CN115590505B - Data anomaly analysis method for intelligent motion monitoring device - Google Patents

Data anomaly analysis method for intelligent motion monitoring device Download PDF

Info

Publication number
CN115590505B
CN115590505B CN202211609597.0A CN202211609597A CN115590505B CN 115590505 B CN115590505 B CN 115590505B CN 202211609597 A CN202211609597 A CN 202211609597A CN 115590505 B CN115590505 B CN 115590505B
Authority
CN
China
Prior art keywords
moment
data
motion data
value
motion
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202211609597.0A
Other languages
Chinese (zh)
Other versions
CN115590505A (en
Inventor
赵永梅
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nannan Juzhi Information Technology Co ltd
Original Assignee
Nannan Juzhi Information Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nannan Juzhi Information Technology Co ltd filed Critical Nannan Juzhi Information Technology Co ltd
Priority to CN202211609597.0A priority Critical patent/CN115590505B/en
Publication of CN115590505A publication Critical patent/CN115590505A/en
Application granted granted Critical
Publication of CN115590505B publication Critical patent/CN115590505B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/01—Measuring temperature of body parts ; Diagnostic temperature sensing, e.g. for malignant or inflamed tissue
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/02—Detecting, measuring or recording pulse, heart rate, blood pressure or blood flow; Combined pulse/heart-rate/blood pressure determination; Evaluating a cardiovascular condition not otherwise provided for, e.g. using combinations of techniques provided for in this group with electrocardiography or electroauscultation; Heart catheters for measuring blood pressure
    • A61B5/0205—Simultaneously evaluating both cardiovascular conditions and different types of body conditions, e.g. heart and respiratory condition
    • A61B5/02055—Simultaneously evaluating both cardiovascular condition and temperature
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
    • A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
    • A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/74—Details of notification to user or communication with user or patient ; user input means
    • A61B5/746—Alarms related to a physiological condition, e.g. details of setting alarm thresholds or avoiding false alarms
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Abstract

The invention relates to the technical field of data processing, in particular to a data anomaly analysis method for an intelligent motion monitoring device, which comprises the following steps: acquiring motion data of a human body at each moment in a complete motion process, and calculating volatility indexes corresponding to the motion data at each moment; acquiring characteristic parameters corresponding to all moments; further calculating a first preferred value, a second preferred value, a third preferred value, a fourth preferred value and a fifth preferred value corresponding to the motion data at each moment; respectively taking the motion data at the time corresponding to the maximum value of the first, second, third, fourth and fifth preference values as a first, second, third, fourth and fifth clustering centers; calculating similarity indexes of the motion data at each moment and each clustering center, and further classifying to obtain five categories; calculating probability indexes of data in all categories, determining normal data and abnormal data, and further calculating abnormal degree; and carrying out early warning according to the abnormal degree. The method has a good effect of analyzing the abnormity of the motion data.

