CN106021877B - A kind of detection method of rule of life exception - Google Patents
A kind of detection method of rule of life exception Download PDFInfo
- Publication number
- CN106021877B CN106021877B CN201610310530.5A CN201610310530A CN106021877B CN 106021877 B CN106021877 B CN 106021877B CN 201610310530 A CN201610310530 A CN 201610310530A CN 106021877 B CN106021877 B CN 106021877B
- Authority
- CN
- China
- Prior art keywords
- detector
- data
- detectors
- value
- rule
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G06F19/3418—
Landscapes
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Alarm Systems (AREA)
Abstract
The present invention relates to a kind of detection methods of rule of life exception, this method acquires normal training data, and according to training data obtain various types data value range, then by server randomly generate it is all kinds of be located at value range in data to form random vector.Then pass through the Euclidean distance of calculating training data and random vector again, and obtain the maximum and minimum value of Euclidean distance, and then detector is generated, recycle the new data of detector detection acquisition whether abnormal, and then judge whether the rule of life for monitoring people exception occurs.The detection method of rule of life exception can more macroscopical, more fully monitor the rule of life abnormal conditions of user, be especially suitable for the strong the elderly of rule of life and use.The detection method of rule of life exception both improves the real-time effectiveness of monitoring, also reduces equipment cost without using Video Supervision Technique.
Description
Technical field
The present invention relates to a kind of computer aided system technical field, more particularly to one kind are abnormal for old man's rule of life
The detection method of situation.
Background technology
China Today enters aging population society, with the process of social senilization, to the treatment problem of old man
Through as social concern urgently to be resolved hurrily.Since life of elderly person is more regular, when deviateing rule of life, then often there is body
Body is uncomfortable or other fortuitous events.At this time it is necessary to remind its relatives to give necessary care, avoid being worse off.
Existing various the elderlys show loving care in system scheme, have the scheme based on video monitoring or wireless sense network, such as
Authorization Notice No. is the Chinese utility model patent of CN204990604 (application No. is 201520616749.9)《Household safety-protection and
Family endowment monitoring and alarming system》, disclosed in technical solution i.e. used video monitoring equipment, health data monitoring set
Standby, security monitor equipment, intelligent gateway and client terminal etc., the system can alarm automatically when found the abnormal situation.
But there is the features such as layout is cumbersome, of high cost in such scheme, it is difficult to which universal use.The also side based on wearable device
Case, such as the Chinese utility model patent that Authorization Notice No. is CN205031234 (application No. is 201520692943.5)《A kind of use
In the body temperature measurable Intelligent heart rate bracelet of old man》, disclosed in be built-in with pulse detection sensor, temperature sensing in bracelet
Device can alarm to the abnormal conditions of old man's heart rate and body temperature.Due to the limit of installation sensor in such scheme
System, function is relatively single, not ideal enough to the comprehensive monitoring effect of abnormal conditions.
Invention content
The technical problem to be solved by the present invention is to provide one kind for the above-mentioned prior art can comprehensively and accurately realize
To the monitoring of old man's abnormal conditions and the detection method of rule of life exception at low cost.
Technical solution is used by the present invention solves the above problems:A kind of detection method of rule of life exception, it is special
Sign is to include the following steps:
(1), it is trained data acquisition, i.e. acquisition time data, while acquiring the position data of monitoring people, Activity Type
Grouped data, kinematic parameter, physical sign parameters, so formed training data group TR=[T, L, C, S, B], then to training data into
Line flag is for using;
Wherein, T is time data, T=[T1,T2,T3,…,Ti], wherein i is natural number;
L is the position data for monitoring people, L=[L1,L2,L3,…,Li], wherein i is natural number;
C is Activity Type grouped data, C=[C1,C2,C3,…,Ci], wherein i is natural number;
S is kinematic parameter, S=[S1,S2,S3,…,Si], wherein i is natural number;
B is physical sign parameters, B=[B1,B2,B3,…,Bi], wherein i is natural number;
Then each training vector in training data group can be expressed as TRa=[Ta,La,Ca,Sa,Ba], wherein a is nature
Number, 1≤a≤i;
(2), training data group is sent in server and is stored in the database;
(3), the maximum value and minimum value for calculating each dimension data in training data group, to form the range of each dimension data
Data group TL={ TE, LE, CE, SE, BE };
Wherein, TE is the value range of time data, TE=(MAX [T1,T2,T3,…,Ti],MIN[T1,T2,T3,…,
Ti]);
LE is the value range for detecting the position data of people, LE=(MAX [L1,L2,L3,…,Li],MIN[L1,L2,
