Indicator threshold determination method, apparatus and device, storage medium and program product

Abstract

An indicator threshold determination method, apparatus and device, a storage medium and a program product. The method comprises: performing weighted calculation using feature values having different statistical characteristics on historical indicator values in a historical indicator value sequence and predicted indicator values in a predicted indicator value sequence to obtain a sensitivity value; when the sensitivity value is within a preset threshold interval, extracting from the historical indicator value sequence a plurality of first indicator values satisfying a preset alarm rule, extracting from a predicted threshold sequence a plurality of first predicted thresholds having the same timestamp as the plurality of first indicator values, and performing fitting calculation on the plurality of first indicator values and the plurality of first predicted thresholds to obtain a first tolerance coefficient; and, on the basis of the first tolerance coefficient, adjusting the predicted threshold sequence to obtain a target predicted threshold sequence. The present solution can adaptively adjust thresholds on the basis of trend changes or jitters of different indicator data, thereby providing a reliable indicator threshold for indicator-based early warning.

Description

Methods, devices, equipment, storage media, and program products for determining indicator thresholds

[0001] This application claims priority to Chinese Patent Application No. 202510321702.8, filed on March 18, 2025, entitled “Method, Apparatus, Device, Storage Medium and Program Product for Determining Indicator Thresholds”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of computer technology, and in particular to a method, apparatus, device, storage medium, and program product for determining an index threshold. Background Technology

[0003] With the rapid development of the internet industry, different internet companies have built their own large-scale business systems, which generate massive and diverse metrics data during operation. For these key metrics covering business performance, system performance, and infrastructure status, internet companies urgently need to build an efficient monitoring and anomaly detection mechanism to ensure high service stability and improve the efficiency of operations and maintenance.

[0004] The relevant technologies use dynamic threshold algorithms for indicator monitoring, which can flexibly adjust the threshold setting according to the real-time changes of indicator data. However, due to the diversity of trends and fluctuations among different indicator data, the threshold prediction results of specific dynamic threshold algorithms are relatively fixed and cannot adapt to the trend changes or fluctuations of different indicator data to adjust the threshold. This results in low accuracy of indicator warnings based on the predicted threshold, which is prone to false alarms and missed alarms, and needs to be improved. Summary of the Invention

[0005] This application provides a method, apparatus, device, storage medium, and program product for determining indicator thresholds. It solves the problem that the threshold prediction results of dynamic threshold algorithms in related technologies are relatively fixed and cannot adapt to the trend changes or fluctuations of different indicator data to adjust the threshold. This results in low accuracy of indicator warnings based on the threshold and is prone to false alarms and missed alarms. The method can adapt to the trend changes or fluctuations of different indicator data to adjust the threshold, providing a reliable indicator threshold for indicator warnings and improving the accuracy of warnings.

[0006] In a first aspect, embodiments of this application provide a method for determining an indicator threshold, the method comprising:

[0007] Obtain a historical indicator value sequence, and calculate the historical indicator value sequence according to the set sequence prediction algorithm to obtain a predicted indicator value sequence and a predicted threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values.

[0008] A sensitivity value is obtained by weighting the feature values ​​of the multiple historical index values ​​and the multiple predicted index values ​​with different statistical characteristics. The sensitivity value is used to adjust the tolerance level of the predicted threshold sequence.

[0009] When the sensitivity value is within a preset threshold range, multiple first index values ​​that satisfy preset alarm rules are extracted from the historical index value sequence, and multiple first prediction thresholds with the same timestamp as the multiple first index values ​​are extracted from the prediction threshold sequence. The multiple first index values ​​and the multiple first prediction thresholds are fitted and calculated to obtain a first tolerance coefficient.

[0010] The target prediction threshold sequence is obtained by adjusting the prediction threshold sequence based on the first tolerance coefficient, and is used for real-time abnormal indicator value warning. The target prediction threshold sequence includes multiple indicator thresholds.

[0011] Secondly, embodiments of this application also provide an indicator threshold determination device, the device comprising:

[0012] The prediction sequence determination module is configured to acquire a historical indicator value sequence, calculate the historical indicator value sequence according to a set sequence prediction algorithm, and obtain a prediction indicator value sequence and a prediction threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the prediction indicator value sequence includes multiple prediction indicator values.

[0013] The sensitivity value determination module is configured to perform a weighted calculation of the feature values ​​of the plurality of historical index values ​​and the plurality of predicted index values ​​with different statistical characteristics to obtain a sensitivity value, wherein the sensitivity value is used to adjust the tolerance level of the predicted threshold sequence;

[0014] The tolerance coefficient determination module is configured to extract multiple first index values ​​that satisfy preset alarm rules from the historical index value sequence when the sensitivity value is within a preset threshold range, and to extract multiple first prediction thresholds that have the same timestamp as the multiple first index values ​​from the prediction threshold sequence, and to perform fitting calculation on the multiple first index values ​​and the multiple first prediction thresholds to obtain a first tolerance coefficient.

[0015] The first indicator threshold determination module is configured to adjust the predicted threshold sequence based on the first tolerance coefficient to obtain a target predicted threshold sequence for real-time abnormal indicator values. The target predicted threshold sequence includes multiple indicator thresholds.

[0016] Thirdly, embodiments of this application also provide an indicator threshold determination device, the device comprising:

[0017] One or more processors;

[0018] Storage device, configured to store one or more programs,

[0019] When the one or more programs are executed by the one or more processors, the one or more processors implement the indicator threshold determination method described in the embodiments of this application.

[0020] Fourthly, embodiments of this application also provide a non-volatile storage medium for storing computer-executable instructions, which, when executed by a computer processor, are configured to perform the index threshold determination method described in embodiments of this application.

[0021] Fifthly, embodiments of this application also provide a computer program product, which includes a computer program stored in a computer-readable storage medium. At least one processor of the device reads from the computer-readable storage medium and executes the computer program, causing the device to perform the indicator threshold determination method described in embodiments of this application.

[0022] In this embodiment, a historical indicator value sequence is obtained, and a predicted indicator value sequence and a predicted threshold sequence are calculated based on a set sequence prediction algorithm. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values. A sensitivity value is obtained by weighting the feature values ​​of the multiple historical indicator values ​​and the multiple predicted indicator values ​​with different statistical characteristics. When the sensitivity value is within a preset threshold range, multiple first indicator values ​​that meet preset alarm rules are extracted from the historical indicator value sequence, and multiple first predicted thresholds with the same timestamps as the multiple first indicator values ​​are extracted from the predicted threshold sequence. A first tolerance coefficient is obtained by fitting the multiple first indicator values ​​and the multiple first predicted thresholds. The predicted threshold sequence is adjusted based on the first tolerance coefficient to obtain a target predicted threshold sequence for real-time abnormal indicator value warning. In the above scheme, by calculating the historical indicator value sequence according to the set sequence prediction algorithm, a preliminary predicted indicator value sequence and a predicted threshold sequence that conform to the changes in historical indicator data can be determined. A sensitivity value is obtained by weighting the feature values ​​of different statistical characteristics based on the historical indicator values ​​in the historical indicator value sequence and the predicted indicator values ​​in the predicted indicator value sequence. This effectively combines the correlation characteristics between the historical indicator value sequence and the predicted indicator value sequence, and reasonably determines the sensitivity value of the tolerance level for adjusting the predicted threshold sequence. By determining the threshold interval in which the sensitivity value lies, a corresponding tolerance coefficient can be calculated. Based on this first tolerance coefficient, the predicted threshold sequence can be adjusted to the target predicted threshold sequence. This allows for threshold adjustment to adapt to the trend changes of different indicator data, providing reliable indicator thresholds for indicator early warning and improving the accuracy of early warning. Attached Figure Description

[0023] Figure 1 is a flowchart of an index threshold determination method provided in an embodiment of this application;

[0024] Figure 2 is a flowchart of an index threshold determination method provided in an embodiment of this application, which includes the process of calculating the upper limit ratio coefficient and the upper limit offset coefficient;

[0025] Figure 3 is a flowchart of an index threshold determination method provided in an embodiment of this application, which includes the process of calculating the lower limit ratio coefficient and the lower limit offset coefficient;

[0026] Figure 4 is a flowchart of an index threshold determination method provided in an embodiment of this application, which includes a process of adjusting the prediction threshold sequence based on a second tolerance coefficient;

[0027] Figure 5 is a flowchart of an index threshold determination method provided in an embodiment of this application, which includes a process of adjusting the prediction threshold sequence based on a third tolerance coefficient;

[0028] Figure 6 is a flowchart of an index threshold determination method that includes a process of calculating sensitivity values, provided in an embodiment of this application.

