CN115166636B - Factory personnel supervision method based on multi-feature channel state information edge calculation - Google Patents

Factory personnel supervision method based on multi-feature channel state information edge calculation Download PDF

Info

Publication number
CN115166636B
CN115166636B CN202211075973.2A CN202211075973A CN115166636B CN 115166636 B CN115166636 B CN 115166636B CN 202211075973 A CN202211075973 A CN 202211075973A CN 115166636 B CN115166636 B CN 115166636B
Authority
CN
China
Prior art keywords
personnel
reference signal
signal receiver
person
subcarrier
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
CN202211075973.2A
Other languages
Chinese (zh)
Other versions
CN115166636A (en
Inventor
谈玲
孙雷
夏景明
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.)
Nanjing University of Information Science and Technology
Original Assignee
Nanjing University of Information Science and Technology
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 Nanjing University of Information Science and Technology filed Critical Nanjing University of Information Science and Technology
Priority to CN202211075973.2A priority Critical patent/CN115166636B/en
Publication of CN115166636A publication Critical patent/CN115166636A/en
Application granted granted Critical
Publication of CN115166636B publication Critical patent/CN115166636B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S5/00—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations
    • G01S5/02—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations using radio waves
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S1/00—Beacons or beacon systems transmitting signals having a characteristic or characteristics capable of being detected by non-directional receivers and defining directions, positions, or position lines fixed relatively to the beacon transmitters; Receivers co-operating therewith
    • G01S1/02—Beacons or beacon systems transmitting signals having a characteristic or characteristics capable of being detected by non-directional receivers and defining directions, positions, or position lines fixed relatively to the beacon transmitters; Receivers co-operating therewith using radio waves
    • G01S1/08—Systems for determining direction or position line
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • H04B17/309—Measuring or estimating channel quality parameters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H04W4/029—Location-based management or tracking services
    • 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
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • General Physics & Mathematics (AREA)
  • Signal Processing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Theoretical Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Quality & Reliability (AREA)
  • Electromagnetism (AREA)
  • Testing Or Calibration Of Command Recording Devices (AREA)

Abstract

The invention relates to a factory personnel supervision method based on multi-feature channel state information edge calculation, which is characterized in that the position information of a reference point is represented by adopting a distance correlation coefficient of amplitude entropy, phase difference and amplitude between carriers in CSI (channel State information), and three feature data are fused and input to a DQN (differential Quadrature reference network) algorithm to carry out operation of judging areas of workers, so that the robustness and the accuracy of the DQN algorithm are improved; considering that the dimension of the decision space in the DQN algorithm is larger, the burden of an edge server is increased in actual operation, and the operation delay is increased, the invention adopts a cross-domain method, and the dimension of the decision space is timely adjusted through the pressure values reflected by the FlexPresure pressure sensors, so that the problem of larger dimension of the decision space in the DQN algorithm is solved; the implementation of the method needs to be carried out in a continuous time period, and the method is more suitable for the method considering that the time delay performance of the DQN algorithm in deep reinforcement learning is not high.