L3,…, Li]);
CE is the value range of Activity Type grouped data, CE=(MAX [C1,C2,C3,…,Ci],MIN[C1,C2,
C3,…,Ci]);
SE is the value range of kinematic parameter, SE=(MAX [S1,S2,S3,…,Si],MIN[S1,S2,S3,…,Si]);
BE is the value range of physical sign parameters, BE=(MAX [B1,B2,B3,…,Bi],MIN[B1,B2,B3,…,Bi]);
(4), the new data of server receiving device acquisition, and then constitute the new state data group N=[T of monitoring peopleN,LN,
CN,SN,BN], wherein TNFor freshly harvested time data, LNFor freshly harvested position data, CNFor freshly harvested Activity Type
Grouped data, SNFor freshly harvested kinematic parameter, BNFreshly harvested physical sign parameters;
(5), server automatically generates a random vector R=[X1,X2,X3,X4,X5], wherein X1∈ TE, X2∈ LE, X3
∈ CE, X4∈ SE, X5∈BE;
(6), each training vector TR in training data group TR is calculatedaWith the Euclidean distance between random vector R, and then obtain
First Euclidean distance array EDR=[EDR1,EDR 2,EDR 3,…,EDRi], it obtains in the first Euclidean distance array EDR most
Big distance value selfmax and lowest distance value selfmin;
(7), a detector D is generated, the corresponding data packet D={ R, selfmax, selfmin, LA } of the detector,
Detector D is added in detectors set Detectors;
Wherein, LA indicates the corresponding Lifetime values of detector D, and the initial value of LA is 2* unit incrementss LA0;
(8), judge whether there is j detector of specified quantity in detectors set Detectors;
If it is not, then cycle carries out step (4) to step (7), until generating j detector of specified quantity, detector collection
Close Detectors=[D1,D2,D3,…,Db,…,Dj], wherein b, j are natural number, 1≤b≤j;
Db={ Rb,selfmaxb,selfminb,LAb};
RbIndicate detector DbData packet in corresponding random vector;
SelfmaxbIndicate random vector RbThe maximum value of Euclidean distance between each training vector;
SelfminbIndicate random vector RbThe minimum value of Euclidean distance between each training vector;
LAbIndicate detector DbCorresponding Lifetime values;
If so, thening follow the steps (9);
(9), a detector D is selected in detectors set Detectorsb, 1≤b≤j calculates most freshly harvested new
Status data group N=[TN,LN,CN,SN,BN] and detector DbMiddle random vector RbBetween Euclidean distance EDb, it is respectively compared EDb
With detector DbMiddle SelfmaxbSize and EDbWith detector DbMiddle SelfminbSize;
(10) if, EDb> SelfmaxbOr EDb< Selfminb, then it represents that detector DbDetect the life of monitoring people
There is exception in rule living, and corresponding server sends out alarm, and return to step (4), while LA to clientbValue be updated to LAb
+LA0;
If Selfminb≤EDb≤Selfmaxb, then it represents that detector DbThe result of detection is to monitor the rule of life of people
In normal range (NR), it is then back to step (9) and carries out cycle detection;
(11) if, detector all in detectors set Detectors cannot all detect the life rule of monitoring people
Abnormal situation is restrained, then the corresponding Lifetime values of detector all in detectors set Detectors are subtracted into LA0, this
When, the detector that Lifetime values are 0 is then rejected in detectors set Detectors;
(12), return to step (4).
For optimizing detection as a result, further including the closeness value J of detector in the data packet of the detector, that is, detect
The initial value of device data packet D={ R, selfmax, selfmin, LA, J }, J are the integer value J of setting0;Correspondingly, detector collection
Close D in Detectorsb={ Rb,selfmaxb,selfminb,LAb,Jb, JbIndicate detector DbCorresponding closeness value;
Either detector Db, closeness value Jb=MIN [selfmaxb-EDb,EDb-selfminb], 1≤b≤j;
Establish parent advantage detectors set Parents=[D1,D2,D3,…,Dm,…,Dn], m, n are natural number, 1≤m
The initial sets of≤n, parent advantage detectors set Parents are the detectors set Detectors being initially formed;
In step (11), when detector all in detectors set Detectors cannot all detect monitoring people's
When the situation of rule of life exception, also increase the detection method for being provided with optimization, the detection method of the optimization includes following step
Suddenly:
(11.1), a detector D of closeness value minimum is taken from parent advantage detectors set Parentsm, to inspection
Survey device DmIn random vector Rm=[X1m,X2m,X3m,X4m,X5m] in random one-dimensional value be changed with obtain it is new it is random to
Measure RNm=[XN1m,XN2m,XN3m,XN4m,XN5m], obtain an interim detector C D={ RNm, selfmaxm,selfminm,
LAm,Jm};
Wherein X1m∈ TE, X2m∈ LE, X3m∈ CE, X4m∈ SE, X5m∈BE;
XN1m∈ TE, XN2m∈ LE, XN3m∈ CE, XN4m∈ SE, XN5m∈BE;
Cycle carries out the step and generates j temporary detecting device to breed, which is stored in temporary detecting
In device set Childen;
(11.2), a temporary detecting device CD is taken out from temporary detecting device set Childenc, CDcIn it is corresponding random
Vector is RNmc, wherein c is natural number, 1≤c≤j;
(11.3), each training vector TR in training data group TR is calculatedaWith random vector RNmcBetween Euclidean distance, into
And obtain the second Euclidean distance array EDC=[EDC1,EDC 2,EDC 3,…,EDCi], it obtains in the second Euclidean distance array DR
Maximum range value c-selfmax and lowest distance value c-selfmin;