Description

Data anomaly analysis method for intelligent motion monitoring device
Technical Field
The invention relates to the technical field of data processing, in particular to a data anomaly analysis method for an intelligent motion monitoring device.
Background
With the continuous development of the current society and the continuous increase of the social pressure, the health problems of people are more and more, more people pay attention to the health, and start to do physical exercises in idle time to improve the physical quality. Meanwhile, an intelligent motion monitoring device capable of monitoring the motion state in real time is also provided. Furthermore, it is very important to analyze the data of the intelligent motion monitoring device for abnormalities.
The traditional motion data anomaly analysis method is usually based on threshold monitoring, namely when one index of the motion data exceeds a set threshold, the current motion data is displayed as an abnormal state. However, for different motions of different users, the motion data is constantly changing, only a single judgment threshold standard is considered, and the change situation of the actual motion state is not considered, so that the effect of performing anomaly analysis on the motion data is poor.
Disclosure of Invention
In order to solve the above technical problems, an object of the present invention is to provide a method for analyzing data abnormality of an intelligent motion monitoring device, wherein the adopted technical scheme is as follows:
acquiring motion data of a human body at each moment in a complete motion process, and obtaining volatility indexes corresponding to the motion data at each moment according to the corresponding difference of the motion data at adjacent moments; for any moment, the number of the moments when the numerical value of the motion data is the same as the moment is obtained and recorded as the characteristic parameter corresponding to the moment;
respectively calculating a first preferred value, a second preferred value and a third preferred value corresponding to the motion data at the moment according to the numerical value of the motion data and the corresponding moment, fluctuation index and characteristic parameter; respectively calculating a fourth preferred value and a fifth preferred value corresponding to the motion data at the moment according to the moment, the volatility index and the characteristic parameter corresponding to the motion data;
respectively taking the motion data at the time corresponding to the maximum value of the first, second, third, fourth and fifth preference values as a first, second, third, fourth and fifth clustering centers; calculating similarity indexes of the motion data at each moment and each clustering center, and classifying the motion data at each moment according to the similarity indexes to obtain five categories;
obtaining probability indexes corresponding to the motion data of each moment according to the similarity indexes corresponding to the motion data of each moment in each category, determining normal data and abnormal data according to the probability indexes and a probability threshold value, and calculating the abnormal degree according to the difference between the normal data and the abnormal data; and carrying out early warning according to the abnormal degree.
Preferably, the obtaining of the volatility index corresponding to the motion data at each time according to the difference corresponding to the motion data at adjacent times specifically includes:
selecting any moment as the current moment, respectively acquiring a fixed number of adjacent moments before and after the current moment, respectively calculating the difference between the motion data of the adjacent moments and the motion data of the current moment, obtaining the volatility index corresponding to the motion data of the current moment according to the difference, and further obtaining the volatility index corresponding to the motion data of each moment.
Preferably, the method for acquiring the first preferred value, the second preferred value and the third preferred value specifically comprises:
recording any one time as a target time; recording the ratio of the target time to the time length required for finishing the movement as a first ratio, and recording the ratio of the characteristic parameter corresponding to the movement data of the target time to the time length required for finishing the movement as a second ratio; obtaining a first preferred value corresponding to the motion data of the target moment according to the difference value between the 1 and the first ratio, the second ratio, the value of the motion data and the volatility index; obtaining a second optimized value corresponding to the motion data at the target moment according to the first ratio, the second ratio, the value of the motion data and the volatility index; and calculating an absolute value of a difference value between half of the time length required for finishing the movement and the target moment, and obtaining a third optimal value corresponding to the movement data of the target moment according to the absolute value of the difference value, the second ratio, the numerical value of the movement data and the volatility index.
Preferably, the fourth preferred value and the fifth preferred value are obtained by a method specifically comprising:
obtaining a fourth preferred value corresponding to the motion data at the target moment according to the difference value between 1 and the first ratio, the difference value between 1 and the second ratio and the fluctuation index of the motion data; and obtaining a fifth preferred value corresponding to the motion data of the target moment according to the first ratio, the difference value between 1 and the second ratio and the fluctuation index of the motion data.
Preferably, the calculating of the similarity index between the motion data at each time and each cluster center specifically includes:
recording any moment as a selected moment, calculating a difference value between the motion data of the selected moment and the motion data of the clustering center, calculating a difference value between the selected moment and the moment corresponding to the clustering center, and calculating similarity indexes of the selected moment and the first clustering center, the second clustering center and the third clustering center respectively according to the two difference values; obtaining difference values of the motion data of the selected moment and each adjacent moment, recording the maximum value of the difference values as the maximum amplitude value, and recording the minimum value of the difference values as the minimum amplitude value; acquiring an initial moment and a termination moment of movement, and acquiring a similarity index between the selected moment and a fourth clustering center according to a difference between the selected moment and the initial moment, a difference between a maximum amplitude and a minimum amplitude and a difference between the selected moment and movement data of the fourth clustering center; and obtaining a similarity index between the selected time and the fifth clustering center according to the difference between the time and the termination time, the difference between the maximum amplitude and the minimum amplitude and the difference between the motion data of the selected time and the motion data of the fifth clustering center, and further obtaining the similarity index between the motion data of each time and each clustering center.
Preferably, the determining normal data and abnormal data according to the probability index and the probability threshold specifically includes:
and recording the motion data at the moment corresponding to the probability index being greater than the probability threshold as abnormal data, and recording the motion data at the moment corresponding to the probability index being less than or equal to the probability threshold as normal data.
Preferably, the method for acquiring the degree of abnormality specifically includes:
in a category corresponding to any one clustering center, acquiring abnormal data with the same time interval to form an abnormal data sequence, acquiring normal data with the same element number as that in the abnormal data sequence according to a fixed time interval to form a normal data sequence, wherein the time interval between the normal data in the normal data sequence is equal to the time interval between the abnormal data in the abnormal data sequence; and calculating the abnormal degree corresponding to the category according to the difference between the corresponding position elements in the abnormal data sequence and the normal data sequence.
Preferably, the early warning according to the abnormal degree specifically includes:
acquiring comprehensive abnormal degree of the motion data according to the abnormal degree, and setting a first threshold, a second threshold and a third threshold;
when the comprehensive abnormal degree is larger than a first threshold value, a red early warning is sent out;
when the comprehensive abnormal degree is smaller than or equal to the first threshold and larger than the second threshold, an orange early warning is sent out;
when the comprehensive abnormal degree is smaller than or equal to the second threshold and larger than the third threshold, a yellow early warning is sent out;
and when the comprehensive abnormal degree is less than or equal to the third threshold, not sending out early warning.
The embodiment of the invention at least has the following beneficial effects:
the method obtains the motion data of the human body at each moment in the one-time complete motion process, and further analyzes the fluctuation condition of the motion data at each moment to obtain the volatility index; acquiring the number of moments with the same numerical value of the motion data as characteristic parameters corresponding to all moments, and representing the frequency situation of the motion data at the corresponding moments; furthermore, the motion data at each moment is calculated as the optimal value of the clustering centers of the corresponding classes in different stages, the motion data in different stages are considered to have different characteristics, and then the similarity index is calculated to divide the motion data at each moment into the classes corresponding to different motion stages, namely, the invention carries out self-adaptive clustering operation according to the motion data corresponding to different motion stages to obtain the better classification result of the motion data, namely, the motion data is divided into five stage classes, the motion data in different stages are respectively analyzed for the abnormal degree, and corresponding early warning is made, so that the judgment precision of abnormal data is improved, the precision of real-time motion monitoring is further improved, the possibility of motion danger is reduced, and the effect of carrying out abnormal analysis on the motion data is better.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of a method for analyzing data abnormality of an intelligent motion monitoring device according to the present invention.
Detailed Description
To further illustrate the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description of the data anomaly analysis method of the intelligent motion monitoring device according to the present invention, its specific implementation, structure, features and effects, with reference to the accompanying drawings and preferred embodiments, is provided below. In the following description, the different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following describes a specific scheme of the data anomaly analysis method for the intelligent motion monitoring device provided by the invention in detail with reference to the accompanying drawings.
Example (b):
referring to fig. 1, a flowchart of a method for analyzing data anomaly of an intelligent motion monitoring device according to an embodiment of the present invention is shown, where the method includes the following steps:
acquiring motion data of a human body at each moment in a complete motion process, and obtaining a volatility index corresponding to the motion data at each moment according to the difference between the motion data at adjacent moments; and for any moment, the number of the moments when the numerical value of the acquired motion data is the same as that of the moment is recorded as the characteristic parameter corresponding to the moment.
Firstly, motion data of a human body at each moment in a one-time complete motion process needs to be acquired, wherein the one-time complete motion process refers to a time period corresponding to the human body from the beginning of motion to the end of motion, namely the motion data corresponding to each moment in the time period is acquired. Meanwhile, in the embodiment, the exercise data comprise heart rate data, body temperature data, pulse data and respiration data of the human body in the exercise process, an implementer can select more interesting data to analyze according to different exercises, various exercise data acquisition methods are diversified, and the implementer can select different intelligent exercise monitoring devices to acquire according to actual conditions.