[0029] Figure 7 is a flowchart of an index threshold determination method that includes a process of calculating index feature values, provided in an embodiment of this application.

[0030] Figure 8 is a structural block diagram of an index threshold determination device provided in an embodiment of this application;

[0031] Figure 9 is a schematic diagram of the structure of an index threshold determination device provided in an embodiment of this application. Detailed Implementation

[0032] The embodiments of this application will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely illustrative of the embodiments of this application and are not intended to limit the scope of the embodiments. Furthermore, it should be noted that, for ease of description, only the parts relevant to the embodiments of this application are shown in the accompanying drawings, not the entire structure.

[0033] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0034] The indicator threshold determination method provided in this application can obtain a sensitivity value by weighting the feature values ​​of historical indicator values ​​in the historical indicator value sequence and predicted indicator values ​​in the predicted indicator value sequence with different statistical characteristics. It then determines the threshold interval in which the sensitivity value falls, selects the corresponding first indicator value and the first predicted threshold for fitting calculation to obtain a first tolerance coefficient. This allows for effective adjustment of the predicted threshold sequence based on the first tolerance coefficient, providing a reliable indicator threshold for indicator early warning and improving the accuracy of early warning. Relevant indicators may include business indicators, performance indicators, and infrastructure indicators, etc., which are not limited in this application. The indicator threshold determination method, apparatus, device, storage medium, and program product provided in this application aim to solve the problem that the threshold prediction results of specific dynamic threshold algorithms in related technologies are relatively fixed and cannot adapt to the trend changes or fluctuations of different indicator data for threshold adjustment, resulting in low accuracy of indicator early warning based on this threshold and a tendency for false alarms and missed alarms.

[0035] The indicator threshold determination method provided in this application embodiment can be executed by a computer device. The computer device refers to any electronic device with data calculation, processing and storage capabilities, such as a server. This application embodiment does not limit this.

[0036] Figure 1 is a flowchart of a method for determining an index threshold provided in an embodiment of this application. As shown in Figure 1, it includes the following steps:

[0037] Step S101: Obtain the historical indicator value sequence, calculate the historical indicator value sequence according to the set sequence prediction algorithm, and obtain the predicted indicator value sequence and the predicted threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values.

[0038] The historical indicator value sequence can be a data sequence formed by arranging multiple historical indicator values ​​collected at multiple set time points in chronological order. The sequence prediction algorithm can be the Prophet algorithm, AutoARIMA (Autoregressive Integrated Moving Average), etc., which can predict multiple predicted indicator values ​​corresponding to each historical indicator value at future time points based on the changing patterns of historical indicator values ​​in the historical indicator value sequence. These multiple predicted indicator values ​​can be arranged in chronological order to form a predicted indicator value sequence. Furthermore, prediction thresholds corresponding to each predicted indicator value can be given based on set confidence intervals. These prediction thresholds can be arranged in chronological order to form a prediction threshold sequence, specifically including upper and lower prediction thresholds, constituting the uncertainty interval of the prediction. However, since the threshold prediction results of the sequence prediction algorithm are relatively fixed and do not involve threshold adjustment with different tolerance levels, it cannot adapt to the trend changes or fluctuations of different indicator data to provide relatively accurate prediction thresholds. Further tolerance-related threshold adjustment is required through subsequent steps.

[0039] Step S102: The sensitivity value is obtained by weighting the feature values ​​of multiple historical index values ​​and multiple predicted index values ​​with different statistical characteristics. The sensitivity value is used to adjust the tolerance of the predicted threshold sequence.

[0040] Because different indicator data exhibit diverse trends and fluctuations, indicator data with clear regularities shows smaller fluctuations, requiring less tolerance for adjusting the corresponding prediction threshold. This results in higher warning sensitivity, suitable for issuing anomaly alerts when indicator data fluctuates slightly. Conversely, indicator data with irregularities shows larger fluctuations, requiring greater tolerance for adjusting the corresponding prediction threshold. This results in lower warning sensitivity, suitable for issuing anomaly alerts when indicator data fluctuates significantly. Therefore, a sensitivity value measuring the tolerance for adjusting the prediction threshold sequence can be obtained by weighting the feature values ​​of different statistical characteristics using historical indicator value sequences and predicted indicator value sequences. These statistical characteristics can include the fluctuation characteristics of the historical indicator value sequence itself, the trend similarity characteristics between the historical and predicted indicator value sequences, and absolute similarity characteristics, used to measure the stability of the indicator data. In one embodiment, the statistical characteristics used can be fluctuation characteristics and trend similarity characteristics. The fluctuation characteristic value can be obtained by calculating the standard deviation of historical indicator values ​​in the historical indicator value sequence, as shown in the following formula:

[0041] Where Volatility is the volatility characteristic value, x i Here, represents the historical index value in the historical index value sequence, u is the average of the historical index values ​​in the historical index value sequence, and n is the number of elements in the historical index value sequence. The larger this volatility characteristic value, the greater the volatility of the index data, the lower the required sensitivity, and the greater the tolerance for adjusting the prediction threshold sequence.

[0042] The trend similarity feature value can be obtained by calculating the coefficient of determination between the historical indicator value series and the predicted indicator value series. This trend similarity feature value can measure the trend similarity between the historical indicator value series and the predicted indicator value series. The relevant calculation formula is as follows:

[0043] Where Similarity is the trend similarity feature value, y i These are historical indicator values ​​within a historical indicator value sequence. For the predicted index values ​​in the predicted index value sequence, is the average of the historical indicator values ​​in the historical indicator value sequence, and n is the number of elements in the historical indicator value sequence. The larger this trend similarity feature value is, the closer the historical indicator value sequence and the predicted indicator value sequence are, the smaller the fluctuation of the indicator data, the higher the required sensitivity, and the smaller the tolerance for adjusting the prediction threshold sequence.

[0044] Finally, the sensitivity value can be obtained by weighting the volatility characteristic value and the trend similarity characteristic value. The relevant calculation formula is as follows: Sensitivity=k1×(1-Volatility)+k2×Similarity,

[0045] Wherein, Sensitivity is the sensitivity value, Volatility is the volatility characteristic value, Similarity is the trend similarity characteristic value, and k1 and k2 are weights, for example, k1 = 0.5, k2 = 0.5, which are not limited in this application.

[0046] In one embodiment, the statistical features used can be volatility features and absolute similarity features. The volatility feature value can be obtained by calculating the standard deviation and mean of historical indicator values ​​in the historical indicator value sequence, and then dividing the standard deviation by the mean. The relevant calculation formula is as follows:

[0047] Where Volatility is the volatility characteristic value, SD is the standard deviation of the historical index values ​​in the historical index value sequence, and u is the average value of the historical index values ​​in the historical index value sequence. The larger the volatility characteristic value, the greater the volatility of the index data, the lower the required sensitivity, and the greater the tolerance for adjusting the prediction threshold sequence.

