preset value, the second preset value is determined as the abnormal threshold corresponding to the N+1th moment.
14. The device according to claim 10, characterized in that When the processing module determines whether the running status information at the N+1th moment is abnormal according to the abnormal probability corresponding to the running status information at the N+1th moment and the abnormal threshold, it is specifically used to: If the abnormal probability corresponding to the running status information at the N+1th moment is greater than the abnormal threshold, determining that the N+1th data is abnormal data; If the abnormal probability corresponding to the running status information at the N+1th moment is less than or equal to the abnormal threshold, it is determined that the N+1th data is normal data.
15. The device according to claim 9, characterized in that When determining the abnormal probability corresponding to the running status information at the N+1th moment, the processing module is specifically used to: Calculate an estimated value of the running status information at the N+1th moment according to the running status information at the N moments and data features of the running status information at the N moments; Obtaining the true value of the running status information at the N+1th moment; The abnormal probability corresponding to the running status information at the N+1th moment is determined according to the estimated value of the running status information at the N+1th moment and the true value of the running status information at the N+1th moment.
16. The device according to claim 10, characterized in that Also includes: The output module is used to output alarm information when it is determined that the operating status information at the N+1th moment is abnormal.
17. A data anomaly detection device, characterized in that: include: A memory for storing program instructions; A processor, configured to call and execute program instructions in the memory to perform the method according to any one of claims 1 to 8.
18. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.
Citation Information
Patent Citations
Residual error posterior-based abnormal value online detection and confidence degree assessment method
CN107092582A
An anomaly detection method and apparatus
CN109213654A
Abnormal data detection method, device and equipment and computer readable storage medium
CN109684118A
Dynamic setting method and system of differential threshold for online monitoring of transformers
CN109856299A
Method and device for monitoring abnormal transactions
CN110189228A