It should be noted that, considering that the same motion data may also generate larger fluctuations in different phases of the motion, if the judgment is performed only by a single threshold, normal data in the motion process may be erroneously judged as abnormal data, in the embodiment of the present invention, the motion is divided into different motion phases, adaptive clustering processing is performed according to the motion data corresponding to the different motion phases, a better classification result of the motion data is obtained, the motion data of different classes are respectively analyzed, the judgment accuracy of the abnormal data is improved, further, the accuracy of real-time monitoring on the motion is improved, and the possibility of motion danger is reduced.
Then, in this embodiment, the motion process is analyzed, and the motion process is divided into five different stages according to the characteristics of the motion, including:
in the initial stage, the human body starts to move, and the fluctuation of the motion data of the human body corresponding to each moment is relatively gentle and is slowly rising.
And in the rising stage, after a period of time passes by the movement of the human body, the movement data of the human body corresponding to each moment have a certain fluctuation degree compared with the initial stage, the phenomenon that the human body starts to do strenuous movement so as to achieve the effect of exercising the body is represented, and the movement data corresponding to the human body rises rapidly.
In the training stage, after the human body moves for a relatively long period of time, the motion data of the human body corresponding to each moment reach a stable state again, for example, after the human body moves for a relatively long period of time, the heart rate data continuously increases from a small value to a certain degree and stops increasing, which indicates that the human body gradually adapts to the motion, and the physical sign data of the human body keeps fluctuating at a certain amplitude.
And a descending stage, in which the human body moves for a long period of time, the state is that the motion is about to be finished, and the human body motion data corresponding to each moment has a certain fluctuation degree compared with the training stage, for example, after the human body moves for a long period of time, the heart rate data continuously descends from a larger value in the training stage, and the descending speed is slower.
And a termination stage, in which the human body stops moving, and the motion data of the human body corresponding to each moment gradually returns to normal until the motion data is stable.
Furthermore, according to the variation characteristics of the motion data in the five stages, the motion data at each time is divided into five categories, and then the motion data at each time needs to be analyzed first, so as to analyze the effect that the motion data at each time is respectively used as the clustering centers of the categories corresponding to the five stages.
Specifically, since the exercise data includes a plurality of kinds of data of the human body, and the same processing is performed for each kind of data, in the present embodiment, the heart rate data of the human body is exemplified. And acquiring the time length of the human body in the whole movement process, namely the time length is the total number of moments contained in the whole movement process.
Because the motion data of the human body have different fluctuation change characteristics in different motion stages, the fluctuation change characteristics of the motion data at each moment can be analyzed through the difference between the motion data at adjacent moments. That is, any one time is selected to be recorded as the current time, and a fixed number of adjacent times are respectively obtained before and after the current time, where the fixed number is 4 in this embodiment, for example, assuming that the current time is the tth time, the t-4 th time, the t-3 th time, the t-2 th time, the t-1 st time, the t +1 th time, the t +2 th time, the t +3 th time, and the t +4 th time are all the adjacent times.
Respectively calculating the difference between the motion data of each adjacent moment and the motion data of the current moment, obtaining the volatility index corresponding to the motion data of the current moment according to the difference, and expressing the volatility index by a formula as follows:
Figure DEST_PATH_IMAGE001
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE003
the fluctuation index corresponding to the motion data at the t-th moment is shown,
Figure DEST_PATH_IMAGE005
representing the motion data at the t-th instant,
Figure DEST_PATH_IMAGE007
representing the total number of adjacent time instants, in this embodiment
Figure 650372DEST_PATH_IMAGE007
The value of (a) is 8,
Figure DEST_PATH_IMAGE009
the motion data of the ith adjacent time is represented, for example, if the t-4 th time is the first adjacent time, the motion data of the first adjacent time is the motion data of the t-4 th time.
Figure DEST_PATH_IMAGE011
The difference between the motion data of the adjacent time and the motion data of the current time is represented, and the larger the difference is, the larger the value of the fluctuation index corresponding to the motion data of the current time is, which indicates that the larger the difference between the motion data of the current time and the motion data of the adjacent time is, the larger the fluctuation of the motion data is.
And obtaining the fluctuation indexes corresponding to the motion data at all times according to the method, wherein the fluctuation degrees of the motion data corresponding to different stages are different, so that specific analysis needs to be carried out subsequently according to specific conditions.
And finally, for any moment, the number of the moments when the numerical value of the motion data is the same as that of the moment is acquired and recorded as the characteristic parameter corresponding to the moment, and further the characteristic parameter corresponding to each moment is acquired. For example, before counting the heart rate data of the human body, let m =0, when the heart rate data of some other time is equal to the heart rate data of the current time, let m = m +1, otherwise, let m = m, traverse the motion data of all other times except the current time, and obtain a final value of m, which is the characteristic parameter corresponding to the current time.
Step two, respectively calculating a first preferred value, a second preferred value and a third preferred value corresponding to the motion data at the moment according to the numerical value of the motion data and the corresponding moment, fluctuation index and characteristic parameter; and respectively calculating a fourth preferred value and a fifth preferred value corresponding to the movement data at the moment according to the moment, the volatility index and the characteristic parameter corresponding to the movement data.
First, it should be noted that when the human body is in the initial stage of movement, the fluctuation of the movement data at each time is relatively gentle and is rising slowly. Therefore, when the motion data is selected as the clustering center of the corresponding category in the initial stage, the fluctuation degree corresponding to the motion data should be small, and meanwhile, in order to make the classification result good, the value of the characteristic parameter corresponding to the motion data should be large.
Based on this, any one time is taken as a target time, the ratio of the target time to the time length required for completing the exercise is taken as a first ratio, the ratio of the characteristic parameter corresponding to the exercise data of the target time to the time length required for completing the exercise is taken as a second ratio, and a first preferred value corresponding to the exercise data of the target time is obtained according to the difference between 1 and the first ratio, the second ratio, the numerical value of the exercise data and the volatility index, and is expressed as follows by a formula:
Figure 551201DEST_PATH_IMAGE012
wherein the content of the first and second substances,
Figure 799779DEST_PATH_IMAGE014
a first preferred value corresponding to the motion data at the tth moment is represented, t is a numerical value of a moment serial number corresponding to the tth moment, for example, the moment serial number corresponding to the 3 rd moment is 3; t represents the length of time required to complete the movement, i.e. the total number of moments contained within the time period required to complete the movement,
Figure 21420DEST_PATH_IMAGE016
representing a first ratio;
Figure DEST_PATH_IMAGE017
the fluctuation index corresponding to the motion data at the t-th moment is shown,
Figure 366950DEST_PATH_IMAGE018
representing the motion data at the t-th instant,
Figure 782888DEST_PATH_IMAGE020
the characteristic parameter corresponding to the t-th moment is shown,
Figure 377818DEST_PATH_IMAGE022
a second ratio is indicated.
First preferred value
Figure 45559DEST_PATH_IMAGE014
And representing the optimization degree of the motion data at the t-th moment as the clustering center of the corresponding category in the initial stage, wherein the larger the value of the first optimization value is, the larger the optimization degree of the motion data at the t-th moment as the clustering center of the corresponding category in the initial stage is, and the better the classification effect by using the clustering center is.
First ratio
Figure 543799DEST_PATH_IMAGE016
And representing the ratio of the t moment in the time length required for finishing the movement, wherein if the t moment belongs to the initial stage of the movement, the value of the first ratio is smaller. On the basis of this, the method is suitable for the production,
Figure 865059DEST_PATH_IMAGE024
the larger the value of (2) is, the smaller the value of the first ratio is, the more likely the tth moment belongs to the initial stage of the motion, the larger the value of the first preferred value is, the larger the preferred degree of the motion data at the tth moment as the clustering center of the corresponding category of the initial stage is, and the better the effect of classifying by using the clustering center is. I.e. the relationship between the first ratio and the first preferred value is a negative correlation.
The volatility index represents the fluctuation condition of the motion data at the current moment, and if the tth moment belongs to the initial stage of motion, the value of the volatility index corresponding to the motion data at the moment is small. Based on this, the smaller the value of the volatility index is, the smaller the difference between the motion data at the current moment and the motion data at the adjacent moment is, the smaller the fluctuation of the motion data at the t moment is, the larger the value of the corresponding first preferred value is, the larger the preferred degree of the motion data at the t moment as the clustering center of the corresponding category at the initial stage is, and the better the effect of classification by using the clustering center is. I.e. the relation between the volatility indicator and the first preferred value is a negative correlation, the denominator plus 1 is to prevent the denominator being 0.
Second ratio
Figure 681705DEST_PATH_IMAGE022
The method comprises the steps of representing the ratio of the number of moments which are the same as motion data at the t-th moment to the length of time required for finishing motion, wherein the larger the value of the second ratio is, the larger the number of moments which are the same as the motion data at the t-th moment in other moments is, the larger the value of the first preferred value is, the larger the preferred degree of the motion data at the t-th moment as a clustering center of a corresponding category at an initial stage is, and the better the classification effect by utilizing the clustering center is.
Since the value of the motion data is smaller than the values of the motion data in other phases when the motion data is in the initial phase of the motion, the relationship between the motion data and the first preferred value is a negative correlation relationship, that is, the smaller the value of the motion data is, the larger the value of the first preferred value is.
Then, it should be noted that when the human body is in the termination stage of the movement, the movement data at each time gradually returns to normal until stable. Therefore, when the motion data is selected as the clustering center of the corresponding category at the termination stage, the fluctuation degree corresponding to the motion data should be small, and meanwhile, in order to make the classification result good, the value of the characteristic parameter corresponding to the motion data should be large.
Based on the first ratio, the second ratio, the value of the motion data and the volatility index, a second optimal value corresponding to the motion data of the target moment is obtained, and the optimal value is expressed by a formula as follows:
Figure DEST_PATH_IMAGE025
wherein, the first and the second end of the pipe are connected with each other,
Figure DEST_PATH_IMAGE027
a second preferred value corresponding to the motion data at the t-th moment is represented, where t is a numerical value of a moment serial number corresponding to the t-th moment, for example, a moment serial number corresponding to the 3 rd moment is 3; t represents the length of time required to complete the movement, i.e. the total number of moments contained within the time period required to complete the movement,
Figure 776307DEST_PATH_IMAGE016
representing a first ratio;
Figure 627588DEST_PATH_IMAGE003
the fluctuation index corresponding to the motion data at the t-th moment is shown,
Figure 526274DEST_PATH_IMAGE005
representing the motion data at the t-th instant,