[0048] The absolute similarity feature value can be obtained by calculating the average absolute percentage error between the historical indicator value series and the predicted indicator value series. This absolute similarity feature value can measure the degree of closeness between the absolute distance between the historical indicator value series and the predicted indicator value series. The relevant calculation formula is as follows:

[0049] Where MAPE is the absolute similarity eigenvalue, y i These are historical indicator values ​​within a historical indicator value sequence. Let be the predicted index value in the predicted index value sequence, and n be the number of elements in the historical index value sequence. The larger this absolute similarity feature value is, the less similar the historical index value sequence and the predicted index value sequence are, the greater the fluctuation of the index data, the lower the required sensitivity, and the greater the tolerance for adjusting the prediction threshold sequence.

[0050] Finally, the sensitivity value can be obtained by weighting the volatility characteristic value and the trend similarity characteristic value. The relevant calculation formula is as follows: Sensitivity=k3×(1-Volatility)+k4×(1-MAPE),

[0051] Wherein, Sensitivity is the sensitivity value, Volatility is the fluctuation characteristic value, MAPE is the absolute similarity characteristic value, and k3 and k4 are weights, for example, k3 = 0.5 and k4 = 0.5, which are not limited in this application.

[0052] Of course, other different statistical features can be combined to perform eigenvalue weighted calculations, but this application does not limit this.

[0053] Step S103: When the sensitivity value is within the preset threshold range, extract multiple first index values ​​that meet the preset alarm rules from the historical index value sequence, and extract multiple first prediction thresholds with the same timestamp as the multiple first index values ​​from the prediction threshold sequence. Fit the multiple first index values ​​and the multiple first prediction thresholds to obtain the first tolerance coefficient.

[0054] Sensitivity values ​​can be categorized using pre-set threshold ranges to allow for different calculation methods of the tolerance coefficient, each corresponding to a different level of tolerance. If the sensitivity value falls within the pre-set threshold range, for example, 0.3 to 0.6, it can be considered relatively moderate. To adjust the tolerance of the prediction threshold sequence, a first indicator value satisfying a pre-set alarm rule can be extracted from the historical indicator value sequence. This pre-set alarm rule might be to consider every pre-set number of historical indicator values ​​exceeding the prediction threshold as the first indicator value, for example, every 4 or 5 historical indicator values. The number of first indicator values ​​will be fewer than the number of historical indicator values ​​exceeding the prediction threshold. Furthermore, the tolerance coefficient can be calculated by combining the predicted indicator values ​​in the prediction indicator value sequence with the same timestamp as the first indicator value. Adjusting the prediction threshold based on this tolerance coefficient ensures that the original first indicator values ​​satisfying the pre-set alarm rule are no longer considered alarm targets, guaranteeing appropriate alarm sensitivity suitable for indicator data with moderate fluctuations. The specific fitting calculation can employ linear or quadratic function relationships. For linear function fitting, the corresponding first tolerance coefficient may include a proportionality coefficient and an offset coefficient. For quadratic function fitting, the corresponding first tolerance coefficient may include a first proportionality coefficient, a second proportionality coefficient, and an offset coefficient. The specific function mapping relationship can be selected by the developers based on the fitting accuracy requirements of the actual application scenario, and this application does not impose any limitations on it. If the sensitivity value is greater than the upper limit of the preset threshold range, it can be considered that the sensitivity value is too large, and the tolerance of the predicted threshold sequence needs to be adjusted to be small. The target threshold sequence can be obtained by adjusting the predicted index values ​​in the predicted index value sequence according to the standard deviation of a preset multiple. The tolerance coefficient is then calculated by fitting the target threshold sequence and the predicted index value sequence. After adjusting the predicted threshold based on this tolerance coefficient, it can follow the statistical numerical range distribution, ensuring high alarm sensitivity, and is suitable for index data with small fluctuations. If the sensitivity value is less than the lower limit of the preset threshold range, it can be considered that the sensitivity value is small and the tolerance of the prediction threshold sequence needs to be adjusted. A second indicator value that exceeds the prediction threshold limit can be extracted from the historical indicator value sequence, and the tolerance coefficient can be obtained by fitting the prediction indicator value with the same timestamp as the second indicator value in the prediction indicator value sequence. After adjusting the prediction threshold based on the tolerance coefficient, the second indicator value no longer exceeds the prediction threshold limit, thereby reducing the alarm sensitivity. This is suitable for indicator data with large fluctuations.

[0055] Step S104: Adjust the predicted threshold sequence based on the first tolerance coefficient to obtain the target predicted threshold sequence, which is used for real-time abnormal indicator value warning.

[0056] In this process, the first tolerance coefficient corresponding to the preset fitting relationship can be obtained after the fitting calculation in step S103. In one embodiment, the fitting calculation can be based on a linear function relationship. The first tolerance coefficient includes a proportionality coefficient and an offset coefficient. The target predicted threshold sequence is obtained by adjusting the predicted threshold sequence based on the first tolerance coefficient, specifically as follows:

[0057] Based on the scaling factor, offset factor, and the established linear function relationship, each prediction threshold in the prediction threshold sequence is integrated to obtain the target prediction threshold sequence. The formula for adjusting the prediction threshold sequence is as follows: y1 = a1x + b1,

[0058] Where y1 is the index threshold, x is the prediction threshold in the prediction threshold sequence, a1 is the proportional coefficient, and b1 is the offset coefficient.

[0059] In one embodiment, fitting calculation can be performed based on a quadratic function relationship. Correspondingly, the first tolerance coefficient may include a first proportionality coefficient, a second proportionality coefficient, and a bias coefficient. The target predicted threshold sequence is obtained by adjusting the predicted threshold sequence based on the first tolerance coefficient, which can be specifically as follows:

[0060] Based on the first proportionality coefficient, the second proportionality coefficient, the offset coefficient, and the set quadratic function relationship, each predicted threshold in the predicted threshold sequence is integrated to obtain the target predicted threshold sequence. The relevant formula for adjusting the predicted threshold sequence is as follows: y² = a²x 2 +b2x+c1,

[0061] Where y2 is the index threshold, x is the prediction threshold in the prediction threshold sequence, a2 is the first proportional coefficient, b2 is the second proportional coefficient, and c1 is the offset coefficient.

[0062] Therefore, by adjusting the prediction threshold sequence, a target prediction threshold sequence can be obtained. For real-time indicator values ​​that exceed the threshold limits in the target prediction threshold sequence, anomaly warnings can be issued, triggering relevant background notifications to achieve the purpose of monitoring indicators.

[0063] The above describes a process where, by acquiring historical indicator value sequences and calculating them using a set sequence prediction algorithm, a predicted indicator value sequence and a predicted threshold sequence are obtained. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values. A sensitivity value is obtained by weighting the feature values ​​of the multiple historical and predicted indicator values ​​with different statistical characteristics. If the sensitivity value falls within a preset threshold range, multiple first indicator values ​​that satisfy preset alarm rules are extracted from the historical indicator value sequence, and multiple first predicted thresholds with the same timestamps as the multiple first indicator values ​​are extracted from the predicted threshold sequence. A first tolerance coefficient is obtained by fitting the multiple first indicator values ​​and the multiple first predicted thresholds. The predicted threshold sequence is then adjusted based on the first tolerance coefficient to obtain a target predicted threshold sequence, which is used for real-time abnormal indicator value warnings. In the above scheme, by calculating the historical indicator value sequence according to the set sequence prediction algorithm, the predicted indicator value sequence and the predicted threshold sequence that conform to the changes in historical indicator data can be initially determined. By weighting the feature values ​​of different statistical characteristics based on the historical indicator values ​​in the historical indicator value sequence and the predicted indicator values ​​in the predicted indicator value sequence, the sensitivity value can be obtained. This can effectively combine the correlation characteristics between the historical indicator value sequence and the predicted indicator value sequence, and reasonably determine the sensitivity value of the tolerance level of the specific adjustment of the predicted threshold sequence. By judging the threshold interval in which the sensitivity value is located, the corresponding tolerance coefficient can be calculated. Based on the first tolerance coefficient, the predicted threshold sequence can be adjusted to the target predicted threshold sequence. This can adapt to the trend changes of different indicator data to adjust the threshold, provide reliable indicator thresholds for indicator early warning, and improve the accuracy of early warning.

