Heart rate monitoring data intelligent analysis method based on deep learning

Abstract

The invention discloses a heart rate monitoring data intelligent analysis method based on deep learning, relates to the technical field of heart rate monitoring, and solves the problems that an original monitoring mode of this kind is one-sided, and cycle analysis cannot be performed to determine the numerical value change condition of a cycle. Characteristic analysis is carried out on heart rate data monitored by a heart rate monitor, correlation confirmation of the same frequency band is carried out on a heart rate data change curve generated in real time on the basis of specific numerical value characteristics, and then the numerical value characteristics of the same frequency band are marked in a mode of generating a correlation center line on the basis of a plurality of determined same frequency bands. According to the feature marking mode, the numerical features of the same frequency band can be fully displayed, a good feature determination effect is achieved, and the relevant accuracy of numerical feature determination is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of heart rate monitoring technology, and in particular to a method for intelligent analysis of heart rate monitoring data based on deep learning. Background Art

[0002] Patent application number CN117874689A relates to the field of data processing technology, specifically to a method for intelligent processing of heart rate monitoring data, comprising: obtaining a heart rate data sequence, obtaining a number of growth intervals and a number of outlier intervals based on the fluctuation of the heart rate data sequence, obtaining a merging standard for the outlier interval based on the difference in heart rate data in the growth interval and the outlier interval, and then obtaining a target heart rate interval, obtaining the degree of abnormality of the target heart rate interval based on the fluctuation of the heart rate data in the target heart rate interval and the proportion of the target heart rate interval in the mean of the heart rate data sequence, and obtaining the possibility of the target heart rate interval being an abnormal heart rate interval based on the difference in abnormality degree and duration between the target heart rate interval and adjacent target heart rate intervals, and then obtaining abnormal heart rate data. The present invention is based on a region growing algorithm to classify heart rate data, and obtains accurate abnormal heart rate data by analyzing the differences between different categories, which facilitates external personnel to better assess health status.