(11.4), one is generated newly for detector ND, the corresponding data packet ND={ RN of the new generation detectormc,c-
Selfmax, c-selfmin, NLA, NJ }, wherein NLA indicates that NLA's is initial newly for detector ND corresponding Lifetime values
Value is 2* unit incrementss LA0, newly for the closeness value of detector, the initial value of NJ is the integer value J of setting for NJ expressions0;
It will be newly added to newly for detectors set NewDetectors for detector ND;
(11.5), circulation step (11.2) to (11.4) is until go through all over temporary detectings all detectors set Childen
Device, then NewDetectors=[D1,D2,D3,…,Dd,…,Dj], wherein d is natural number, 1≤d≤j;
Dd={ RNmcd,c-selfmaxd,c-selfmin d,NLAd,NJd};
RNmcdIt indicates newly for detector DdData packet in corresponding random vector;
c-selfmaxdIndicate random vector RNmcdThe maximum value of Euclidean distance between each training vector;
c-selfmin dIndicate random vector RNmcdThe minimum value of Euclidean distance between each training vector;
NLAdIt indicates newly for detector DdCorresponding Lifetime values;
NJdIt indicates newly for detector DdCloseness value;
(11.6), from newly for taken in detectors set NewDetectors one newly for detector NDd, wherein d is nature
Number, 1≤d≤j calculate most freshly harvested new state data group N=[TN,LN,CN,SN,BN] and newly for detector NDdIn at random to
Measure RNmcdBetween Euclidean distance NEDd, it is respectively compared NEDdWith detector NDdMiddle c-selfmaxdSize and NEDdWith inspection
Survey device DbMiddle c-selfmindSize;
(11.7) if, NEDd> c-selfmaxdOr NEDd< c-selfmind, then it represents that newly for detector NDdInspection
There is exception in the rule of life for measuring monitoring people, NDdDetectors set Detectors is added, corresponding server is to client
End sends out alarm, and executes step (12), while NLAdValue be updated to NLAd+LA0;
If c-selfmind≤NEDd≤c-selfmaxd, then it represents that newly for detector NDdThe result of detection is monitoring people
Rule of life be in normal range (NR), be then back to step (11.6) carry out cycle detection;
It calculates simultaneously newly for detector NDdCorresponding closeness value NJd, NJd=MIN [c-selfmaxd-NEDd,NEDd
-c-selfmin d];
(11.8) if, newly for detector all in detectors set NewDetectors all cannot detect monitoring people
Rule of life exception situation, then from newly for detectors set NewDetectors and parent advantage detectors set
The preceding j detector that closeness value minimum is taken out in all detectors of Parents, as new parent advantage detector
Set Parents is emptied newly for detector all in detectors set NewDetectors;
Return to step (11.1), cycle detection Count times, Count are the setting cycle detection time of the detection method of optimization
Number, if be still not detected monitoring people rule of life abnormal conditions, can recognize that for monitor people rule of life it is normal, so
Step (12) is executed afterwards.
Preferably, the time data include the date, when, divided data.
Preferably, the position data includes indoor location data and outdoor location data;
By position data in indoor WiFi localization methods collection room, outdoor location data are acquired by GPS positioning method.
Preferably, the kinematic parameter includes direction of motion data and angular movement speed data, the Activity Type classification number
According to including grouped data that is static, walking, run, drive.
Preferably, the physical sign parameters include blood oxygen saturation data, pulse data, temperature data.
Compared with the prior art, the advantages of the present invention are as follows:The detection method of rule of life exception can be more macroscopical, more
The rule of life abnormal conditions of comprehensive monitoring user are especially suitable for the strong the elderly of rule of life and use.The life is advised
Abnormal detection method is restrained without using Video Supervision Technique, both improves the real-time effectiveness of monitoring, also reduce equipment at
This.
Description of the drawings
Fig. 1 is the flow chart of the detection method of rule of life exception in the embodiment of the present invention.
Specific implementation mode
Below in conjunction with attached drawing embodiment, present invention is further described in detail.
As shown in Figure 1, the detection method of the rule of life exception in the present embodiment, includes the following steps:
(1), it is trained data acquisition, i.e. acquisition time data, while acquiring the position data of monitoring people, Activity Type
Grouped data, kinematic parameter, physical sign parameters, so formed training data group TR=[T, L, C, S, B], then to training data into
Line flag is for using;
Wherein, T is time data, T=[T1,T2,T3,…,Ti], wherein i is natural number;
L is the position data for monitoring people, L=[L1,L2,L3,…,Li], wherein i is natural number;
C is Activity Type grouped data, C=[C1,C2,C3,…,Ci], wherein i is natural number;
S is kinematic parameter, S=[S1,S2,S3,…,Si], wherein i is natural number;
B is physical sign parameters, B=[B1,B2,B3,…,Bi], wherein i is natural number;
Then each training vector in training data group can be expressed as TRa=[Ta,La,Ca,Sa,Ba], wherein a is nature
Number, 1≤a≤i.