Figure 95796DEST_PATH_IMAGE020
the characteristic parameter corresponding to the t-th moment is shown,
Figure 698815DEST_PATH_IMAGE022
the second ratio is indicated.
Second preferred value
Figure 811128DEST_PATH_IMAGE027
And representing the preference degree of the motion data at the t-th moment as the clustering center of the corresponding category at the termination stage, wherein the larger the value of the second preference value is, the larger the preference degree of the motion data at the t-th moment as the clustering center of the corresponding category at the termination stage is, and the better the effect of classifying by using the clustering center is.
First ratio
Figure 975655DEST_PATH_IMAGE016
And representing the ratio of the t moment in the time length required for finishing the movement, wherein if the t moment belongs to the termination stage of the movement, the value of the first ratio is larger than the value belonging to the initial stage. Based on this, the larger the value of the first ratio is, the more likely the tth moment belongs to the motion termination stage, the larger the value of the second preferred value is, the larger the preferred degree of the motion data at the tth moment as the cluster center of the corresponding category of the termination stage is, and the better the effect of classifying by using the cluster center is. I.e. the relationship between the first ratio and the second preferred value is a positive correlation.
The volatility index represents the fluctuation condition of the motion data at the current moment, and if the tth moment belongs to the termination stage of motion, the value of the volatility index corresponding to the motion data at the moment should be smaller. Based on this, the smaller the value of the volatility index is, the smaller the fluctuation of the motion data at the t-th moment is, and the more stable the motion data at the moment is, the larger the value of the second preferred value is, the greater the preferred degree of the motion data at the t-th moment as the clustering center of the corresponding category at the termination stage is, and the better the classification effect by using the clustering center is. I.e. the relation between the volatility indicator and the second preferred value is a negative correlation, the denominator plus 1 is to prevent the denominator being 0.
The larger the value of the second ratio is, the more the number of the moments which are the same as the motion data of the t-th moment in other moments is, the larger the value of the second preferred value is; when the motion data is in the termination stage of the motion, the value of the motion data is smaller than the values of the motion data in other stages, so that the relationship between the value of the motion data and the second preferred value is a negative correlation relationship, that is, the smaller the value of the motion data is, the larger the value of the second preferred value is, the greater the preferred degree of the motion data at the t-th moment as the clustering center of the corresponding category in the termination stage is, and the better the classification effect by using the clustering center is.
Secondly, it should be noted that when the human body is in the training phase of exercise, the exercise data at each moment reaches a stable state again, and has a small fluctuation condition. Therefore, when the motion data is selected as the clustering center of the corresponding category in the training stage, the fluctuation degree corresponding to the motion data should be small, and meanwhile, in order to make the classification result good, the value of the characteristic parameter corresponding to the motion data should be large.
Based on the above, calculating the absolute value of the difference between half of the time length required for completing the movement and the target time, obtaining a third preferred value corresponding to the movement data of the target time according to the absolute value of the difference, the second ratio, the numerical value of the movement data and the volatility index, and expressing the third preferred value as follows:
Figure 173418DEST_PATH_IMAGE028
wherein the content of the first and second substances,
Figure 580129DEST_PATH_IMAGE030
a third preferred value corresponding to the motion data at the t-th moment is represented, where t is a numerical value of a moment serial number corresponding to the t-th moment, and for example, a moment serial number corresponding to the 3 rd moment is 3; t represents the length of time required to complete the movement, i.e. the total number of moments contained in the time period required to complete the movement;
Figure DEST_PATH_IMAGE031
the fluctuation index corresponding to the motion data at the t-th moment is shown,
Figure 874844DEST_PATH_IMAGE018
representing the motion data at the t-th instant,
Figure 974387DEST_PATH_IMAGE020
the characteristic parameter corresponding to the t-th moment is shown,
Figure 485878DEST_PATH_IMAGE022
representing the second ratio, e being a natural constant.
Third preferred value
Figure 961858DEST_PATH_IMAGE030
The optimization degree of the motion data at the t-th moment as the clustering center of the corresponding category in the training phase is represented, and the larger the value of the third optimization value is, the larger the optimization degree of the motion data at the t-th moment as the clustering center of the corresponding category in the training phase is, the better the effect of classification by using the clustering center is.
Figure DEST_PATH_IMAGE033
And if the tth moment belongs to the training stage, the tth moment is positioned at a more middle moment of the whole movement time, and the value of the time difference is smaller. On the basis of this, it is possible to provide,
Figure 845501DEST_PATH_IMAGE034
the smaller the value of (a) is, the more the time characteristic of the training phase is met at the t-th moment, the larger the value of the third preferred value is, the greater the preferred degree of the motion data at the t-th moment as the clustering center of the corresponding category of the training phase is, and the better the effect of classification by using the clustering center is.
The volatility index represents the fluctuation condition of the motion data at the tth moment, and if the tth moment belongs to the training stage of motion, the value of the volatility index corresponding to the motion data at the moment should be smaller. Based on this, the smaller the value of the volatility index is, the smaller the fluctuation of the motion data at the t-th moment is, the larger the value of the corresponding third preferred value is, the larger the preferred degree of the motion data at the t-th moment as the cluster center of the corresponding category in the training stage is, and the better the effect of classification by using the cluster center is. That is, the relationship between the volatility index and the third preferred value is a negative correlation, and the denominator is increased by 1 to prevent the case where the denominator is 0.
The larger the value of the second ratio is, the larger the number of times with the same motion data as the t-th time in other times is, the larger the value of the third preferred value is; when the motion data is in the training phase of the motion, the value of the motion data is larger than the values of the motion data in other phases, so that the relationship between the value of the motion data and the third preferred value is a positive correlation relationship, that is, the larger the value of the motion data is, the larger the value of the third preferred value is, the larger the preferred degree of the motion data at the t-th moment as the clustering center of the corresponding class in the training phase is, and the better the effect of classifying by using the clustering center is.
Furthermore, it should be noted that, when the human body is in the ascending stage of the movement, the movement data at each moment has a larger fluctuation degree than that in the initial stage, which represents that the human body starts to make a violent movement to achieve the effect of exercising the body, and the movement data corresponding to the human body gradually rises. Therefore, when the motion data is selected as the cluster center of the corresponding category in the ascending stage, the fluctuation degree corresponding to the motion data should be larger, and meanwhile, in order to make the classification result better, the value of the characteristic parameter corresponding to the motion data should be smaller.
Based on this, a fourth preferred value corresponding to the motion data at the target time is obtained according to the difference between 1 and the first ratio, the difference between 1 and the second ratio and the fluctuation index of the motion data, and is expressed by a formula as follows:
Figure 115945DEST_PATH_IMAGE036
wherein the content of the first and second substances,
Figure 383241DEST_PATH_IMAGE038
a fourth preferred value corresponding to the motion data at the tth moment is represented, t is a numerical value of a moment serial number corresponding to the tth moment, for example, the moment serial number corresponding to the 3 rd moment is 3; t represents the length of time required to complete the movement, i.e. the total number of moments contained within the time period required to complete the movement,
Figure 7120DEST_PATH_IMAGE016
representing a first ratio;
Figure 807586DEST_PATH_IMAGE003
the fluctuation index corresponding to the motion data at the t-th moment is shown,
Figure 983352DEST_PATH_IMAGE020
the characteristic parameter corresponding to the t-th moment is shown,
Figure 705320DEST_PATH_IMAGE022
a second ratio is indicated.
Fourth preferred value
Figure 83956DEST_PATH_IMAGE038
The preference degree of the motion data at the t-th moment as the clustering center of the corresponding category at the ascending stage is represented, and the larger the value of the fourth preference value is, the greater the preference degree of the motion data at the t-th moment as the clustering center of the corresponding category at the ascending stage is, the better the effect of classifying by using the clustering center is.
First ratio
Figure 473349DEST_PATH_IMAGE016
And if the tth moment belongs to the rising stage of the movement, the tth moment is in the first half period of the movement, and the value of the first ratio is smaller. On the basis of this, the method is suitable for the production,
Figure 226541DEST_PATH_IMAGE024
the larger the value of (3) is, the smaller the value of the first ratio is, the more likely the tth moment belongs to the rising stage of the motion, the larger the value of the fourth preferred value is, the larger the preferred degree of the motion data at the tth moment serving as the cluster center of the corresponding category of the rising stage is, and the better the effect of classifying by using the cluster center is. I.e. the relationship between the first ratio and the fourth preferred value is a negative correlation.
The volatility index represents the fluctuation condition of the motion data at the tth moment, and if the tth moment belongs to the ascending stage of motion, the value of the volatility index corresponding to the motion data at the moment is larger. Based on this, the larger the value of the volatility index is, the larger the fluctuation of the motion data at the t-th moment is, the larger the value of the corresponding fourth preferred value is, the larger the preferred degree of the motion data at the t-th moment as the cluster center of the corresponding category at the ascending stage is, and the better the effect of classification by using the cluster center is. That is, the relationship between the volatility index and the fourth preferred value is a positive correlation, and 1 is added to prevent the value of the volatility index from being 0 and further influencing the judgment of other factors.
The larger the value of the second ratio is, the larger the number of times that are the same as the motion data of the t-th time among other times is, and when the motion data is in the ascending stage of the motion, the motion number of each time gradually increases, so that the motion data of each time in the ascending stage is changed all the time. Therefore, when the value of the second ratio is smaller, the smaller the number of times that are the same as the motion data at the t-th time in other times is, the larger the value of the fourth preferred value is, the larger the preferred degree of the motion data at the t-th time as the cluster center of the corresponding category at the ascending stage is, and the better the effect of classification by using the cluster center is.
Finally, it should be noted that, when the human body is in the descending stage of the movement, the movement data at each moment has a larger fluctuation degree than that in the training stage, and represents that the human body is in a state that the movement is about to be finished after a long period of time. Therefore, when the motion data is selected as the cluster center of the corresponding category in the descending stage, the fluctuation degree corresponding to the motion data should be larger, and meanwhile, in order to make the classification result better, the value of the characteristic parameter corresponding to the motion data should be smaller.
Based on the first ratio, the difference value between 1 and the second ratio and the fluctuation index of the motion data, obtaining a fifth preferred value corresponding to the motion data at the target moment, and expressing the fifth preferred value as follows by using a formula:
Figure 966964DEST_PATH_IMAGE040
wherein, the first and the second end of the pipe are connected with each other,
Figure 824324DEST_PATH_IMAGE042
a fifth preferred value corresponding to the motion data at the tth moment is represented, t is a numerical value of a moment serial number corresponding to the tth moment, for example, the moment serial number corresponding to the 3 rd moment is 3; t denotes the length of time required to complete the movement, i.e. the total number of moments contained in the time period required to complete the movement,
Figure 474748DEST_PATH_IMAGE016
representing a first ratio;
Figure 992317DEST_PATH_IMAGE003
the fluctuation index corresponding to the motion data at the t-th moment is shown,
Figure 220036DEST_PATH_IMAGE020
the characteristic parameters corresponding to the t-th time are shown,
Figure 254988DEST_PATH_IMAGE022
a second ratio is indicated.
Fifth preferred value
Figure 618973DEST_PATH_IMAGE042
And representing the optimization degree of the motion data at the t-th moment as the clustering center of the corresponding category in the descending stage, wherein the larger the value of the fifth optimization value is, the larger the optimization degree of the motion data at the t-th moment as the clustering center of the corresponding category in the descending stage is, and the better the classifying effect by using the clustering center is.
First ratio of
Figure 829416DEST_PATH_IMAGE016
And representing the ratio of the t moment in the time length required by finishing the movement, wherein if the t moment belongs to the descending stage of the movement, the t moment is in the latter half time period of the movement, and the value of the first ratio is larger. Based on this, the larger the value of the first ratio is, the more likely the tth moment belongs to the descending stage of the movement, the larger the value of the fifth preferred value is, and the movement number at the tth moment is describedThe higher the preference degree of the cluster center corresponding to the category in the descending stage is, the better the classification effect by using the cluster center is. That is, the relationship between the first ratio and the fifth preferred value is a positive correlation.
The volatility index represents the fluctuation condition of the motion data at the tth moment, and if the tth moment belongs to the descending stage of motion, the value of the volatility index corresponding to the motion data at the moment is larger. Based on this, the larger the value of the volatility index is, the larger the fluctuation of the motion data at the t-th moment is, the larger the value of the corresponding fifth preferred value is, the larger the preferred degree of the motion data at the t-th moment as the cluster center of the corresponding category at the descending stage is, and the better the effect of classification by using the cluster center is. That is, the relationship between the volatility index and the fifth preferred value is a positive correlation, and 1 is added to prevent the value of the volatility index from being 0 and further influencing the judgment of other factors.
The larger the value of the second ratio is, the larger the number of the times which are the same as the motion data of the t-th time among other times is, and when the motion data is in the descending stage of the motion, the motion number of each time is gradually reduced, so that the motion data of each time is changed all the time in the descending stage. When the value of the second ratio is smaller, the smaller the number of times which are the same as the motion data at the t-th time in other times is, the larger the value of the fifth preferred value is, the larger the preferred degree of the motion data at the t-th time as the cluster center of the corresponding category at the descending stage is, and the better the effect of classifying by using the cluster center is.
Step three, the motion data at the moment corresponding to the maximum value of the first, second, third, fourth and fifth preference values are respectively used as a first, second, third, fourth and fifth clustering centers; and calculating similarity indexes of the motion data at each moment and each clustering center, and classifying the motion data at each moment according to the similarity indexes to obtain five categories.
First, the motion data at the time corresponding to the maximum value of the first preferred value is taken as a first clustering center, the motion data at the time corresponding to the maximum value of the second preferred value is taken as a second clustering center, the motion data at the time corresponding to the maximum value of the third preferred value is taken as a third clustering center, the motion data at the time corresponding to the maximum value of the fourth preferred value is taken as a fourth clustering center, and the motion data at the time corresponding to the maximum value of the fifth preferred value is taken as a fifth clustering center.
The category corresponding to the first clustering center is an initial stage of movement, the category corresponding to the second clustering center is a termination stage of movement, the category corresponding to the third clustering center is a training stage of movement, the category corresponding to the fourth clustering center is an ascending stage of movement, and the category corresponding to the fifth clustering center is a descending stage of movement.
Then, in the initial stage, the termination stage and the training stage of the exercise, the exercise data at each moment are kept in a stable state, that is, the difference between the exercise data at each moment is small, and meanwhile, each moment is relatively close. Based on this, any one time is taken as a selected time, the difference between the motion data of the selected time and the motion data of the cluster center is calculated, the difference between the selected time and the time corresponding to the cluster center is calculated, the similarity indexes of the selected time and the first cluster center, the second cluster center and the third cluster center are respectively calculated according to the two differences, and then the similarity indexes of the motion data of each time and the first cluster center, the second cluster center and the third cluster center are obtained, and the similarity indexes are expressed by a formula as follows:
Figure DEST_PATH_IMAGE043
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE045
representing a similarity indicator between the motion data at the s-th time instant and the first cluster center, the s-th time instant being a selected time instant,
Figure DEST_PATH_IMAGE047
representing the motion data at the s-th instant,
Figure DEST_PATH_IMAGE049
representing the motion data corresponding to the first clustering center, s representing the time sequence number corresponding to the s-th time, z1 representing the time sequence number corresponding to the first clustering center, for example, the time sequence number corresponding to the 5-th time is 5; max () represents a function of taking the maximum value.
Figure DEST_PATH_IMAGE051
The larger the value of (a) is, the larger the difference between the motion data at the s-th moment and the motion data corresponding to the first clustering center is, the smaller the value of the corresponding similarity index is, the more unlikely the motion data at the s-th moment belongs to the category corresponding to the first clustering center, that is, the relationship between the difference between the motion data and the similarity index is a negative correlation relationship.
Figure DEST_PATH_IMAGE053
The larger the value of (a) is, the longer the time distance between the s-th moment and the moment corresponding to the first clustering center is, the smaller the value of the corresponding similarity index is, the less possibility that the motion data at the s moments belong to the category corresponding to the first clustering center is,
Figure DEST_PATH_IMAGE055
indicating that the difference between the time instants is normalized.
The method for calculating the similarity index between each of the other times and the second cluster center and the similarity index between each of the other times and the third cluster center are the same as those described above, that is, the method for calculating the similarity index between each of the other times and the first cluster center is the same as that described above.
Further, for the ascending phase and the descending phase of the exercise, the exercise data at each time point is in a gradually increasing state or in a gradually decreasing state, and then there is a certain difference between the exercise data at each time point. For the rising stage, the increasing speed of the exercise data at each moment is fast, for example, after the human body starts aerobic exercise for a period of time, the heart rate value starts to rise to a large value in a short time, and then the high heart rate can be maintained all the time during subsequent training. For the descending phase, the speed of the decrease of the motion data at each moment is slower than the speed of the increase of the motion data in the ascending phase, for example, after the training is completed, the human body gradually relaxes, and the heart rate value gradually decreases with the passage of time.
It should be noted that, in the ascending stage of the movement, the speed of the human body to ascend the movement data such as the heart rate value of the human body by the movement is fast, and in the descending stage of the movement, the movement data such as the heart rate value of the human body naturally descends with time, so that the descending data of the movement data in the descending stage is slow compared with the ascending speed of the movement data in the ascending stage.
Based on the difference, the difference between the motion data of the selected moment and each adjacent moment is obtained, the maximum value of the difference is recorded as the maximum amplitude, and the minimum value of the difference is recorded as the minimum amplitude. That is, any one time is selected to be the selected time, and a fixed number of adjacent times are respectively obtained before and after the selected time, where the fixed number is 4 in this embodiment, for example, if the selected time is the s-th time, the s-4 th time, the s-3 th time, the s-2 th time, the s-1 th time, the s +2 th time, the s +3 th time, and the s +4 th time are all the adjacent times. And then respectively calculating the difference values of the motion data of 8 adjacent moments and the selected moment, obtaining the maximum value of the difference values and recording the maximum value as the maximum amplitude value, and obtaining the minimum value of the difference values and recording the minimum amplitude value. The motion data at each moment corresponds to a maximum amplitude and a minimum amplitude.
Acquiring an initial moment and a termination moment of movement, and acquiring a similarity index between the selected moment and a fourth clustering center according to a difference between the selected moment and the initial moment, a difference between a maximum amplitude and a minimum amplitude and a difference between the selected moment and movement data of the fourth clustering center; and obtaining a similarity index between the time and the fifth clustering center according to the difference between the selected time and the termination time, the difference between the maximum amplitude and the minimum amplitude and the difference between the selected time and the motion data of the fifth clustering center. And then obtaining similarity indexes of the motion data at each moment, the fourth clustering center and the fifth clustering center.
The similarity indexes between the selected time and the fourth clustering center and the fifth clustering center can be expressed by a formula as follows:
Figure DEST_PATH_IMAGE057
Figure DEST_PATH_IMAGE059
Figure 154218DEST_PATH_IMAGE060
Figure DEST_PATH_IMAGE061
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE063
representing a similarity index between the motion data at the s-th time and the fourth cluster center, the s-th time being a selected time,
Figure DEST_PATH_IMAGE065
representing a similarity measure between the motion data at the s-th instant and the fifth cluster center,
Figure 209506DEST_PATH_IMAGE047
representing the motion data at the s-th instant,
Figure DEST_PATH_IMAGE067
representing the motion data corresponding to the fourth cluster center,
Figure DEST_PATH_IMAGE069
representing motion data, s, corresponding to a fifth cluster centerThe time sequence number corresponding to the s-th time is shown,
Figure DEST_PATH_IMAGE071
a time number corresponding to the initial time is indicated,
Figure DEST_PATH_IMAGE073
the time number corresponding to the termination time, e is a natural constant, max () represents a function for obtaining a maximum value, and min () represents a function for obtaining a minimum value.
Figure DEST_PATH_IMAGE075
And
Figure DEST_PATH_IMAGE077
respectively representing the maximum amplitude and the minimum amplitude at the s-th instant,
Figure DEST_PATH_IMAGE079
and
Figure DEST_PATH_IMAGE081
respectively representing the maximum amplitude and the minimum amplitude of the corresponding moment of the fourth cluster center,
Figure DEST_PATH_IMAGE083
and
Figure DEST_PATH_IMAGE085
respectively representing the maximum amplitude and the minimum amplitude of the corresponding moment of the fifth cluster center.
Figure DEST_PATH_IMAGE087