[0003] Some people generally adjust the intensity of exercise and improve fat burning efficiency during exercise. However, as the exercise intensity changes, their heart rate will also change accordingly, which will also cause the heart rate data monitored by the heart rate monitor to change dramatically. In the process of intelligent analysis and processing of the heart rate data related to the heart rate monitor, the monitored heart rate data is generally compared with the set threshold. Based on the comparison results, it is analyzed whether the heart rate data is too high. However, this original monitoring method is relatively one-sided and cannot perform periodic analysis to determine the numerical changes of the period and determine the periodic characteristics, so as to facilitate relevant personnel to understand the changes in the heart rate data. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides an intelligent analysis method for heart rate monitoring data based on deep learning, which solves the problem that the original monitoring method of this type is relatively one-sided and cannot perform periodic analysis to determine the numerical changes of the period.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for intelligent analysis of heart rate monitoring data based on deep learning, comprising the following steps: S1) Define a set of monitoring periods, confirm the heart rate data monitored by the heart rate monitor at different times during the monitoring period, and assess whether the heart rate parameters of the wearer are normal based on the real-time monitored heart rate data. The specific sub-steps are as follows: Define a set of monitoring periods T, where T is a preset value. Calibrate the real-time heart rate data within this monitoring period T as XL, and compare this heart rate data XL with the preset value Y1. When XL ≥ Y1, continue normal monitoring. When XL < Y1, generate an abnormal heart rate signal and display it on the display terminal, where Y1 is the preset value. S2) Based on the different heart rate data monitored at different times during this monitoring period, a heart rate data change curve is generated in real time. Then, based on the change characteristics of the heart rate data between adjacent times, the same-frequency feature segments within the heart rate data change curve are confirmed one by one. The specific sub-steps are: S21. Based on the different heart rate data monitored at different times during the monitoring period, a heart rate data change curve of the monitoring period is generated in real time, where the horizontal axis of the curve is the timeline and the vertical axis of the curve is the heart rate parameter; S22. Mark the turning points that appear in the heart rate data change curve one by one. The trend of the line segments on both sides of the turning point is opposite. Based on the marked turning points, the heart rate data X corresponding to different turning points are confirmed in turn. i , where i represents different turning points, i=1, 2, ..., n. When i is 1, it represents the heart rate data corresponding to the first group of turning points. When i is 2, it represents the heart rate data corresponding to the second group of turning points. When i is n, it represents the heart rate data corresponding to the last group of turning points of this heart rate data change curve. S23, heart rate data X corresponding to the first set of turning points i First, identify the change difference Cz between the heart rate data between adjacent turning points, where Cz=|X j -X j+1 |, j∈i, and j≤n-1, then determine the interval Ss between two sets of adjacent turning points, and use PD=Cz÷Ss to determine its evaluation value PD. If PD≤Y2, where Y2 is a preset value, the line segment between the two sets of turning points is marked as a same-frequency feature segment. If the same-frequency feature segment appears continuously, the same-frequency feature segments that appear continuously are classified as the same-frequency feature segments of the same type; if PD>Y2, no calibration is performed; S3) After the same-frequency characteristic segments within the heart rate data change curve are confirmed, the associated midline within each group of same-frequency characteristic segments is locked. Based on the numerical changes of the associated midline, the periodic characteristics of this monitoring period are calibrated and displayed. The specific sub-steps are as follows: S31. Based on the determined same-frequency feature segment, construct a set of associated midlines in the middle of the same-frequency feature segment, and move the associated midlines up and down. The associated midlines are straight lines and perpendicular to the vertical coordinate axis. The line segments above the associated midlines within the same-frequency feature segment are marked as upper feature segments, and the line segments below the associated midlines within the same-frequency feature segment are marked as lower feature segments. S32. Determine the vertical distances between a plurality of points within the upper feature segment and the associated center line, then sum the plurality of vertical distances to determine the sum of the upper distances. Perform the same process on the lower feature segment to determine the sum of the lower distances. S33. When the position of the associated midline moves up and down, the numerical difference between the sum of the upper distance and the sum of the lower distance during each movement is recorded, and the numerical difference is ≥ 0. When the numerical difference drops to the minimum value, the specific position of the associated midline at this moment is recorded, and the heart rate data corresponding to this associated midline is calibrated as SJ; S34. Confirm the associated midlines within different same-frequency feature segments one by one, and calibrate the heart rate data corresponding to different same-frequency feature segments as SJ k , where k = 1, 2, ..., m, where m represents the total number of different associated midlines. According to the temporal relationship of different same-frequency feature segments, several groups of heart rate data SJ k Sort and confirm the heart rate data sequence {SJ1, SJ2, ..., SJ m},use: CL=(SL m -SL m-1 )+(SL m-1 -SL m-2 )+……+(SJ2-SJ1)Confirm the cycle characteristic CL of this monitoring cycle. If CL>0, this monitoring cycle is calibrated as an ascending cycle. If CL<0, this monitoring cycle is calibrated as a descending cycle. If CL=0, no calibration is performed. S4), based on the heart rate data change curve generated in this monitoring period and the determined same-frequency characteristic segments, the value change segments between the same-frequency characteristic segments are confirmed, and then based on the value change characteristics within the value change segments, the rising cycle or the falling cycle is recalibrated; the specific sub-steps are: S41. Based on the determined numerical change segment, confirm the change amplitude values of the heart rate data at adjacent moments within the numerical change segment. If the change amplitude value is ≥ 0, calibrate the change amplitude values of different adjacent moments as F. If F> Y3, where Y3 is a preset value, calibrate the change segment between adjacent moments as an abnormal segment. If F≤ Y3, no calibration is performed. S42. Different abnormal segments within different value change segments are confirmed one by one, and the line lengths of different abnormal segments are correlatively confirmed. The line lengths of several groups of different abnormal segments are then summed to determine a total value ZZ of the abnormal segment line lengths. The total value ZS of the line lengths of several groups of value change segments is then identified, and the specific proportion ZB of the abnormal segments is determined using ZB=ZZ÷ZS. S43, compare the specific proportion value ZB with the preset value Y4: If ZB>Y4, it means that this monitoring period is an abnormal fluctuation period. If this monitoring period is an increasing period, this monitoring period is marked as an abnormal fluctuation increasing period. If this monitoring period is a decreasing period, this monitoring period is marked as an abnormal fluctuation decreasing period. If ZB≤Y4, it means that this monitoring period is a normal fluctuation period. If this monitoring period is an increasing period, this monitoring period is calibrated as a normal fluctuation increasing period. If this monitoring period is a decreasing period, this monitoring period is calibrated as a normal fluctuation decreasing period.