The time data acquired in the present embodiment include the date, when, divided data.
The position data acquired in the present embodiment includes indoor location data and outdoor location data, wherein indoor location number
It is acquired according to by indoor WiFi localization methods, outdoor location data are then acquired by GPS positioning method.
Kinematic parameter in the present embodiment includes direction of motion data and angular movement speed data, Activity Type grouped data packet
Include grouped data that is static, walking, run, driving.
Physical sign parameters in the present embodiment include blood oxygen saturation data, pulse data, temperature data.
(2), training data group is sent in server and is stored in the database.
(3), the maximum value and minimum value for calculating each dimension data in training data group, to form the range of each dimension data
Data group TL={ TE, LE, CE, SE, BE };
Wherein, TE is the value range of time data, TE=(MAX [T1,T2,T3,…,Ti],MIN[T1,T2,T3,…,
Ti]);
LE is the value range for detecting the position data of people, LE=(MAX [L1,L2,L3,…,Li],MIN[L1,L2,
L3,…, Li]);
CE is the value range of Activity Type grouped data, CE=(MAX [C1,C2,C3,…,Ci],MIN[C1,C2,
C3,…,Ci]);
SE is the value range of kinematic parameter, SE=(MAX [S1,S2,S3,…,Si],MIN[S1,S2,S3,…,Si]);
BE is the value range of physical sign parameters, BE=(MAX [B1,B2,B3,…,Bi],MIN[B1,B2,B3,…,Bi])。
(4), the new data of server receiving device acquisition, and then constitute the new state data group N=[T of monitoring peopleN,LN,
CN,SN,BN], wherein TNFor freshly harvested time data, LNFor freshly harvested position data, CNFor freshly harvested Activity Type
Grouped data, SNFor freshly harvested kinematic parameter, BNFreshly harvested physical sign parameters.
(5), server automatically generates a random vector R=[X1,X2,X3,X4,X5], wherein X1∈ TE, X2∈ LE, X3
∈ CE, X4∈ SE, X5∈BE。
(6), each training vector TR in training data group TR is calculatedaWith the Euclidean distance between random vector R, and then obtain
First Euclidean distance array EDR=[EDR1,EDR 2,EDR 3,…,EDRi], it obtains in the first Euclidean distance array EDR most
Big distance value selfmax and lowest distance value selfmin.
(7), a detector D is generated, the detector corresponding data packet D=R, selfmax, selfmin, LA,
J }, detector D is added in detectors set Detectors.
Wherein, LA indicates the corresponding Lifetime values of detector D, and the initial value of LA is 2* unit incrementss LA0;J is indicated
The D closeness values of detector, the initial value of J are the integer value J of setting0;
(8), judge whether there is j detector of specified quantity in detectors set Detectors;
If it is not, then cycle carries out step (4) to step (7), until generating j detector of specified quantity, detector collection
Close Detectors=[D1,D2,D3,…,Db,…,Dj], wherein b, j are natural number, 1≤b≤j;
Db={ Rb,selfmaxb,selfminb,LAb,Jb};
RbIndicate detector DbData packet in corresponding random vector;
SelfmaxbIndicate random vector RbThe maximum value of Euclidean distance between each training vector;
SelfminbIndicate random vector RbThe minimum value of Euclidean distance between each training vector;
LAbIndicate detector DbCorresponding Lifetime values;
JbIndicate detector DbCorresponding closeness value;
Either detector DbCloseness value Jb=MIN [selfmaxb-EDb,EDb-selfminb], 1≤b≤j;
If so, thening follow the steps (9);
Establish parent advantage detectors set Parents=[D1,D2,D3,…,Dm,…,Dn], m, n are natural number, 1≤m
The initial sets of≤n, parent advantage detectors set Parents are the detectors set Detectors being initially formed.
(9), a detector D is selected in detectors set Detectorsb, 1≤b≤j calculates most freshly harvested new
Status data group N=[TN,LN,CN,SN,BN] and detector DbMiddle random vector RbBetween Euclidean distance EDb, it is respectively compared EDb
With detector DbMiddle SelfmaxbSize and EDbWith detector DbMiddle SelfminbSize;
(10) if, EDb> SelfmaxbOr EDb< Selfminb, then it represents that detector DbDetect the life of monitoring people
There is exception in rule living, and corresponding server sends out alarm, and return to step (4), while LA to clientbValue be updated to LAb
+LA0;
If Selfminb≤EDb≤Selfmaxb, then it represents that detector DbThe result of detection is to monitor the rule of life of people
In normal range (NR), it is then back to step (9) and carries out cycle detection.