[0064] Figure 2 is a flowchart of an index threshold determination method provided in an embodiment of this application, which includes a process for calculating an upper limit ratio coefficient and an upper limit offset coefficient. As shown in Figure 2, it includes the following steps:

[0065] Step S201: Obtain the historical indicator value sequence, calculate the historical indicator value sequence according to the set sequence prediction algorithm, and obtain the predicted indicator value sequence and the predicted threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values.

[0066] Step S202: The sensitivity value is obtained by weighting the feature values ​​of multiple historical index values ​​and multiple predicted index values ​​with different statistical characteristics. The sensitivity value is used to adjust the tolerance of the predicted threshold sequence.

[0067] Step S203: When the sensitivity value is within a preset threshold range, extract multiple first indicator values ​​that meet the preset alarm rules from the historical indicator value sequence, and extract multiple first prediction thresholds with the same timestamp as the multiple first indicator values ​​from the prediction threshold sequence, wherein the multiple first prediction thresholds include multiple upper limit prediction thresholds.

[0068] The upper limit prediction threshold can represent the maximum predicted value obtained based on a sequence prediction algorithm. If a historical indicator value is greater than the upper limit prediction threshold corresponding to the same timestamp, it can be considered an indicator value that may trigger an anomaly warning in the future. Accordingly, the preset alarm rule can be to regard every preset number of consecutive historical indicator values ​​that are greater than the upper limit prediction threshold as the first indicator value.

[0069] Step S204: Calculate the standard deviation of multiple historical index values.

[0070] Step S205: Calculate the first ratio for each first indicator value and the upper limit prediction threshold with the same timestamp to obtain multiple first result values, and determine the maximum value among the multiple first result values ​​as the upper limit ratio coefficient.

[0071] In this embodiment, a linear function relationship is used for fitting calculation, requiring the corresponding calculation of the upper limit proportional coefficient and upper limit offset coefficient. The first ratio calculation can be achieved by dividing each first indicator value by the upper limit prediction threshold with the same timestamp to obtain multiple first result values, or by subtracting each first indicator value from the standard deviation and dividing the result by the upper limit prediction threshold with the same timestamp. This application does not limit the calculation. The first result value can be used to characterize the proportion of the first indicator value relative to the upper limit prediction threshold with the same timestamp. By determining the maximum value among the multiple first result values ​​as the upper limit proportional coefficient, the product of the calculated upper limit proportional coefficient and each upper limit prediction threshold can cover the corresponding first indicator value with the same timestamp. Optionally, a maximum proportional coefficient can be set. If the calculated upper limit proportional coefficient exceeds this maximum proportional coefficient, the upper limit proportional coefficient is adjusted to match the maximum proportional coefficient to prevent extreme values ​​in the first indicator values ​​from affecting the accuracy of the upper limit proportional coefficient.

[0072] Step S206: Based on the standard deviation and the upper limit ratio coefficient, perform a first integration calculation on the upper limit prediction threshold with the same timestamp for each first indicator value to obtain multiple second result values, and determine the maximum value among the multiple second result values ​​as the upper limit offset coefficient.

[0073] The first integrated calculation can be based on the inverse calculation of a linear function relationship to obtain multiple second result values. Specifically, the relevant formula for the first integrated calculation is as follows: b′=y+k′Δb-a′*x,

[0074] Wherein, b′ is the second result value, y is the first indicator value, Δb is the standard deviation, a′ is the upper limit proportional coefficient, x is the upper limit prediction threshold with the same timestamp as the first indicator value, and k′ is a preset multiple, for example, k′=1, which is not limited in this application. The standard deviation Δb can be used to compensate for data fluctuations, making the calculated second result value more stable. The second result value can be used to characterize the offset value of the first indicator value relative to the upper limit prediction threshold with the same timestamp. By determining the maximum value among multiple second result values ​​as the upper limit offset coefficient, the product of the upper limit proportional coefficient and each upper limit prediction threshold, plus the upper limit offset coefficient, can cover the corresponding first indicator value with the same timestamp. Optionally, a maximum offset coefficient can be set. If the calculated upper limit offset coefficient exceeds the maximum offset coefficient, the upper limit offset coefficient is adjusted to be consistent with the maximum offset coefficient to prevent extreme values ​​in the first indicator value from affecting the accuracy of the upper limit offset coefficient.

[0075] Step S207: Adjust the predicted threshold sequence based on the upper limit ratio coefficient and the upper limit offset coefficient to obtain the target predicted threshold sequence, which is used for real-time abnormal indicator value warning.

[0076] The above-mentioned method of determining the upper limit ratio coefficient and upper limit offset coefficient through the first ratio calculation and the first integration calculation can effectively match multiple first indicator values ​​that meet the preset alarm rules, determine the tolerance adjustment range of the preset threshold sequence, and make the indicator threshold more closely match the actual fluctuation of the indicator data.

[0077] Figure 3 is a flowchart of an index threshold determination method provided in an embodiment of this application, which includes the process of calculating the lower limit proportional coefficient and the lower limit offset coefficient. As shown in Figure 3, it includes the following steps:

[0078] Step S301: Obtain the historical indicator value sequence, calculate the historical indicator value sequence according to the set sequence prediction algorithm, and obtain the predicted indicator value sequence and the predicted threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values.

[0079] Step S302: The sensitivity value is obtained by weighting the feature values ​​of multiple historical index values ​​and multiple predicted index values ​​with different statistical characteristics. The sensitivity value is used to adjust the tolerance of the predicted threshold sequence.

[0080] Step S303: When the sensitivity value is within a preset threshold range, extract multiple first indicator values ​​that meet the preset alarm rules from the historical indicator value sequence, and extract multiple first prediction thresholds with the same timestamp as the multiple first indicator values ​​from the prediction threshold sequence, wherein the multiple first prediction thresholds include multiple lower limit prediction thresholds.

[0081] The lower limit prediction threshold can represent the minimum predicted value obtained based on a sequence prediction algorithm. If a historical indicator value is less than the lower limit prediction threshold corresponding to the same timestamp, it can be considered an indicator value that may trigger an anomaly warning in the future. Accordingly, the preset alarm rule can be to regard every preset number of consecutive historical indicator values ​​that are less than the lower limit prediction threshold as the first indicator value.

[0082] Step S304: Calculate the standard deviation of multiple historical index values.

[0083] Step S305: Based on the standard deviation, calculate the second ratio for each first indicator value and the lower limit prediction threshold with the same timestamp to obtain multiple third result values. Determine the minimum value among the multiple third result values ​​as the lower limit ratio coefficient.

[0084] In this embodiment, a linear function relationship is used for fitting calculation, requiring the corresponding calculation of the lower limit proportional coefficient and the lower limit offset coefficient. The second ratio calculation can be achieved by subtracting each first indicator value from its standard deviation, dividing the result by the lower limit prediction threshold with the same timestamp, to obtain multiple third result values. The standard deviation can be used to compensate for data fluctuations, making the calculated third result values ​​more stable. Alternatively, it can be achieved by subtracting each first indicator value from its standard deviation by a preset multiple, dividing the result by the lower limit prediction threshold with the same timestamp, to obtain multiple third result values. This application does not limit the specific third result value. The third result value can be used to characterize the proportion of the first indicator value relative to the lower limit prediction threshold with the same timestamp. By determining the minimum value among the multiple third result values ​​as the lower limit proportional coefficient, the product of the calculated lower limit proportional coefficient and each lower limit prediction threshold can cover the corresponding first indicator value with the same timestamp. Optionally, a minimum proportional coefficient can be set. If the calculated lower limit proportional coefficient is less than this minimum proportional coefficient, the lower limit proportional coefficient is adjusted to be consistent with the minimum proportional coefficient to prevent extreme values ​​in the first indicator values ​​from affecting the accuracy of the lower limit proportional coefficient.