[0006] This invention provides a deep learning-based intelligent analysis method for heart rate monitoring data. Compared with the existing technology, it has the following advantages: The present invention performs feature analysis on the heart rate data monitored by the heart rate monitor. Based on specific numerical features, the real-time generated heart rate data change curve is correlated with the same frequency band. Then, based on the determined groups of same frequency bands, the numerical features of the same frequency bands are marked by generating associated midlines. This feature marking method can fully display the numerical features of the same frequency band, achieve better feature determination results, and ensure the correlation accuracy of the numerical feature determination. The associated center line within the corresponding heart rate data change curve is confirmed, and the periodic characteristics of its monitoring period are identified based on the numerical change characteristics before and after the corresponding associated center line. The periodic characteristics are further refined through the numerical change segments between the same-frequency feature segments, and the determined periodic characteristics are displayed. Subsequently, external personnel can promptly confirm its periodic characteristics based on the calibrated monitoring period, which is convenient for the wearer to perform feature recognition of the heart rate data. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 Schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0008] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0009] Example 1 See also Figure 1 , this application provides a deep learning-based intelligent analysis method for heart rate monitoring data, including the following steps: S1) Define a set of monitoring periods, confirm the heart rate data monitored by the heart rate monitor at different times during the monitoring period, and assess whether the wearer's heart rate parameters are normal based on the real-time monitored heart rate data. The specific sub-steps of the assessment are as follows: S11. Define a set of monitoring periods T, where T is a preset value, the specific value of which is determined by the operator based on experience. Calibrate the real-time heart rate data within this monitoring period T as XL, and compare this heart rate data XL with a preset value Y1. When XL ≥ Y1, continue normal monitoring. When XL < Y1, generate an abnormal heart rate signal and display it on the display terminal. Y1 is a preset value, the specific value of which is determined by the operator based on experience. S2), based on the different heart rate data monitored at different times during the monitoring period, a heart rate data change curve is generated in real time, and then based on the change characteristics of the heart rate data between adjacent times, the same-frequency characteristic segments within the heart rate data change curve are confirmed one by one. Specifically, different heart rate data characteristics are monitored at different times. When the heart rate data changes, there are same-frequency characteristic change segments, that is, numerical segments with basically consistent numerical change characteristics. In order to analyze the specific changes of the heart rate data during the monitoring period, it is necessary to perform numerical analysis on the changes of different same-frequency characteristic change segments and determine the relevant period characteristics; The specific sub-steps for confirming the same-frequency characteristic segments one by one are as follows: S21. Based on the different heart rate data monitored at different times during the monitoring period, a heart rate data change curve for the monitoring period is generated in real time. The horizontal coordinate axis of the curve is the timeline, and the vertical coordinate axis of the curve is the heart rate parameter. Since different times correspond to different heart rate data, corresponding coordinate points can be selected in a two-dimensional coordinate system, and the heart rate data change curve can be generated based on the changes in the coordinate points. S22. Mark the turning points that appear in the heart rate data change curve one by one. The trend of the line segments on both sides of the turning point is opposite. When the front line segment of the turning point climbs up, the rear line segment will fall down. When the front line segment of the turning point falls down, the rear line segment will climb up. Based on the marked turning points, the heart rate data X corresponding to different turning points are confirmed in turn. i , where i represents different turning points, i=1, 2, ..., n. When i is 1, it represents the heart rate data corresponding to the first group of turning points. When i is 2, it represents the heart rate data corresponding to the second group of turning points. When i is n, it represents the heart rate data corresponding to the last group of turning points of this heart rate data change curve. S23, heart rate data X corresponding to the first set of turning points i First, identify the change difference Cz between the heart rate data between adjacent turning points, where Cz=|X j -Xj+1 |, j∈i, and j≤n-1, then determine the interval Ss between two sets of adjacent turning points, and use PD=Cz÷Ss to determine its evaluation value PD. If PD≤Y2, where Y2 is a preset value, the specific value is determined by the operator based on experience. The line segment between the two sets of turning points is marked as a same-frequency feature segment. If PD>Y2, no calibration is performed. If the same-frequency feature segment appears continuously, the same-frequency feature segment that appears continuously is classified as the same-frequency feature segment of the same type. For example: Assume that the marked The turning points are A, B, C, D, E, F, and G. The line segment between the two turning points A and B belongs to the same-frequency characteristic segment, and the line segment between the two turning points B and C also belongs to the same-frequency characteristic segment. Then AC are all the same-frequency characteristic segments, which means that their same-frequency change characteristics are basically the same, so they belong to the same-frequency characteristic segments. When the evaluation values between CD do not meet the relevant conditions, then the first group of same-frequency characteristic segments belongs to AC. Starting from the turning point D, the same-frequency characteristic segments are confirmed one by one. S3) After the same-frequency characteristic segments within the heart rate data change curve are confirmed, the associated midline within each group of same-frequency characteristic segments is locked. Based on the numerical changes of the associated midline, the periodic characteristics of the monitoring period are calibrated and displayed. The specific sub-steps for determining the periodic characteristics of the monitoring period are as follows: S31. Based on the determined same-frequency feature segment, construct a set of associated midlines in the middle of the same-frequency feature segment, and move the associated midlines up and down. The associated midlines are straight lines and perpendicular to the vertical coordinate axis. The line segments above the associated midlines within the same-frequency feature segment are marked as upper feature segments, and the line segments below the associated midlines within the same-frequency feature segment are marked as lower feature segments. S32. Determine the vertical distances between a plurality of points within the upper feature segment and the associated center line, then sum the plurality of vertical distances to determine the sum of the upper distances. Perform the same process on the lower feature segment to determine the sum of the lower distances. S33. When the position of the associated midline moves up and down, the numerical difference between the sum of the upper distance and the sum of the lower distance during each movement is recorded, and the numerical difference is ≥ 0. When the numerical difference drops to the minimum value, the specific position of the associated midline at this moment is recorded, and the heart rate data corresponding to this associated midline is calibrated as SJ; S34. Confirm the associated midlines within different same-frequency feature segments one by one, and calibrate the heart rate data corresponding to different same-frequency feature segments as SJ k , where k = 1, 2, ..., m, where m represents the total number of different associated midlines. According to the temporal relationship of different same-frequency feature segments, several groups of heart rate data SJ k Sort and confirm the heart rate data sequence {SJ1, SJ2, ..., SJ m},use: CL=(SL m -SL m-1 )+(SL m-1 -SL m-2 )+……+(SJ2-SJ1)Confirm the cycle characteristic CL of this monitoring cycle. If CL>0, this monitoring cycle is calibrated as an ascending cycle. If CL<0, this monitoring cycle is calibrated as a descending cycle. If CL=0, no calibration is performed. Specifically, each group of iso-frequency feature segments corresponds to heart rate data at different times. Therefore, the iso-frequency feature segments can be divided into several groups of points. For each group of points, a perpendicular line to the associated center line can be identified. Based on the perpendicular line, the vertical distance is determined. The associated center line is moved up and down. When the difference between the vertical distances is minimized during the movement, the associated center line is located at the middle characteristic position of the iso-frequency feature segment. The associated center line can fully display the characteristics of the iso-frequency feature segment. By using the correlation center line to identify its data characteristics, the numerical performance of the same-frequency characteristic segment can be effectively confirmed, which is convenient for the overall confirmation of the periodic characteristics of this monitoring period. The calculation amount is small and the accuracy of its numerical value can be guaranteed at the same time, so as to achieve better intelligent analysis effect of heart rate data.