(11) if, detector all in detectors set Detectors cannot all detect the life rule of monitoring people
Abnormal situation is restrained, then the corresponding Lifetime values of detector all in detectors set Detectors are subtracted into LA0, this
When, the detector that Lifetime values are 0 is then rejected in detectors set Detectors;
When detector all in detectors set Detectors cannot all detect the rule of life exception of monitoring people
When situation, also increase the detection method for being provided with optimization, the detection method of the optimization includes the following steps:
(11.1), a detector D of closeness value minimum is taken from parent advantage detectors set Parentsm, to inspection
Survey device DmIn random vector Rm=[X1m,X2m,X3m,X4m,X5m] in random one-dimensional value be changed with obtain it is new it is random to
Measure RNm=[XN1m,XN2m,XN3m,XN4m,XN5m], obtain an interim detector C D={ RNm, selfmaxm,selfminm,
LAm,Jm};
Wherein X1m∈ TE, X2m∈ LE, X3m∈ CE, X4m∈ SE, X5m∈BE;
XN1m∈ TE, XN2m∈ LE, XN3m∈ CE, XN4m∈ SE, XN5m∈BE;
Cycle carries out the step and generates j temporary detecting device to breed, which is stored in temporary detecting
In device set Childen;
(11.2), a temporary detecting device CD is taken out from temporary detecting device set Childenc, CDcIn it is corresponding random
Vector is RNmc, wherein c is natural number, 1≤c≤j;
(11.3), each training vector TR in training data group TR is calculatedaWith random vector RNmcBetween Euclidean distance, into
And obtain the second Euclidean distance array EDC=[EDC1,EDC 2,EDC 3,…,EDCi], it obtains in the second Euclidean distance array DR
Maximum range value c-selfmax and lowest distance value c-selfmin;
(11.4), one is generated newly for detector ND, the corresponding data packet ND={ RN of the new generation detectormc,c-
Selfmax, c-selfmin, NLA, NJ }, wherein NLA indicates that NLA's is initial newly for detector ND corresponding Lifetime values
Value is 2* unit incrementss LA0, newly for the closeness value of detector, the initial value of NJ is the integer value J of setting for NJ expressions0;
It will be newly added to newly for detectors set NewDetectors for detector ND;
(11.5), circulation step (11.2) to (11.4) is until go through all over temporary detectings all detectors set Childen
Device, then NewDetectors=[D1,D2,D3,…,Dd,…,Dj], wherein d is natural number, 1≤d≤j;
Dd={ RNmcd,c-selfmaxd,c-selfmin d,NLAd,NJd};
RNmcdIt indicates newly for detector DdData packet in corresponding random vector;
c-selfmaxdIndicate random vector RNmcdThe maximum value of Euclidean distance between each training vector;
c-selfmin dIndicate random vector RNmcdThe minimum value of Euclidean distance between each training vector;
NLAdIt indicates newly for detector DdCorresponding Lifetime values;
NJdIt indicates newly for detector DdCloseness value;
(11.6), from newly for taken in detectors set NewDetectors one newly for detector NDd, wherein d is nature
Number, 1≤d≤j calculate most freshly harvested new state data group N=[TN,LN,CN,SN,BN] and newly for detector NDdIn at random to
Measure RNmcdBetween Euclidean distance NEDd, it is respectively compared NEDdWith detector NDdMiddle c-selfmaxdSize and NEDdWith inspection
Survey device DbMiddle c-selfmindSize;
(11.7) if, NEDd> c-selfmaxdOr NEDd< c-selfmind, then it represents that newly for detector NDdInspection
There is exception in the rule of life for measuring monitoring people, and corresponding server sends out alarm to client, NDdDetectors set is added
Detectors, and step (12) is executed, while NLAdValue be updated to NLAd+LA0;
If c-selfmind≤NEDd≤c-selfmaxd, then it represents that newly for detector NDdThe result of detection is monitoring people
Rule of life be in normal range (NR), be then back to step (11.6) carry out cycle detection;
It calculates simultaneously newly for detector NDdCorresponding closeness value NJd, NJd=MIN [c-selfmaxd-NEDd,NEDd
-c-selfmin d];
(11.8) if, newly for detector all in detectors set NewDetectors all cannot detect monitoring people
Rule of life exception situation, then from newly for detectors set NewDetectors and parent advantage detectors set
The preceding j detector that closeness value minimum is taken out in all detectors of Parents, as new parent advantage detector
Set Parents is emptied newly for detector all in detectors set NewDetectors;
Return to step (11.1), cycle detection Count times, Count are the setting cycle detection time of the detection method of optimization
Number, if be still not detected monitoring people rule of life abnormal conditions, can recognize that for monitor people rule of life it is normal, so
Step (12) is executed afterwards.
(12), return to step (4).