[0085] Step S306: Based on the lower limit ratio coefficient, perform a second integration calculation on the lower limit prediction threshold of each first indicator value and the same timestamp to obtain multiple fourth result values, and determine the minimum value among the multiple fourth result values ​​as the lower limit offset coefficient.

[0086] The second integrated calculation can be based on the inverse calculation of a linear function relationship to obtain multiple fourth result values. Specifically, the relevant formula for the second integrated calculation is as follows: b″=ya″*x+c,

[0087] Wherein, b″ is the second result value, y is the first index value, a″ is the lower limit proportional coefficient, x is the lower limit prediction threshold with the same timestamp as the first index value, and c is a compensation term, for example, c = 0, which is not limited in this application. This fourth result value can be used to characterize the offset of the first index value relative to the lower limit prediction threshold with the same timestamp. By determining the minimum value among multiple fourth result values ​​as the lower limit offset coefficient, the product of the lower limit proportional coefficient and each lower limit prediction threshold, plus the lower limit offset coefficient, can cover the corresponding first index value with the same timestamp. Optionally, a minimum offset coefficient can be set. If the calculated lower limit offset coefficient is less than this minimum offset coefficient, then the lower limit offset coefficient is adjusted to be consistent with the maximum minimum offset coefficient to prevent extreme values ​​in the first index value from affecting the accuracy of the lower limit offset coefficient.

[0088] Step S307: Adjust the predicted threshold sequence based on the lower limit proportional coefficient and the lower limit offset coefficient to obtain the target predicted threshold sequence, which is used for real-time abnormal indicator value warning.

[0089] The above-mentioned calculation of the lower limit ratio coefficient and the lower limit offset coefficient by the second ratio calculation and the second integration calculation can effectively match multiple first indicator values ​​that meet the preset alarm rules, determine the tolerance adjustment range of the preset threshold sequence, and make the indicator threshold more in line with the actual fluctuation of the indicator data.

[0090] Figure 4 is a flowchart of an index threshold determination method provided in an embodiment of this application, which includes a process of adjusting the prediction threshold sequence based on a second tolerance coefficient. As shown in Figure 4, it includes the following steps:

[0091] Step S401: Obtain the historical indicator value sequence, calculate the historical indicator value sequence according to the set sequence prediction algorithm, and obtain the predicted indicator value sequence and the predicted threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values.

[0092] Step S402: The sensitivity value is obtained by weighting the feature values ​​of multiple historical index values ​​and multiple predicted index values ​​with different statistical characteristics. The sensitivity value is used to adjust the tolerance of the predicted threshold sequence.

[0093] Step S403: When the sensitivity value is within the preset threshold range, extract multiple first index values ​​that meet the preset alarm rules from the historical index value sequence, and extract multiple first prediction thresholds with the same timestamp as the multiple first index values ​​from the prediction threshold sequence. Fit the multiple first index values ​​and the multiple first prediction thresholds to obtain the first tolerance coefficient.

[0094] Step S404: Adjust the prediction threshold sequence based on the first tolerance coefficient to obtain the target prediction threshold sequence, which is used for real-time abnormal indicator value warning. The target prediction threshold sequence includes multiple indicator thresholds.

[0095] Step S405: When the sensitivity value is greater than the upper limit of the preset threshold range, calculate the standard deviation of multiple historical index values, and adjust the predicted index values ​​in the predicted index value sequence according to the standard deviation of the preset multiple to obtain the target threshold sequence.

[0096] If the sensitivity value is greater than the upper limit of the preset threshold range, it can be considered that the sensitivity value is too large, and the tolerance of the predicted threshold sequence is small. Therefore, a suitable preset multiple of the standard deviation can be selected based on the statistical distribution of numerical intervals, and the target threshold can be calculated using the predicted index value as a benchmark. For example, since the probability of the numerical distribution within the interval (μ-3σ, μ+3σ) is approximately 0.9974, where μ is the mean and σ is the standard deviation, approximately 99.74% of the data values ​​will fall within three standard deviations of the mean. This means that under the assumption of normal distribution, the probability of data appearing outside three standard deviations of the mean is very small and can be considered outliers. Therefore, the preset multiple can be set to 3, and the target threshold sequence can be set to include a target upper threshold sequence and a target lower threshold sequence. The corresponding calculation formula is as follows:

[0097] in, The target upper limit threshold in the target upper limit threshold sequence. y is the target lower limit threshold in the target lower limit threshold sequence. predict Let σ be the predicted index value in the predicted index value sequence, and σ be the standard deviation of the historical index values ​​in the historical index value sequence.

[0098] Step S406: Fit the target threshold sequence and the predicted index value sequence to obtain the second tolerance coefficient.

[0099] The second tolerance coefficient may include an upper tolerance coefficient and a lower tolerance coefficient. The upper tolerance coefficient may include an upper proportional coefficient and an upper offset coefficient, and the lower tolerance coefficient may include a lower proportional coefficient and a lower offset coefficient. This fitting calculation can employ methods such as least squares or gradient descent, which are not limited in this application. Taking the least squares method as an example, the relevant calculation formula is as follows:

[0100] Among them, a up b is the upper limit proportionality coefficient. up This is the upper limit offset coefficient. The target upper limit threshold in the target upper limit threshold sequence. For the predicted index values ​​in the predicted index value sequence, This represents the average of the target upper limit thresholds in the target upper limit threshold sequence. Let be the average value of the predicted index values ​​in the predicted index value sequence, and n be the number of elements in the predicted index value sequence. Similarly, the lower limit proportional coefficient and the lower limit offset coefficient can be calculated, which will not be elaborated here.

[0101] Step S407: Adjust the prediction threshold sequence based on the second tolerance coefficient to obtain the target prediction threshold sequence.

[0102] As mentioned above, when the sensitivity value is greater than the upper limit of the preset threshold range, the statistical numerical range distribution can be followed, and a tolerance coefficient calculation method with a smaller tolerance level can be selected. This is beneficial to ensure higher alarm sensitivity and is suitable for application scenarios with small data fluctuations.

[0103] Figure 5 is a flowchart of an index threshold determination method provided in an embodiment of this application, which includes a process of adjusting the prediction threshold sequence based on a third tolerance coefficient. As shown in Figure 5, it includes the following steps:

[0104] Step S501: Obtain the historical indicator value sequence, calculate the historical indicator value sequence according to the set sequence prediction algorithm, and obtain the predicted indicator value sequence and the predicted threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values.

[0105] Step S502: The sensitivity value is obtained by weighting the feature values ​​of multiple historical index values ​​and multiple predicted index values ​​with different statistical characteristics. The sensitivity value is used to adjust the tolerance of the predicted threshold sequence.

[0106] Step S503: When the sensitivity value is within the preset threshold range, extract multiple first index values ​​that meet the preset alarm rules from the historical index value sequence, and extract multiple first prediction thresholds with the same timestamp as the multiple first index values ​​from the prediction threshold sequence. Fit the multiple first index values ​​and the multiple first prediction thresholds to obtain the first tolerance coefficient.

[0107] Step S504: Adjust the predicted threshold sequence based on the first tolerance coefficient to obtain the target predicted threshold sequence, which is used for real-time abnormal indicator value warning.

[0108] Step S505: When the sensitivity value is less than the lower limit of the preset threshold range, extract multiple second index values ​​that meet the preset anomaly rules from the historical index value sequence, and extract multiple second prediction thresholds with the same timestamp as the multiple second index values ​​from the prediction threshold sequence. Fit the multiple second index values ​​and the multiple second prediction thresholds to obtain the third tolerance coefficient.

[0109] If the sensitivity value is less than the lower limit of the preset threshold range, it can be considered that the sensitivity value is small and the tolerance of the prediction threshold sequence needs to be adjusted. Multiple second index values ​​that satisfy a preset anomaly rule can be extracted from the historical index value sequence. This preset anomaly rule can be to regard historical index values ​​that exceed the preset threshold limit as second index values. The specific fitting calculation process can be referred to the relevant description of the first tolerance coefficient in the aforementioned embodiments, and will not be repeated here.