Representing the difference between the motion data at the s-th moment and the motion data of the fourth clustering center, wherein the smaller the difference is, the larger the corresponding similarity index value is, and the more likely the motion data at the s-th moment belongs to the category corresponding to the fourth clustering center;
Figure DEST_PATH_IMAGE089
representing the motion data of the s-th time and the motion number of the fifth cluster centerAccording to the difference between the motion data and the motion data, the smaller the difference is, the larger the corresponding similarity index value is, and the more likely the motion data at the s-th moment belongs to the fifth clustering center.
Figure 224735DEST_PATH_IMAGE090
And
Figure DEST_PATH_IMAGE091
indicating that the difference between the motion data is normalized.
Figure DEST_PATH_IMAGE093
The difference between the maximum amplitude and the minimum amplitude at the s-th time is represented, the change speed of the motion data at the s-th time is reflected, the larger the difference is, the faster the change speed of the motion data at the s-th time is, and the smaller the difference is, the slower the change speed of the motion data at the s-th time is.
Figure DEST_PATH_IMAGE095
And
Figure DEST_PATH_IMAGE097
the difference between the maximum amplitude and the minimum amplitude of the time corresponding to the fourth clustering center and the fifth clustering center is respectively represented, and the change speed of the motion data of the time corresponding to the fourth clustering center and the fifth clustering center is reflected. Therefore, it is
Figure DEST_PATH_IMAGE099
Representing the change situation of the motion data at the s-th moment and the moment corresponding to the fourth clustering center,
Figure DEST_PATH_IMAGE101
and representing the change conditions of the motion data at the s-th moment and the moment corresponding to the fifth clustering center.
The category corresponding to the fourth clustering center is a motion ascending stage, so that the change speed of the motion data at the moment corresponding to the fourth clustering center is high, and then
Figure 894226DEST_PATH_IMAGE099
The larger the value of (a) is, the faster the change of the motion data at the s-th moment and the moment corresponding to the fourth clustering center is, and the larger the value of the corresponding similarity index is, the more likely the s-th moment belongs to the category corresponding to the fourth clustering center. The category corresponding to the fifth clustering center is a descending stage, so that the change speed of the motion data at the moment corresponding to the fifth clustering center is slow, and then
Figure 830958DEST_PATH_IMAGE101
The smaller the value of (a) is, the slower the change of the motion data at the s-th moment and the moment corresponding to the fifth clustering center is, and the larger the value of the corresponding similarity index is, the more likely the s-th moment belongs to the category corresponding to the fifth clustering center.
Figure DEST_PATH_IMAGE103
And the difference between the s-th moment and the initial moment is shown, the smaller the difference is, the closer the s-th moment is to the initial moment, the more likely the s-th moment belongs to the rising stage, the larger the corresponding similarity index value is, and the more likely the s-th moment belongs to the category corresponding to the fourth clustering center.
Figure DEST_PATH_IMAGE105
And the difference between the s-th moment and the termination moment is represented, the smaller the difference is, the closer the s-th moment is to the termination moment is, the more likely the s-th moment belongs to a descending stage, the larger the value of the corresponding similarity index is, and the more likely the s-th moment belongs to the category corresponding to the fifth clustering center.
Figure 630549DEST_PATH_IMAGE106
And
Figure DEST_PATH_IMAGE107
indicating that the difference between the time instants is normalized.
And finally, classifying the motion data at each moment according to the similarity indexes to obtain five categories, namely recording the moment as a corresponding category by using a clustering center corresponding to the maximum value of the similarity indexes at each moment. For example, the cluster center corresponding to the maximum value of the similarity index of the motion data at the s-th time is the first cluster center, and the motion data at the s-th time belongs to the category corresponding to the first cluster center, that is, the initial stage of motion. The maximum value of the similarity index of the motion data at each moment is the maximum value of the similarity index between the motion data at each moment and the first, second, third, fourth and fifth cluster centers respectively. At this point, the classification of the motion data is completed.
Obtaining probability indexes corresponding to the motion data of each moment according to the similarity indexes corresponding to the motion data of each moment in each category, determining normal data and abnormal data according to the probability indexes and a probability threshold value, and calculating the abnormal degree according to the difference between the normal data and the abnormal data; and carrying out corresponding early warning according to the abnormal degree.
Firstly, for any time, calculating the reciprocal of the similarity index between the motion data of the time and the cluster center of the category, and performing weighted summation on the reciprocal corresponding to each type of motion data to obtain the probability index corresponding to the motion data of the time. Specifically, in this embodiment, the exercise data includes heart rate data, body temperature data, pulse data, and respiration data of the human body during exercise, each data corresponds to a reciprocal of a similarity index, and the four data are weighted and summed to obtain a probability index corresponding to the exercise data at that time. The weights are equal in value, that is, the weight value is 0.25, and an implementer can set the weights according to the attention degrees of different types of data.
Based on this, the probability indexes corresponding to the motion data at each moment are obtained by combining various data, and the larger the value of the probability index is, the smaller the similarity index between the moment and the class clustering center where the moment is located is, the more likely the motion data at the moment is abnormal data. And setting a probability threshold, recording the motion data at the moment corresponding to the probability index larger than the probability threshold as abnormal data, and recording the motion data at the moment corresponding to the probability index smaller than or equal to the probability threshold as normal data. The probability threshold implementer may set the probability threshold according to a specific implementation scenario, for example, the difference between motion data of different people is different, and then different probability thresholds need to be set.
And then, analyzing the abnormal data condition in each category according to the normal data and the abnormal data. For the category corresponding to each clustering center, the motion data at each moment in the category can be divided into two types of motion data, namely normal data and abnormal data, and according to the priori knowledge, the abnormal data usually only occupies a smaller part of the whole data.
Therefore, the abnormal data which are continuous in time are obtained in one category, and the abnormal data which are the same in time interval are obtained to form an abnormal data sequence. For example, if the first abnormal data and the second abnormal data are separated by two time instants, and the second abnormal data and the third abnormal data are also separated by two time instants, the first abnormal data, the second abnormal data and the third abnormal data form an abnormal data sequence.
And acquiring normal data with the same number as the elements in the abnormal data sequence at fixed time intervals to form a normal data sequence, namely acquiring the normal data at the fixed time interval before the first abnormal data in the abnormal data sequence or acquiring the normal data at the fixed time interval after the last abnormal data in the abnormal data sequence. For example, in the present embodiment, the normal data is acquired at the time positions spaced 5 times before the first abnormal data. Or after the third abnormal data, acquiring normal data at 5 time positions, wherein the time interval between the normal data is also two times. Namely, the time interval between each normal data in the normal data sequence is equal to the time interval between each abnormal data in the abnormal data sequence.
And further calculating the abnormal degree corresponding to the category according to the difference between the corresponding position elements in the abnormal data sequence and the normal data sequence, wherein the abnormal degree is expressed by a formula as follows:
Figure 172389DEST_PATH_IMAGE108
wherein, V represents the degree of abnormality,
Figure 468241DEST_PATH_IMAGE110
indicating the total amount of anomalous data contained in the anomalous data sequence,
Figure 390805DEST_PATH_IMAGE112
representing the nth anomalous data in the anomalous data sequence,
Figure 305671DEST_PATH_IMAGE114
indicates the nth normal data in the normal data sequence,
Figure 764334DEST_PATH_IMAGE116
indicating the n +1 th anomalous data in the anomalous data sequence,
Figure 965508DEST_PATH_IMAGE118
represents the n +1 th normal data in the normal data sequence, and max () represents a function of taking the maximum value.
Figure 876833DEST_PATH_IMAGE120
Indicates the difference between the nth abnormal data and the normal data, the larger the difference is, the larger the difference between the data is,
Figure 329811DEST_PATH_IMAGE122
indicating the difference between the adjacent anomaly data,
Figure 410024DEST_PATH_IMAGE124
the difference between the adjacent normal data is represented, the amplitude change condition between the adjacent data is represented, and then the difference degree between the data is obtained by calculating the difference of the amplitude change of the adjacent data, namely the difference is larger, and the value of the difference degree is larger. It can be seen that the degree of difference not only characterizes the difference between the abnormal data and the normal data, but also characterizes the abnormal data and the normal dataThe difference in amplitude variation of adjacent data.
It should be noted that, for different categories, abnormal degrees corresponding to abnormal data sequences may be obtained, and multiple abnormal data sequences may exist in the same category, each abnormal data sequence needs to be calculated identically, and finally, all the abnormal degrees obtained through calculation are added to obtain a comprehensive abnormal degree, which can relatively comprehensively reflect the abnormal condition of the current complete motion data.
Finally, different degrees of early warning are given according to the comprehensive abnormal degree of the motion data, in this embodiment, early warning operations of four color levels are set, specifically, a first threshold, a second threshold, and a third threshold are set, values in this embodiment are 0.8,0.4, and 0.1, respectively, and an implementer can set the early warning operations according to a specific implementation scenario.
When the degree of comprehensive abnormality is greater than a first threshold value, i.e.
Figure 47679DEST_PATH_IMAGE126
When the abnormal degree of the current motion data is over high, the highest level red early warning is sent out; when the degree of comprehensive abnormality is less than or equal to the first threshold value and greater than the second threshold value, that is
Figure 180720DEST_PATH_IMAGE128
When the current movement data are abnormal, the abnormal degree of the current movement data is higher, and an orange early warning is sent out; when the degree of comprehensive abnormality is less than or equal to the second threshold value and greater than the third threshold value, that is
Figure 702968DEST_PATH_IMAGE130
If the abnormal degree of the current motion data is relatively low, a yellow early warning is sent; when the degree of comprehensive abnormality is less than or equal to the third threshold value, i.e.
Figure 605065DEST_PATH_IMAGE132
And when the current motion data are abnormal, the abnormal degree of the current motion data is low, and no early warning is sent out.
According to the intelligent motion monitoring device, the motion data of the human body in the motion process is collected in real time through the intelligent motion monitoring device, the motion is divided into different motion stages according to different characteristics of the motion data in different stages, self-adaptive clustering processing is carried out according to the motion data corresponding to the different motion stages, a better classification result of the motion data is obtained, the motion data in different types are analyzed respectively, the judgment precision of abnormal data is improved, the precision of real-time motion monitoring is improved, the possibility of motion danger is reduced, and the effect of abnormal motion data analysis is better.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present application, and not for limiting the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; the modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present application, and are included in the protection scope of the present application.