Claims (6)
1. a kind of detection method of rule of life exception, it is characterised in that include the following steps:
(1), it is trained data acquisition, i.e. acquisition time data, while acquiring the position data of monitoring people, Activity Type classification
Data, kinematic parameter, physical sign parameters, and then training data group TR=[T, L, C, S, B] is formed, then to training data into rower
Remember for using;
Wherein, T is time data, T=[T1,T2,T3,…,Ti], wherein i is natural number;
L is the position data for monitoring people, L=[L1,L2,L3,…,Li], wherein i is natural number;
C is Activity Type grouped data, C=[C1,C2,C3,…,Ci], wherein i is natural number;
S is kinematic parameter, S=[S1,S2,S3,…,Si], wherein i is natural number;
B is physical sign parameters, B=[B1,B2,B3,…,Bi], wherein i is natural number;
Then each training vector in training data group can be expressed as TRa=[Ta,La,Ca,Sa,Ba], wherein a be natural number, 1≤
a≤i;
(2), training data group is sent in server and is stored in the database;
(3), the maximum value and minimum value for calculating each dimension data in training data group, to form the range data of each dimension data
Group TL={ TE, LE, CE, SE, BE };
Wherein, TE is the value range of time data, TE=(MAX [T1,T2,T3,…,Ti],MIN[T1,T2,T3,…,Ti]);
LE is the value range for detecting the position data of people, LE=(MAX [L1,L2,L3,…,Li],MIN[L1,L2,L3,…,
Li]);
CE is the value range of Activity Type grouped data, CE=(MAX [C1,C2,C3,…,Ci],MIN[C1,C2,C3,…,
Ci]);
SE is the value range of kinematic parameter, SE=(MAX [S1,S2,S3,…,Si],MIN[S1,S2,S3,…,Si]);
BE is the value range of physical sign parameters, BE=(MAX [B1,B2,B3,…,Bi],MIN[B1,B2,B3,…,Bi]);
(4), the new data of server receiving device acquisition, and then constitute the new state data group N=[T of monitoring peopleN,LN,CN,SN,
BN], wherein TNFor freshly harvested time data, LNFor freshly harvested position data, CNFor the classification number of freshly harvested Activity Type
According to SNFor freshly harvested kinematic parameter, BNFreshly harvested physical sign parameters;
(5), server automatically generates a random vector R=[X1,X2,X3,X4,X5], wherein X1∈ TE, X2∈ LE, X3∈ CE,
X4∈ SE, X5∈BE;
(6), each training vector TR in training data group TR is calculatedaWith the Euclidean distance between random vector R, and then obtain first
Euclidean distance array EDR=[EDR1,EDR2,EDR3,…,EDRi], obtain the maximum distance in the first Euclidean distance array EDR
Value selfmax and lowest distance value selfmin;
(7), a detector D is generated, the corresponding data packet D={ R, selfmax, selfmin, LA } of the detector will be examined
Device D is surveyed to be added in detectors set Detectors;
Wherein, LA indicates the corresponding Lifetime values of detector D, and the initial value of LA is 2* unit incrementss LA0;
(8), judge whether there is j detector of specified quantity in detectors set Detectors;
If it is not, then cycle carries out step (4) to step (7), until generating j detector of specified quantity, detectors set
Detectors=[D1,D2,D3,…,Db,…,Dj], wherein b, j are natural number, 1≤b≤j;
Db={ Rb,selfmaxb,selfminb,LAb};
RbIndicate detector DbData packet in corresponding random vector;
SelfmaxbIndicate random vector RbThe maximum value of Euclidean distance between each training vector;
SelfminbIndicate random vector RbThe minimum value of Euclidean distance between each training vector;
LAbIndicate detector DbCorresponding Lifetime values;
If so, thening follow the steps (9);
(9), a detector D is selected in detectors set Detectorsb, 1≤b≤j, the most freshly harvested new state number of calculating
According to a group N=[TN,LN,CN,SN,BN] and detector DbMiddle random vector RbBetween Euclidean distance EDb, it is respectively compared EDbWith detection
Device DbMiddle SelfmaxbSize and EDbWith detector DbMiddle SelfminbSize;
(10) if, EDb> SelfmaxbOr EDb< Selfminb, then it represents that detector DbDetect the life rule of monitoring people
There is exception in rule, and corresponding server sends out alarm, and return to step (4), while LA to clientbValue be updated to LAb+
LA0;
If Selfminb≤EDb≤Selfmaxb, then it represents that detector DbThe result of detection is that the rule of life of monitoring people is in
Normal range (NR) is then back to step (9) and carries out cycle detection;
(11) if, detector all in detectors set Detectors cannot all detect that the rule of life of monitoring people is different
The corresponding Lifetime values of detector all in detectors set Detectors are then subtracted LA by normal situation0, at this point, raw
The detector that life periodic quantity is 0 is then rejected in detectors set Detectors;
(12), return to step (4).