[0110] Step S506: Adjust the predicted threshold sequence based on the third tolerance coefficient to obtain the target predicted threshold sequence.

[0111] As described above, by adjusting the prediction threshold sequence based on the third tolerance coefficient, the original second indicator value that meets the preset anomaly rules can no longer be regarded as an anomaly, ensuring a smaller alarm sensitivity, which is suitable for application scenarios with large data fluctuations.

[0112] Figure 6 is a flowchart of an index threshold determination method including a process for calculating a sensitivity value, provided in an embodiment of this application. As shown in Figure 6, it includes the following steps:

[0113] Step S601: Obtain the historical indicator value sequence, calculate the historical indicator value sequence according to the set sequence prediction algorithm, and obtain the predicted indicator value sequence and the predicted threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values.

[0114] Step S602: Calculate multiple indicator feature values ​​corresponding to different statistical characteristics based on multiple historical indicator values ​​and multiple predicted indicator values. The indicator feature values ​​are used to represent the degree of fluctuation of historical indicator values ​​or the degree of similarity between the historical indicator value sequence and the predicted indicator value sequence.

[0115] The selected statistical features can be used to represent the magnitude of fluctuation in historical indicator values, or to represent the degree of similarity between historical indicator value sequences and predicted indicator value sequences; this application does not impose any limitations on this. Each statistical feature can be used to calculate an indicator feature value. For example, for fluctuation features, the standard deviation or coefficient of variation can be calculated as the fluctuation feature value; for similarity features, the coefficient of determination or mean absolute percentage error can be calculated as the trend similarity feature value. Of course, there are other methods for calculating features that measure the trend changes and fluctuations of indicator data, which are not limited in this application.

[0116] Step S603: Calculate the sensitivity value by weighting the feature values ​​of multiple indicators.

[0117] Step S604: When the sensitivity value is within the preset threshold range, extract multiple first index values ​​that meet the preset alarm rules from the historical index value sequence, and extract multiple first prediction thresholds with the same timestamp as the multiple first index values ​​from the prediction threshold sequence. Fit the multiple first index values ​​and the multiple first prediction thresholds to obtain the first tolerance coefficient.

[0118] Step S605: Adjust the predicted threshold sequence based on the first tolerance coefficient to obtain the target predicted threshold sequence, which is used for real-time abnormal indicator value warning.

[0119] As mentioned above, by calculating multiple indicator feature values ​​corresponding to different statistical characteristics, the trend changes and fluctuations of indicator data can be effectively measured. Furthermore, by weighting multiple indicator feature values ​​to obtain a sensitivity value, the appropriate sensitivity value can be determined by reasonably integrating feature information from different dimensions, adapting to the actual trend changes and fluctuations of the indicators, and deciding the tolerance level of the prediction threshold sequence.

[0120] Figure 7 is a flowchart of an index threshold determination method provided in an embodiment of this application, which includes a process for calculating index feature values. As shown in Figure 7, it includes the following steps:

[0121] Step S701: Obtain the historical indicator value sequence, calculate the historical indicator value sequence according to the set sequence prediction algorithm, and obtain the predicted indicator value sequence and the predicted threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values.

[0122] Step S702: Calculate the standard deviation and mean of multiple historical indicator values, divide the standard deviation by the mean to obtain the fluctuation characteristic value, perform linear correlation calculation on multiple historical indicator values ​​and multiple predicted indicator values ​​to obtain the trend similarity characteristic value, and perform relative error calculation on multiple historical indicator values ​​and multiple predicted indicator values ​​to obtain the absolute similarity characteristic value.

[0123] The linear correlation calculation can be performed by calculating the coefficient of determination, Pearson correlation coefficient, etc., as trend similarity feature values, and the relative error calculation can be performed by calculating the mean absolute percentage error, mean square error, etc., as absolute similarity feature values. This application does not limit the calculation.

[0124] Step S703: The sensitivity value is obtained by weighting the fluctuation feature value, trend similarity feature value and absolute similarity feature value.

[0125] The specific weighted calculation formula is as follows: Sensitivity=k′1×(1-Volatility)+k2′×Similarity+k3′×(1-MAPE),

[0126] Wherein, Sensitivity is the sensitivity value, Volatility is the fluctuation characteristic value, Similarity is the trend similarity characteristic value, MAPE is the absolute similarity characteristic value, and k′1, k′2 and k′3 are weights, for example, k′1 = 1 / 3, k′2 = 1 / 3, k′3 = 1 / 3, which are not limited in this application.

[0127] Step S704: When the sensitivity value is within the preset threshold range, extract multiple first index values ​​that meet the preset alarm rules from the historical index value sequence, and extract multiple first prediction thresholds with the same timestamp as the multiple first index values ​​from the prediction threshold sequence. Fit the multiple first index values ​​and the multiple first prediction thresholds to obtain the first tolerance coefficient.

[0128] Step S705: Adjust the predicted threshold sequence based on the first tolerance coefficient to obtain the target predicted threshold sequence, which is used for real-time abnormal indicator value warning.

[0129] The sensitivity value is obtained by weighting the fluctuation feature value, trend similarity feature value, and absolute similarity feature value. This allows for the effective selection of representative statistical features for sensitivity value calculation, ensuring the reliability of sensitivity measurement in determining the tolerance level of the prediction threshold sequence and improving the accuracy of early warning.

[0130] Figure 8 is a structural block diagram of an indicator threshold determination device provided in an embodiment of this application. This device is configured to execute the indicator threshold determination method provided in the above embodiment, and has corresponding functional modules and beneficial effects for executing the method. As shown in Figure 8, the device includes:

[0131] The prediction sequence determination module 101 is configured to obtain a historical indicator value sequence, calculate the historical indicator value sequence according to the set sequence prediction algorithm, and obtain a prediction indicator value sequence and a prediction threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the prediction indicator value sequence includes multiple prediction indicator values.

[0132] The sensitivity value determination module 102 is configured to perform a weighted calculation of the feature values ​​of multiple historical index values ​​and multiple predicted index values ​​with different statistical characteristics to obtain a sensitivity value. The sensitivity value is used to adjust the tolerance level of the predicted threshold sequence.

[0133] The tolerance coefficient determination module 103 is configured to extract multiple first index values ​​that meet the preset alarm rules from the historical index value sequence when the sensitivity value is within the preset threshold range, and to extract multiple first prediction thresholds that have the same timestamp as the multiple first index values ​​from the prediction threshold sequence, and to perform fitting calculation on the multiple first index values ​​and the multiple first prediction thresholds to obtain the first tolerance coefficient.

[0134] The first indicator threshold determination module 104 is configured to adjust the predicted threshold sequence based on the first tolerance coefficient to obtain the target predicted threshold sequence for use in real-time abnormal indicator value warning. The target predicted threshold sequence includes multiple indicator thresholds.