Claims (6)

1. A data anomaly analysis method for an intelligent motion monitoring device is characterized by comprising the following steps:
acquiring motion data of a human body at each moment in a complete motion process, and obtaining volatility indexes corresponding to the motion data at each moment according to the corresponding difference of the motion data at adjacent moments; for any moment, the number of the moments when the numerical value of the motion data is the same as the moment is obtained and recorded as the characteristic parameter corresponding to the moment;
respectively calculating a first preferred value, a second preferred value and a third preferred value corresponding to the motion data at the moment according to the numerical value of the motion data and the corresponding moment, fluctuation index and characteristic parameter; respectively calculating a fourth preferred value and a fifth preferred value corresponding to the motion data at the moment according to the moment, the volatility index and the characteristic parameter corresponding to the motion data;
respectively taking the motion data at the time corresponding to the maximum value of the first, second, third, fourth and fifth preference values as a first, second, third, fourth and fifth clustering centers; calculating similarity indexes of the motion data at each moment and each clustering center, and classifying the motion data at each moment according to the similarity indexes to obtain five categories;
obtaining probability indexes corresponding to the motion data of each moment according to the similarity indexes corresponding to the motion data of each moment in each category, determining normal data and abnormal data according to the probability indexes and a probability threshold value, and calculating the abnormal degree according to the difference between the normal data and the abnormal data; carrying out early warning according to the abnormal degree;
the method for acquiring the first preferred value, the second preferred value, the third preferred value, the fourth preferred value and the fifth preferred value specifically comprises the following steps:
recording any moment as a target moment, recording a ratio of the target moment to the time length required for finishing the movement as a first ratio, recording a ratio of the characteristic parameter corresponding to the movement data of the target moment to the time length required for finishing the movement as a second ratio, and obtaining a first preferred value corresponding to the movement data of the target moment according to a difference value between 1 and the first ratio, the second ratio, a numerical value of the movement data and a volatility index, wherein the first preferred value is expressed by a formula:
Figure QLYQS_1
wherein the content of the first and second substances,
Figure QLYQS_2
representing a first preferred value corresponding to the motion data at the tth moment, wherein t is a numerical value of a sequence number at the moment corresponding to the tth moment; t represents the length of time required to complete the movement,
Figure QLYQS_3
representing a first ratio;
Figure QLYQS_4
the fluctuation index corresponding to the motion data at the t-th moment is shown,
Figure QLYQS_5
representing the motion data at the t-th instant,
Figure QLYQS_6
the characteristic parameter corresponding to the t-th moment is shown,
Figure QLYQS_7
represents a second ratio;
according to the first ratio, the second ratio, the value of the motion data and the volatility index, obtaining a second optimal value corresponding to the motion data of the target moment, and expressing the second optimal value by a formula as follows:
Figure QLYQS_8
wherein the content of the first and second substances,
Figure QLYQS_9
a second preferred value corresponding to the motion data of the t-th time;
calculating the absolute value of the difference between half of the time length required by finishing the movement and the target moment, obtaining a third optimal value corresponding to the movement data of the target moment according to the absolute value of the difference, the second ratio, the numerical value of the movement data and the volatility index, and expressing the third optimal value as follows by using a formula:
Figure QLYQS_10
wherein the content of the first and second substances,
Figure QLYQS_11
a third preferred value corresponding to the motion data at the t-th moment is represented, and e is a natural constant;
obtaining a fourth preferred value corresponding to the motion data of the target moment according to the difference between 1 and the first ratio, the difference between 1 and the second ratio and the volatility index of the motion data, and expressing the fourth preferred value as follows by using a formula:
Figure QLYQS_12
wherein, the first and the second end of the pipe are connected with each other,
Figure QLYQS_13
a fourth preferred value representing the motion data corresponding to the t-th moment;
obtaining a fifth preferred value corresponding to the motion data of the target moment according to the first ratio, the difference value between 1 and the second ratio and the volatility index of the motion data, and expressing the fifth preferred value as follows by using a formula:
Figure QLYQS_14
wherein, the first and the second end of the pipe are connected with each other,
Figure QLYQS_15
a fifth preferred value corresponding to the motion data at the t-th instant.
2. The method for analyzing the data abnormality of the intelligent motion monitoring device according to claim 1, wherein the obtaining of the volatility index corresponding to the motion data at each moment according to the difference corresponding to the motion data at adjacent moments specifically comprises:
selecting any moment to be recorded as the current moment, respectively acquiring a fixed number of adjacent moments before and after the current moment, respectively calculating a difference value between the motion data of the adjacent moments and the motion data of the current moment, obtaining a volatility index corresponding to the motion data of the current moment according to the difference value, and further obtaining the volatility index corresponding to the motion data of each moment.
3. The method for analyzing the data abnormality of the intelligent motion monitoring device according to claim 1, wherein the calculating of the similarity index between the motion data at each moment and each cluster center is specifically as follows:
recording any moment as a selected moment, calculating a difference value between the motion data of the selected moment and the motion data of the clustering center, calculating a difference value between the selected moment and the moment corresponding to the clustering center, and respectively calculating similarity indexes of the selected moment and the first clustering center, the second clustering center and the third clustering center according to the two difference values;
obtaining difference values of the motion data of the selected moment and each adjacent moment respectively, recording the maximum value of the difference values as the maximum amplitude value, and recording the minimum value of the difference values as the minimum amplitude value;
acquiring an initial moment and a termination moment of movement, and acquiring a similarity index between the selected moment and a fourth clustering center according to a difference between the selected moment and the initial moment, a difference between a maximum amplitude and a minimum amplitude and a difference between the selected moment and movement data of the fourth clustering center; and obtaining a similarity index between the selected time and the fifth clustering center according to the difference between the selected time and the termination time, the difference between the maximum amplitude and the minimum amplitude and the difference between the selected time and the motion data of the fifth clustering center, and further obtaining the similarity index between the motion data of each time and each clustering center.
4. The method for analyzing the data abnormality of the intelligent exercise monitoring device according to claim 1, wherein the determining of the normal data and the abnormal data according to the probability index and the probability threshold specifically comprises:
and recording the motion data at the moment corresponding to the probability index being greater than the probability threshold as abnormal data, and recording the motion data at the moment corresponding to the probability index being less than or equal to the probability threshold as normal data.
5. The method for analyzing the data abnormality of the intelligent exercise monitoring device according to claim 1, wherein the method for acquiring the degree of abnormality specifically comprises:
in a category corresponding to any one clustering center, acquiring abnormal data with the same time interval to form an abnormal data sequence, acquiring normal data with the same number of elements in the abnormal data sequence to form a normal data sequence according to a fixed time interval, wherein the time interval between the normal data in the normal data sequence is equal to the time interval between the abnormal data in the abnormal data sequence; and calculating the abnormal degree corresponding to the category according to the difference between the corresponding position elements in the abnormal data sequence and the normal data sequence.
6. The data anomaly analysis method for the intelligent motion monitoring device according to claim 1, wherein the early warning according to the anomaly degree specifically comprises the following steps:
acquiring comprehensive abnormal degree of the motion data according to the abnormal degree, and setting a first threshold, a second threshold and a third threshold;
when the comprehensive abnormal degree is larger than a first threshold value, a red early warning is sent out;
when the comprehensive abnormal degree is smaller than or equal to the first threshold and larger than the second threshold, an orange early warning is sent out;
when the comprehensive abnormal degree is smaller than or equal to the second threshold and larger than the third threshold, a yellow early warning is sent out;
and when the comprehensive abnormal degree is less than or equal to the third threshold, not giving out early warning.
CN202211609597.0A 2022-12-15 2022-12-15 Data anomaly analysis method for intelligent motion monitoring device Active CN115590505B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211609597.0A CN115590505B (en) 2022-12-15 2022-12-15 Data anomaly analysis method for intelligent motion monitoring device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211609597.0A CN115590505B (en) 2022-12-15 2022-12-15 Data anomaly analysis method for intelligent motion monitoring device