2. the detection method of rule of life exception according to claim 1, it is characterised in that:The data packet of the detector
In further include detector closeness value J, i.e. detector data packet D={ R, selfmax, selfmin, LA, J }, J's is initial
Value is the integer value J of setting0;Correspondingly, D in detectors set Detectorsb={ Rb,selfmaxb,selfminb,LAb,
Jb, JbIndicate detector DbCorresponding closeness value;
Either detector DbCloseness value Jb=MIN [selfmaxb-EDb,EDb-selfminb], 1≤b≤j;
Establish parent advantage detectors set Parents=[D1,D2,D3,…,Dm,…,Dn], m, n be natural number, 1≤m≤n,
The initial sets of parent advantage detectors set Parents are the detectors set Detectors being initially formed;
In step (11), when detector all in detectors set Detectors cannot all detect the life of monitoring people
When the situation of rule exception, also increase the detection method for being provided with optimization, the detection method of the optimization includes the following steps:
(11.1), a detector D of closeness value minimum is taken from parent advantage detectors set Parentsm, to detector
DmIn random vector Rm=[X1m,X2m,X3m,X4m,X5m] in random one-dimensional value be changed to obtain new random vector
RNm=[XN1m,XN2m,XN3m,XN4m,XN5m], obtain an interim detector C D={ RNm,selfmaxm,selfminm,
LAm,Jm};
Wherein X1m∈ TE, X2m∈ LE, X3m∈ CE, X4m∈ SE, X5m∈BE;
XN1m∈ TE, XN2m∈ LE, XN3m∈ CE, XN4m∈ SE, XN5m∈BE;
Cycle carries out the step and generates j temporary detecting device to breed, which is stored in temporary detecting device collection
It closes in Childen;
(11.2), a temporary detecting device CD is taken out from temporary detecting device set Childenc, CDcIn corresponding random vector
For RNmc, wherein c is natural number, 1≤c≤j;
(11.3), each training vector TR in training data group TR is calculatedaWith random vector RNmcBetween Euclidean distance, and then obtain
Take the second Euclidean distance array EDC=[EDC1,EDC2,EDC3,…,EDCi], obtain the maximum in the second Euclidean distance array DR
Distance value c-selfmax and lowest distance value c-selfmin;
(11.4), one is generated newly for detector ND, the corresponding data packet ND={ RN of the new generation detectormc,c-selfmax,c-
Selfmin, NLA, NJ }, wherein NLA indicates that, newly for the corresponding Lifetime values of detector ND, the initial value of NLA increases for 2* units
Dosage LA0, newly for the closeness value of detector, the initial value of NJ is the integer value J of setting for NJ expressions0;
It will be newly added to newly for detectors set NewDetectors for detector ND;
(11.5), circulation step (11.2) to (11.4) is until go through all over temporary detecting devices all detectors set Childen, then
NewDetectors=[D1,D2,D3,…,Dd,…,Dj], wherein d is natural number, 1≤d≤j;
Dd={ RNmcd,c-selfmaxd,c-selfmind,NLAd,NJd};
RNmcdIt indicates newly for detector DdData packet in corresponding random vector;
c-selfmaxdIndicate random vector RNmcdThe maximum value of Euclidean distance between each training vector;
c-selfmindIndicate random vector RNmcdThe minimum value of Euclidean distance between each training vector;
NLAdIt indicates newly for detector DdCorresponding Lifetime values;
NJdIt indicates newly for detector DdCloseness value;
(11.6), from newly for taken in detectors set NewDetectors one newly for detector NDd, wherein d be natural number, 1≤
D≤j calculates most freshly harvested new state data group N=[TN,LN,CN,SN,BN] and newly for detector NDdMiddle random vector RNmcd
Between Euclidean distance NEDd, it is respectively compared NEDdWith detector NDdMiddle c-selfmaxdSize and NEDdWith detector Db
Middle c-selfmindSize;
(11.7) if, NEDd> c-selfmaxdOr NEDd< c-selfmind, then it represents that newly for detector NDdIt detects
There is exception in the rule of life for monitoring people, and corresponding server sends out alarm to client, NDdDetectors set is added
Detectors, and step (12) is executed, while NLAdValue be updated to NLAd+LA0;
If c-selfmind≤NEDd≤c-selfmaxd, then it represents that newly for detector NDdThe result of detection is to monitor the life of people
Rule living is in normal range (NR), is then back to step (11.6) and carries out cycle detection;
It calculates simultaneously newly for detector NDdCorresponding closeness value NJd, NJd=MIN [c-selfmaxd-NEDd,NEDd-c-
selfmind];
(11.8) if, newly for detector all in detectors set NewDetectors all cannot detect monitoring people life
The situation of rule exception living, then from newly for detectors set NewDetectors's and parent advantage detectors set Parents
The preceding j detector that closeness value minimum is taken out in all detectors, as new parent advantage detectors set
Parents is emptied newly for detector all in detectors set NewDetectors;
Return to step (11.1), cycle detection Count times, Count are the setting cycle detection number of the detection method of optimization, such as
Fruit be still not detected monitoring people rule of life abnormal conditions, then can recognize that for monitor people rule of life it is normal, then hold
Row step (12).
3. the detection method of rule of life exception according to claim 1 or 2, it is characterised in that:The time data packet
Include the date, when, divided data.