[0010] Example 2 In the specific implementation process of this embodiment, compared with the above-mentioned embodiment 1, the specific implementation process of this embodiment mainly performs correlation analysis on the change segments between the same-frequency characteristic segments to analyze the numerical fluctuation characteristics of this monitoring period, so as to facilitate management personnel to better confirm the characteristics of this monitoring period; The following steps are also included: S4), based on the heart rate data change curve generated in this monitoring period and the determined same-frequency characteristic segments, confirm the value change segments between the same-frequency characteristic segments, and then recalibrate the rising period or falling period based on the value change characteristics within the value change segments. Specifically, if the heart rate data corresponding to the rising period or falling period fluctuates slightly or severely during the change, re-determination is required to ensure the overall intelligent analysis effect of the heart rate data; The specific sub-steps of recalibration are: S41. Based on the determined numerical variation segment, confirm the variation amplitude of the heart rate data at adjacent moments within the numerical variation segment. If the variation amplitude is ≥ 0, calibrate the variation amplitude values at different adjacent moments as F. If F > Y3, where Y3 is a preset value, the specific value of which is determined by the operator based on experience, calibrate the variation segment between adjacent moments as an abnormal segment. If F ≤ Y3, no calibration is performed. S42. Different abnormal segments within different value change segments are confirmed one by one, and the line lengths of different abnormal segments are correlatively confirmed. The line lengths of several groups of different abnormal segments are then summed to determine a total value ZZ of the abnormal segment line lengths. The total value ZS of the line lengths of several groups of value change segments is then identified, and the specific proportion ZB of the abnormal segments is determined using ZB=ZZ÷ZS. S43. Compare the specific proportion ZB with the preset value Y4, where the specific value of Y4 is determined by the operator based on experience: If ZB>Y4, it means that this monitoring period is an abnormal fluctuation period. If this monitoring period is an increasing period, this monitoring period is marked as an abnormal fluctuation increasing period. If this monitoring period is a decreasing period, this monitoring period is marked as an abnormal fluctuation decreasing period. If ZB≤Y4, it means that this monitoring period is a normal fluctuation period. If this monitoring period is an increasing period, this monitoring period is calibrated as a normal fluctuation increasing period. If this monitoring period is a decreasing period, this monitoring period is calibrated as a normal fluctuation decreasing period. In order to facilitate external relevant personnel to identify the monitoring characteristics of such monitoring cycles, the fluctuation state of the monitoring cycle is identified based on the corresponding numerical fluctuation characteristics, and then the fluctuation of the monitoring cycle is calibrated. Subsequently, external personnel can promptly confirm the cycle characteristics based on the calibrated monitoring cycle, which is convenient for management personnel to identify the characteristics of heart rate data.