Publications (2)

Publication Number Publication Date
CN115590505A CN115590505A (en) 2023-01-13
CN115590505B true CN115590505B (en) 2023-03-14

Family

ID=84853932

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211609597.0A Active CN115590505B (en) 2022-12-15 2022-12-15 Data anomaly analysis method for intelligent motion monitoring device

Country Status (1)

Country Link
CN (1) CN115590505B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115861011B (en) * 2023-02-15 2023-05-05 山东优嘉环境科技有限公司 Smart city optimization management method and system based on multi-source data fusion
CN115858794B (en) * 2023-02-20 2023-05-05 北京特立信电子技术股份有限公司 Abnormal log data identification method for network operation safety monitoring
CN116091985B (en) * 2023-04-12 2023-06-13 云钞机器人科技(北京)有限公司 Escort coding method and escort security system
CN116662729B (en) * 2023-08-02 2023-10-31 山东鲁玻玻璃科技有限公司 Low borosilicate glass feeding control data intelligent monitoring method
CN116701886B (en) * 2023-08-07 2023-10-27 知心健(南京)科技有限公司 Intelligent analysis management system for exercise lung test data
CN116820056B (en) * 2023-08-29 2023-11-14 青岛义龙包装机械有限公司 Production process parameter processing method for bag type packaging machine

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112494011A (en) * 2020-11-25 2021-03-16 深圳旭宏医疗科技有限公司 Method and system for monitoring human health index
CN114469041A (en) * 2022-01-30 2022-05-13 北京理工大学 Heart rate change data characteristic analysis method in exercise process
CN114756114A (en) * 2022-04-07 2022-07-15 北京邮电大学 Energy-saving method and equipment for wearable equipment

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3562469B2 (en) * 1995-09-13 2004-09-08 セイコーエプソン株式会社 Health condition management device
US20170188895A1 (en) * 2014-03-12 2017-07-06 Smart Monitor Corp System and method of body motion analytics recognition and alerting
US10842429B2 (en) * 2015-02-26 2020-11-24 Oura Health Oy Method and system for assessing a readiness score of a user
CN109620209B (en) * 2018-12-31 2023-12-19 南京茂森电子技术有限公司 Dynamic electrocardiograph, respiration and motion monitoring system and method
DE102019000608A1 (en) * 2019-01-30 2020-07-30 Drägerwerk AG & Co. KGaA Method, portable device and system for evaluating vector data
WO2021054186A1 (en) * 2019-09-17 2021-03-25 Cyberdyne株式会社 Activity task evaluating system, and activity task evaluating method
TWI779454B (en) * 2021-01-08 2022-10-01 財團法人資訊工業策進會 Motion recognition apparatus and method thereof

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112494011A (en) * 2020-11-25 2021-03-16 深圳旭宏医疗科技有限公司 Method and system for monitoring human health index
CN114469041A (en) * 2022-01-30 2022-05-13 北京理工大学 Heart rate change data characteristic analysis method in exercise process
CN114756114A (en) * 2022-04-07 2022-07-15 北京邮电大学 Energy-saving method and equipment for wearable equipment

Also Published As

Publication number Publication date
CN115590505A (en) 2023-01-13

Similar Documents

Publication Publication Date Title
CN115590505B (en) Data anomaly analysis method for intelligent motion monitoring device
WO2010083562A1 (en) Activity detection
CN112577747A (en) Rolling bearing fault diagnosis method based on space pooling network
CN115670418B (en) Data storage method based on smart watch user information
CN108090508A (en) A kind of classification based training method, apparatus and storage medium
CN117241306B (en) Real-time monitoring method for abnormal flow data of 4G network
CN111368892B (en) Electric energy quality disturbance efficient identification method for generalized S transformation and SVM
CN115982557B (en) Data processing system for gas concentration detection
CN111191722B (en) Method and device for training prediction model through computer
CN112765550A (en) Target behavior segmentation method based on Wi-Fi channel state information
CN116821809A (en) Vital sign data acquisition system based on artificial intelligence
CN109901064B (en) ICA-LVQ-based high-voltage circuit breaker fault diagnosis method
Jin et al. Human activity recognition machine with an anchor-based loss function
CN108038586B (en) Water environment evaluation method and system for river crab culture
Jia et al. Deep learning with convolutional neural networks for sleep arousal detection
CN116662925A (en) Industrial process soft measurement method based on weighted sparse neural network
CN115410267A (en) Statistical algorithm based on interaction action analysis data of human skeleton and muscle
CN114601455A (en) Motion recognition method based on two-stage neural network
CN114680879A (en) Method and device for detecting body motion artifact signal
CN117633480B (en) Quantitative evaluation method and system for exercise training effect
CN112529637A (en) Service demand dynamic prediction method and system based on context awareness
CN115222945B (en) Deep semantic segmentation network training method based on multi-scale self-adaptive course learning
CN109985369A (en) A kind of adaptive stroke recognition methods based on intelligent wrist-worn device
Li et al. Research on Wordle based on time series ARIMA model and XGBoost model
CN115995282B (en) Expiratory flow data processing system based on knowledge graph

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
CP02 Change in the address of a patent holder
CP02 Change in the address of a patent holder

Address after: Room 606, Floor 6, Dongmingjia East Second Ring Store, No. 39, East Second Ring South Road, Yuhua District, Shijiazhuang City, Hebei Province, 050000

Patentee after: Nannan Juzhi Information Technology Co.,Ltd.

Address before: Room 319, 3rd floor, Block C, No. 100 Yuhua East Road, Yuhua District, Shijiazhuang, Hebei Province 050000

Patentee before: Nannan Juzhi Information Technology Co.,Ltd.