4. the detection method of rule of life exception according to claim 1 or 2, it is characterised in that:The position data packet
Include indoor location data and outdoor location data;
By position data in indoor WiFi localization methods collection room, outdoor location data are acquired by GPS positioning method.
5. the detection method of rule of life exception according to claim 1 or 2, it is characterised in that:The kinematic parameter packet
Include direction of motion data and angular movement speed data, the Activity Type grouped data includes static, the classification number walking, run, driving
According to.
6. the detection method of rule of life exception according to claim 1 or 2, it is characterised in that:The physical sign parameters packet
Include blood oxygen saturation data, pulse data, temperature data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610310530.5A CN106021877B (en) | 2016-05-11 | 2016-05-11 | A kind of detection method of rule of life exception |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610310530.5A CN106021877B (en) | 2016-05-11 | 2016-05-11 | A kind of detection method of rule of life exception |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106021877A CN106021877A (en) | 2016-10-12 |
CN106021877B true CN106021877B (en) | 2018-07-20 |
Family
ID=57100489
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610310530.5A Expired - Fee Related CN106021877B (en) | 2016-05-11 | 2016-05-11 | A kind of detection method of rule of life exception |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106021877B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP7157961B2 (en) * | 2018-09-28 | 2022-10-21 | パナソニックIpマネジメント株式会社 | Monitoring system and monitoring method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2001076272A (en) * | 1999-09-03 | 2001-03-23 | Matsushita Electric Works Ltd | Human body detector |
CN102133090A (en) * | 2011-03-01 | 2011-07-27 | 无锡爱维特信息技术有限公司 | Health detecting and monitoring system based on wireless network |
CN103150579A (en) * | 2013-02-25 | 2013-06-12 | 东华大学 | Abnormal human behavior detecting method based on video sequence |
CN103577795A (en) * | 2012-07-30 | 2014-02-12 | 索尼公司 | Detection equipment and method, detector generation equipment and method and monitoring system |
-
2016
- 2016-05-11 CN CN201610310530.5A patent/CN106021877B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2001076272A (en) * | 1999-09-03 | 2001-03-23 | Matsushita Electric Works Ltd | Human body detector |
CN102133090A (en) * | 2011-03-01 | 2011-07-27 | 无锡爱维特信息技术有限公司 | Health detecting and monitoring system based on wireless network |
CN103577795A (en) * | 2012-07-30 | 2014-02-12 | 索尼公司 | Detection equipment and method, detector generation equipment and method and monitoring system |
CN103150579A (en) * | 2013-02-25 | 2013-06-12 | 东华大学 | Abnormal human behavior detecting method based on video sequence |
Non-Patent Citations (1)
Title |
---|
基于混合高斯模型和主成分分析的轨迹分析行为识别方法;田国会 等;《电子学报》;20160115(第1期);第143-149页 * |
Also Published As
Publication number | Publication date |
---|---|
CN106021877A (en) | 2016-10-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US8749391B2 (en) | Fall detection system | |
US12011258B2 (en) | Method and apparatus for determining a fall risk | |
CN107092806A (en) | It is a kind of towards the intelligentized information fusion of old man's household and method for early warning | |
CN113397520B (en) | Information detection method and device for indoor object, storage medium and processor | |
CN103295364B (en) | A kind of fall events detection system based on two mirror | |
CN203812370U (en) | Ground mat type fall-down detection alarm device | |
CN107909771A (en) | A kind of personnel's tumble alarm system and its implementation based on wireless sensor network | |
CN109697830A (en) | A kind of personnel's anomaly detection method based on target distribution rule | |
CN106599802A (en) | Intelligent corridor monitoring system based on cloud technology | |
CN107633883A (en) | A kind of health degree evaluation method based on big data analysis | |
Razum et al. | Optimal threshold selection for threshold-based fall detection algorithms with multiple features | |
CN107018206A (en) | A kind of suitable aging intelligent home system | |
CN114140997B (en) | Monitoring and early warning system and method for residence and physical condition of old people in rest house toilet | |
CN107019501B (en) | Remote tumble detection method and system based on genetic algorithm and probabilistic neural network | |
CN106021877B (en) | A kind of detection method of rule of life exception | |
CN104872001B (en) | A kind of method that live pig existing state is monitored by heart rate data | |
CN109567770A (en) | A kind of measuring of human health method and system | |
CN107661092A (en) | Vital sign state monitoring method and computer-readable recording medium | |
CN105380622A (en) | Multi-data monitoring method for health early warning | |
CN110084081A (en) | A kind of tumble early warning realization method and system | |
CN105404778A (en) | Smart caring method | |
JP7026740B2 (en) | Systems and methods for monitoring position safety | |
Power et al. | Developing a sensor based homecare system: The role of bluetooth low-energy in activity monitoring | |
CN111671397A (en) | Health management system and method based on sleep monitoring | |
JP2020190889A (en) | Monitoring system for care-needing person |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180720 Termination date: 20190511 |