[0011] Example 3 The specific implementation process of this embodiment includes the entire implementation process of the above two groups of embodiments.

[0012] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0013] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An intelligent analysis method for heart rate monitoring data based on deep learning, characterized in that: The following steps are involved: S1) defining a set of monitoring periods, confirming the heart rate data monitored by the heart rate monitor at different times during the monitoring period, and assessing whether the heart rate parameters of the wearer are normal based on the real-time monitored heart rate data; S2), based on the different heart rate data monitored at different times during this monitoring period, generating a heart rate data change curve in real time, and then confirming the same-frequency feature segments within the heart rate data change curve one by one based on the change characteristics of the heart rate data between adjacent times; S3) After the same-frequency characteristic segments within the heart rate data change curve are confirmed, the associated midline within each group of same-frequency characteristic segments is locked, and based on the numerical changes of the associated midline, the periodic characteristics of this monitoring period are calibrated and displayed; S4) Based on the heart rate data change curve generated in this monitoring period and the determined same-frequency characteristic segments, the value change segments between the same-frequency characteristic segments are confirmed, and then based on the value change characteristics within the value change segments, the rising cycle or the falling cycle is recalibrated.

2. The method for intelligent analysis of heart rate monitoring data based on deep learning according to claim 1, characterized in that: In step S1, the specific sub-steps for assessing whether the heart rate parameters of the wearer are normal are: A set of monitoring periods T is defined, where T is a preset value. The real-time monitored heart rate data within this monitoring period T is calibrated as XL, and this heart rate data XL is compared with the preset value Y1. When XL ≥ Y1, normal monitoring is continued. When XL < Y1, an abnormal heart rate signal is generated and displayed through the display terminal, where Y1 is a preset value.

3. The method for intelligent analysis of heart rate monitoring data based on deep learning according to claim 1, characterized in that: In step S2, the specific sub-steps of confirming the same-frequency characteristic segments one by one are: S21. Based on the different heart rate data monitored at different times during the monitoring period, a heart rate data change curve of the monitoring period is generated in real time, where the horizontal axis of the curve is the timeline and the vertical axis of the curve is the heart rate parameter; S22. Mark the turning points that appear in the heart rate data change curve one by one. The trend of the line segments on both sides of the turning point is opposite. Based on the marked turning points, the heart rate data X corresponding to different turning points are confirmed in turn. i , where i represents different turning points, i=1, 2, ..., n. When i is 1, it represents the heart rate data corresponding to the first group of turning points. When i is 2, it represents the heart rate data corresponding to the second group of turning points. When i is n, it represents the heart rate data corresponding to the last group of turning points of this heart rate data change curve. S23, heart rate data X corresponding to the first set of turning points i First, identify the change difference Cz between the heart rate data between adjacent turning points, where Cz=|X j -X j+1 |, j∈i, and j≤n-1, then determine the interval time period Ss between two groups of adjacent turning points, and use PD=Cz÷Ss to determine its evaluation value PD. If PD≤Y2, where Y2 is a preset value, the line segment between the two groups of turning points is marked as a same-frequency feature segment. If the same-frequency feature segment appears continuously and uninterruptedly, the continuously appearing same-frequency feature segments are divided into the same type of same-frequency feature segments.