[0135] The above describes a process where, by acquiring historical indicator value sequences and calculating them using a set sequence prediction algorithm, a predicted indicator value sequence and a predicted threshold sequence are obtained. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values. A sensitivity value is obtained by weighting the feature values ​​of the multiple historical and predicted indicator values ​​with different statistical characteristics. If the sensitivity value falls within a preset threshold range, multiple first indicator values ​​that satisfy preset alarm rules are extracted from the historical indicator value sequence, and multiple first predicted thresholds with the same timestamps as the multiple first indicator values ​​are extracted from the predicted threshold sequence. A first tolerance coefficient is obtained by fitting the multiple first indicator values ​​and the multiple first predicted thresholds. The predicted threshold sequence is then adjusted based on the first tolerance coefficient to obtain a target predicted threshold sequence, which is used for real-time abnormal indicator value warnings. In the above scheme, by calculating the historical indicator value sequence according to the set sequence prediction algorithm, the predicted indicator value sequence and the predicted threshold sequence that conform to the changes in historical indicator data can be initially determined. By weighting the feature values ​​of different statistical characteristics based on the historical indicator values ​​in the historical indicator value sequence and the predicted indicator values ​​in the predicted indicator value sequence, the sensitivity value can be obtained. This can effectively combine the correlation characteristics between the historical indicator value sequence and the predicted indicator value sequence, and reasonably determine the sensitivity value of the tolerance level of the specific adjustment of the predicted threshold sequence. By judging the threshold interval in which the sensitivity value is located, the corresponding tolerance coefficient can be calculated. Based on the first tolerance coefficient, the predicted threshold sequence can be adjusted to the target predicted threshold sequence. This can adapt to the trend changes of different indicator data to adjust the threshold, provide reliable indicator thresholds for indicator early warning, and improve the accuracy of early warning.

[0136] In one possible embodiment, the plurality of first prediction thresholds include a plurality of upper limit prediction thresholds, the first tolerance coefficient includes an upper limit tolerance coefficient, the upper limit tolerance coefficient includes an upper limit proportional coefficient and an upper limit offset coefficient, and the tolerance coefficient determination module 103 is further configured to:

[0137] Calculate the standard deviation of multiple historical indicator values;

[0138] For each first indicator value and the same upper limit prediction threshold with the same timestamp, the first ratio is calculated to obtain multiple first result values. The maximum value among the multiple first result values ​​is determined as the upper limit ratio coefficient.

[0139] Based on the standard deviation and upper limit proportional coefficient, the upper limit prediction threshold with the same time stamp for each first indicator value is integrated and calculated to obtain multiple second result values. The maximum value among the multiple second result values ​​is determined as the upper limit offset coefficient.

[0140] In one possible embodiment, the plurality of first prediction thresholds include a plurality of lower limit prediction thresholds, the first tolerance coefficient includes a lower limit tolerance coefficient, the lower limit tolerance coefficient includes a lower limit proportional coefficient and a lower limit offset coefficient, and the tolerance coefficient determination module 103 is further configured to:

[0141] Calculate the standard deviation of multiple historical indicator values;

[0142] Based on the standard deviation, a second ratio is calculated for each first indicator value and the same timestamp as the lower limit prediction threshold, resulting in multiple third result values. The minimum value among the multiple third result values ​​is determined as the lower limit ratio coefficient.

[0143] Based on the lower limit ratio coefficient, a second integration calculation is performed on the lower limit prediction threshold with the same timestamp for each first indicator value to obtain multiple fourth result values. The minimum value among the multiple fourth result values ​​is determined as the lower limit offset coefficient.

[0144] In one possible embodiment, a second indicator threshold determination module is also included, configured as follows:

[0145] When the sensitivity value is greater than the upper limit of the preset threshold range, the standard deviation of multiple historical index values ​​is calculated, and the predicted index values ​​in the predicted index value sequence are adjusted according to the standard deviation of the preset multiple to obtain the target threshold sequence.

[0146] The second tolerance coefficient is obtained by fitting the target threshold sequence and the predicted index value sequence;

[0147] The target prediction threshold sequence is obtained by adjusting the prediction threshold sequence based on the second tolerance coefficient.

[0148] In one possible embodiment, a third indicator threshold determination module is also included, configured as follows:

[0149] When the sensitivity value is less than the lower limit of the preset threshold range, multiple second index values ​​that meet the preset anomaly rules are extracted from the historical index value sequence, and multiple second prediction thresholds with the same timestamp as the multiple second index values ​​are extracted from the prediction threshold sequence. The multiple second index values ​​and multiple second prediction thresholds are fitted and calculated to obtain the third tolerance coefficient.

[0150] The target prediction threshold sequence is obtained by adjusting the prediction threshold sequence based on the third tolerance coefficient.

[0151] In one possible embodiment, the sensitivity value determination module 102 is further configured to:

[0152] Based on multiple historical indicator values ​​and multiple predicted indicator values, multiple indicator feature values ​​corresponding to different statistical characteristics are calculated. The indicator feature values ​​are used to represent the degree of fluctuation of historical indicator values ​​or the degree of similarity between the historical indicator value sequence and the predicted indicator value sequence.

[0153] The sensitivity value is obtained by weighting the feature values ​​of multiple indicators.

[0154] In one possible embodiment, the sensitivity value determination module 102 is further configured to:

[0155] Calculate the standard deviation and mean of multiple historical indicator values, and divide the standard deviation by the mean to obtain the volatility characteristic value;

[0156] Trend similarity feature values ​​are obtained by performing linear correlation calculations on multiple historical indicator values ​​and multiple predicted indicator values.

[0157] The absolute similarity feature value is obtained by calculating the relative error of multiple historical index values ​​and multiple predicted index values.

[0158] The sensitivity value is obtained by weighting the fluctuation characteristic value, the trend similarity characteristic value, and the absolute similarity characteristic value.

[0159] In one possible embodiment, the first tolerance coefficient includes a scaling factor and an offset factor, and the first index threshold determination module 104 is further configured to:

[0160] The target prediction threshold sequence is obtained by integrating each prediction threshold in the prediction threshold sequence based on the scaling factor, the offset factor, and the set linear function relationship.

[0161] Figure 9 is a schematic diagram of the structure of an index threshold determination device provided in an embodiment of this application. As shown in Figure 9, the device includes a processor 201, a memory 202, an input device 203, and an output device 204. The number of processors 201 in the device can be one or more; Figure 9 shows an example of one processor 201. The processor 201, memory 202, input device 203, and output device 204 in the device can be connected via a bus or other means; Figure 9 shows an example of connection via a bus. The memory 202, as a computer-readable storage medium, can be configured to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the index threshold determination method in the embodiment of this application. The processor 201 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory 202, thereby implementing the aforementioned index threshold determination method. The input device 203 can be configured to receive input digital or character information and generate key signal inputs related to user settings and function control of the device. The output device 204 may include a display screen or other display device.

[0162] This application embodiment also provides a non-volatile storage medium containing computer-executable instructions, which, when executed by a computer processor, are configured to perform an indicator threshold determination method described in the above embodiments. The method includes: acquiring a historical indicator value sequence; calculating the historical indicator value sequence according to a set sequence prediction algorithm to obtain a predicted indicator value sequence and a predicted threshold sequence, wherein the historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values; performing feature value weighting calculation on the multiple historical indicator values ​​and the multiple predicted indicator values ​​using different statistical characteristics to obtain a sensitivity value, wherein the sensitivity value is used to adjust the tolerance level of the predicted threshold sequence; when the sensitivity value is within a preset threshold range, extracting multiple first indicator values ​​that satisfy preset alarm rules from the historical indicator value sequence, and extracting multiple first predicted thresholds with the same timestamps as the multiple first indicator values ​​from the predicted threshold sequence; performing fitting calculation on the multiple first indicator values ​​and the multiple first predicted thresholds to obtain a first tolerance coefficient; and adjusting the predicted threshold sequence based on the first tolerance coefficient to obtain a target predicted threshold sequence for real-time abnormal indicator value warning, wherein the target predicted threshold sequence includes multiple indicator thresholds.

[0163] It is worth noting that in the embodiments of the above-mentioned indicator threshold determination device, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not configured to limit the protection scope of the embodiments of this application.

[0164] In some possible implementations, various aspects of the methods provided in this application can also be implemented as a program product comprising program code that, when run on a computer device, is configured to cause the computer device to perform the steps of the methods according to the various exemplary embodiments of this application described above. For example, the computer device may execute the index threshold determination method described in the embodiments of this application. The program product may be implemented using any combination of one or more readable media.