4. The method for intelligent analysis of heart rate monitoring data based on deep learning according to claim 3, characterized in that: In step S23, if PD>Y2, no calibration is performed.

5. The method for intelligent analysis of heart rate monitoring data based on deep learning according to claim 3, characterized in that: In step S3, the specific sub-steps for determining the periodic characteristics of the monitoring period are: S31. Based on the determined same-frequency feature segment, construct a set of associated midlines in the middle of the same-frequency feature segment, and move the associated midlines up and down. The associated midlines are straight lines and perpendicular to the vertical coordinate axis. The line segments above the associated midlines within the same-frequency feature segment are marked as upper feature segments, and the line segments below the associated midlines within the same-frequency feature segment are marked as lower feature segments. S32. Determine the vertical distances between a plurality of points within the upper feature segment and the associated center line, then sum the plurality of vertical distances to determine the sum of the upper distances. Perform the same process on the lower feature segment to determine the sum of the lower distances. S33. When the position of the associated midline moves up and down, the numerical difference between the sum of the upper distance and the sum of the lower distance during each movement is recorded, and the numerical difference is ≥ 0. When the numerical difference drops to the minimum value, the specific position of the associated midline at this moment is recorded, and the heart rate data corresponding to this associated midline is calibrated as SJ; S34. Confirm the associated midlines within different same-frequency feature segments one by one, and calibrate the heart rate data corresponding to different same-frequency feature segments as SJ k , where k = 1, 2, ..., m, where m represents the total number of different associated midlines. According to the temporal relationship of different same-frequency feature segments, several groups of heart rate data SJ k Sort and confirm the heart rate data sequence {SJ1, SJ2, ..., SJ m },use: CL=(SL m -SL m-1 )+(SL m-1 -SL m-2 )+……+(SJ2-SJ1)confirm the periodic characteristic CL of this monitoring period. If CL>0, this monitoring period is calibrated as a rising period. If CL<0, this monitoring period is calibrated as a falling period.

6. The method for intelligent analysis of heart rate monitoring data based on deep learning according to claim 5, characterized in that: In step S34, if CL=0, no calibration is performed.

7. The method for intelligent analysis of heart rate monitoring data based on deep learning according to claim 5, characterized in that: In step S4, the specific sub-steps for recalibrating the rising period or the falling period are: S41. Based on the determined numerical change segment, confirm the change amplitude values of the heart rate data at adjacent moments within the numerical change segment. If the change amplitude value is ≥ 0, calibrate the change amplitude values of different adjacent moments as F. If F> Y3, where Y3 is a preset value, calibrate the change segment between adjacent moments as an abnormal segment. If F≤ Y3, no calibration is performed. S42. Different abnormal segments within different value change segments are confirmed one by one, and the line lengths of different abnormal segments are correlatively confirmed. The line lengths of several groups of different abnormal segments are then summed to determine a total value ZZ of the abnormal segment line lengths. The total value ZS of the line lengths of several groups of value change segments is then identified, and the specific proportion ZB of the abnormal segments is determined using ZB=ZZ÷ZS. S43, compare the specific proportion value ZB with the preset value Y4: If ZB>Y4, it means that this monitoring period is an abnormal fluctuation period. If this monitoring period is an increasing period, this monitoring period is marked as an abnormal fluctuation increasing period. If this monitoring period is a decreasing period, this monitoring period is marked as an abnormal fluctuation decreasing period. If ZB≤Y4, it means that this monitoring period is a normal fluctuation period. If this monitoring period is an increasing period, this monitoring period is calibrated as a normal fluctuation increasing period. If this monitoring period is a decreasing period, this monitoring period is calibrated as a normal fluctuation decreasing period.

Images