Claims

A method for determining a threshold value of an index, characterized in that include: Obtain a historical indicator value sequence, and calculate the historical indicator value sequence according to the set sequence prediction algorithm to obtain a predicted indicator value sequence and a predicted threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the predicted indicator value sequence includes multiple predicted indicator values. A sensitivity value is obtained by weighting the feature values ​​of the multiple historical index values ​​and the multiple predicted index values ​​with different statistical characteristics. The sensitivity value is used to adjust the tolerance level of the predicted threshold sequence. When the sensitivity value is within a preset threshold range, multiple first index values ​​that satisfy preset alarm rules are extracted from the historical index value sequence, and multiple first prediction thresholds with the same timestamp as the multiple first index values ​​are extracted from the prediction threshold sequence. The multiple first index values ​​and the multiple first prediction thresholds are fitted and calculated to obtain a first tolerance coefficient. The target prediction threshold sequence is obtained by adjusting the prediction threshold sequence based on the first tolerance coefficient, and is used for real-time abnormal indicator value warning. The target prediction threshold sequence includes multiple indicator thresholds. The indicator threshold value determination method according to claim 1, characterized in that The plurality of first prediction thresholds includes a plurality of upper limit prediction thresholds, the first tolerance coefficient includes an upper limit tolerance coefficient, the upper limit tolerance coefficient includes an upper limit proportional coefficient and an upper limit offset coefficient, and the step of fitting and calculating the first tolerance coefficient by the plurality of first index values ​​and the plurality of first prediction thresholds includes: Calculate the standard deviation of the aforementioned historical index values; For each upper limit prediction threshold that has the same first indicator value and timestamp, a first ratio is calculated to obtain multiple first result values, and the maximum value among the multiple first result values ​​is determined as the upper limit ratio coefficient; Based on the standard deviation and the upper limit ratio coefficient, a first integration calculation is performed on the upper limit prediction threshold for each first indicator value and the same timestamp to obtain multiple second result values. The maximum value among the multiple second result values ​​is determined as the upper limit offset coefficient. The indicator threshold value determination method according to claim 1, characterized in that The plurality of first prediction thresholds include a plurality of lower limit prediction thresholds, the first tolerance coefficient includes a lower limit tolerance coefficient, the lower limit tolerance coefficient includes a lower limit proportional coefficient and a lower limit offset coefficient, and the step of fitting and calculating the first tolerance coefficient by the plurality of first index values ​​and the plurality of first prediction thresholds includes: Calculate the standard deviation of the aforementioned historical index values; Based on the standard deviation, a second ratio is calculated for each of the first indicator values ​​and the same timestamp as the lower limit prediction threshold, resulting in multiple third result values. The minimum value among the multiple third result values ​​is determined as the lower limit ratio coefficient. Based on the lower limit ratio coefficient, a second integration calculation is performed on each of the first indicator values ​​and the lower limit prediction threshold with the same timestamp to obtain multiple fourth result values. The minimum value among the multiple fourth result values ​​is determined as the lower limit offset coefficient. The indicator threshold value determination method according to claim 1, characterized in that After obtaining the sensitivity value by weighting the feature values ​​of the multiple historical indicator values ​​and the multiple predicted indicator values ​​according to different statistical characteristics, the method further includes: When the sensitivity value is greater than the upper limit of the preset threshold range, the standard deviation of the multiple historical index values ​​is calculated, and the predicted index values ​​in the predicted index value sequence are adjusted according to the standard deviation of the preset multiple to obtain the target threshold sequence. The second tolerance coefficient is obtained by fitting the target threshold sequence and the predicted index value sequence; The target prediction threshold sequence is obtained by adjusting the prediction threshold sequence based on the second tolerance coefficient. The indicator threshold value determination method according to claim 1, characterized in that After obtaining the sensitivity value by weighting the feature values ​​of the multiple historical indicator values ​​and the multiple predicted indicator values ​​according to different statistical characteristics, the method further includes: When the sensitivity value is less than the lower limit of the preset threshold range, multiple second index values ​​that satisfy the preset anomaly rules are extracted from the historical index value sequence, and multiple second prediction thresholds with the same timestamp as the multiple second index values ​​are extracted from the prediction threshold sequence. The multiple second index values ​​and the multiple second prediction thresholds are fitted and calculated to obtain a third tolerance coefficient. The target prediction threshold sequence is obtained by adjusting the prediction threshold sequence based on the third tolerance coefficient. The indicator threshold value determination method according to claim 1, characterized in that The sensitivity value is obtained by weighting the feature values ​​of the multiple historical indicator values ​​and the multiple predicted indicator values ​​according to different statistical characteristics, including: Based on the multiple historical indicator values ​​and the multiple predicted indicator values, multiple indicator feature values ​​corresponding to different statistical characteristics are calculated. The indicator feature values ​​are used to represent the degree of fluctuation of the historical indicator values, or the degree of similarity between the historical indicator value sequence and the predicted indicator value sequence. The sensitivity value is obtained by weighting the feature values ​​of the multiple indicators. The indicator threshold value determination method according to claim 6, characterized in that The multiple indicator feature values ​​include fluctuation feature values, trend similarity feature values, and absolute similarity feature values. The calculation of multiple indicator feature values ​​corresponding to different statistical characteristics based on the multiple historical indicator values ​​and the multiple predicted indicator values ​​includes: Calculate the standard deviation and average of the multiple historical indicator values, and divide the standard deviation by the average to obtain the fluctuation characteristic value; A trend similarity feature value is obtained by performing linear correlation calculation on the multiple historical indicator values ​​and the multiple predicted indicator values; The absolute similarity feature value is obtained by calculating the relative error between the multiple historical index values ​​and the multiple predicted index values. Accordingly, the step of weighting the multiple indicator feature values ​​to obtain the sensitivity value includes: The sensitivity value is obtained by weighting the fluctuation feature value, the trend similarity feature value, and the absolute similarity feature value. The indicator threshold value determination method according to claim 1, characterized in that The first tolerance coefficient includes a scaling factor and an offset factor. The step of adjusting the predicted threshold sequence based on the first tolerance coefficient to obtain the target predicted threshold sequence includes: Based on the proportional coefficient, the offset coefficient, and the set linear function relationship, each prediction threshold in the prediction threshold sequence is integrated to obtain the target prediction threshold sequence. An index threshold value determination apparatus characterized by comprising: include: The prediction sequence determination module is configured to acquire a historical indicator value sequence, calculate the historical indicator value sequence according to a set sequence prediction algorithm, and obtain a prediction indicator value sequence and a prediction threshold sequence. The historical indicator value sequence includes multiple historical indicator values, and the prediction indicator value sequence includes multiple prediction indicator values. The sensitivity value determination module is configured to perform a weighted calculation of the feature values ​​of the plurality of historical index values ​​and the plurality of predicted index values ​​with different statistical characteristics to obtain a sensitivity value, wherein the sensitivity value is used to adjust the tolerance level of the predicted threshold sequence; The tolerance coefficient determination module is configured to extract multiple first index values ​​that satisfy preset alarm rules from the historical index value sequence when the sensitivity value is within a preset threshold range, and to extract multiple first prediction thresholds that have the same timestamp as the multiple first index values ​​from the prediction threshold sequence, and to perform fitting calculation on the multiple first index values ​​and the multiple first prediction thresholds to obtain a first tolerance coefficient. The first indicator threshold determination module is configured to adjust the predicted threshold sequence based on the first tolerance coefficient to obtain a target predicted threshold sequence for real-time abnormal indicator values. The target predicted threshold sequence includes multiple indicator thresholds. An indicator threshold determination device, the device comprising: One or more processors; A storage device configured to store one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the index threshold determination method according to any one of claims 1-8. A non-volatile storage medium for storing computer-executable instructions, which, when executed by a computer processor, are configured to perform the index threshold determination method according to any one of claims 1-8. A computer program product comprising a computer program, characterized in that The computer program is stored in a computer-readable storage medium, and at least one processor of the device reads from the computer-readable storage medium and executes the computer program, causing the device to perform the index threshold determination method according to any one of claims 1-8.

Images