Description

Factory personnel supervision method based on multi-feature channel state information edge calculation
Technical Field
The invention relates to a factory personnel supervision method based on multi-feature channel state information edge calculation, and belongs to the technical field of edge calculation.
Background
In recent years, more and more competitors emerge in various industries, and the profits of the plant operators are continuously squeezed, so that the demands for improving the competitive advantages and the production efficiency are more and more urgent. However, in an unsupervised factory environment, the working attitude of the staff is not active, and the working efficiency is not guaranteed, so that the production efficiency of the factory is reduced.
Disclosure of Invention
The invention aims to solve the technical problem of providing a factory personnel supervision method based on multi-feature channel state information edge calculation, which can monitor the working state of each personnel in a factory in real time. And the effect of the working efficiency of each person is statistically analyzed.
The invention adopts the following technical scheme for solving the technical problems: the invention designs a factory personnel supervision method based on multi-feature channel state information edge calculation, which is used for supervising the condition of the factory personnel
Figure 888590DEST_PATH_IMAGE001
The target factory area of each working position realizes the aim of the target factory area
Figure 973220DEST_PATH_IMAGE001
Monitoring the position of each person, wherein each person wears a work suit with an embedded receiving antenna and a work boot with a built-in pressure sensor; the method for supervising the factory personnel comprises the following steps of A to B, and a personnel working position judgment model is obtained; according to the step i, the position monitoring of the personnel to be monitored is realized by applying a personnel working position judgment model;
step A, based on that each person respectively corresponds to each working position in the target factory area one by one, continuously broadcasting by using a transmitting antenna which covers the target factory area by signals and is used as a signal transmitter
Figure 797957DEST_PATH_IMAGE002
Each data packet including a sequence
Figure 914293DEST_PATH_IMAGE003
B, sub-carriers, a receiving antenna which is positioned in the target factory area and used as a reference signal receiver, and a receiving antenna which is positioned in a work clothes where each person is positioned and used as a reference signal receiver, sequentially receive each data packet from a signal transmitter respectively, build a fingerprint database, and then enter the step B;
b, collecting amplitude entropy characteristics corresponding to the reference signal receivers of the personnel according to the method of the step A and the preset period duration based on the fact that the personnel are randomly free in the target plant area
Figure 776070DEST_PATH_IMAGE004
Phase difference characteristics
Figure 347996DEST_PATH_IMAGE005
And the corresponding inter-carrier amplitude distance correlation coefficient characteristics between the personnel reference signal receiver and the signal transmitter
Figure 586211DEST_PATH_IMAGE006
Meanwhile, fluctuation information of the sum of the numerical values of the pressure sensors in the two work boots respectively worn by each person under the condition of corresponding cycle duration is periodically acquired
Figure 684617DEST_PATH_IMAGE007
Training a deep reinforcement learning algorithm model by combining a fingerprint library to obtain a personnel working position judgment model;
step i, collecting amplitude entropy characteristics corresponding to a reference signal receiver of a person to be monitored
Figure 717295DEST_PATH_IMAGE004
Phase difference characteristics
Figure 42097DEST_PATH_IMAGE008
And the correlation coefficient characteristic of the amplitude distance between corresponding carriers between the reference signal receiver and the signal transmitter of the personnel to be monitored
Figure 818423DEST_PATH_IMAGE006
Simultaneously acquiring fluctuation information of the sum of the numerical values of the pressure sensors in two work boots worn by the person to be monitored under the condition of corresponding cycle duration
Figure 646702DEST_PATH_IMAGE007
And applying a working position judgment model of the personnel to obtain the working position of the personnel to be monitored.
As a preferred technical scheme of the invention: in the step A, a fingerprint database is built according to the following steps A1 to A5;
step A1, respectively aiming at each subcarrier in each data packet broadcasted by a signal transmitter, acquiring channel state information of the subcarriers between the signal transmitter and a reference signal receiver, and constructing CSI information of a channel between the signal transmitter and the reference signal receiver
Figure 240494DEST_PATH_IMAGE009
The following:
Figure 784083DEST_PATH_IMAGE010
wherein,
Figure 36204DEST_PATH_IMAGE011
，
Figure 984569DEST_PATH_IMAGE012
，
Figure 890208DEST_PATH_IMAGE013
representing signals between signal transmitters and reference signal receiversAbout a
Figure 189602DEST_PATH_IMAGE014
In a data packet
Figure 838889DEST_PATH_IMAGE015
Channel state information of the subcarriers;
meanwhile, aiming at each subcarrier in each data packet broadcasted by a signal transmitter, channel state information about the subcarrier between the signal transmitter and each personnel reference signal receiver is collected, and CSI information of a channel between the signal transmitter and each personnel reference signal receiver is constructed
Figure 907340DEST_PATH_IMAGE016
The following:
Figure 992669DEST_PATH_IMAGE017
wherein,
Figure 513780DEST_PATH_IMAGE018
，
Figure 107704DEST_PATH_IMAGE019
indicating signal transmitters and
Figure 30660DEST_PATH_IMAGE020
CSI information of channels between individual personal reference signal receivers,
Figure 543681DEST_PATH_IMAGE021
indicating signal transmitters and
Figure 817668DEST_PATH_IMAGE020
channel between personal reference signal receivers with respect to
Figure 808758DEST_PATH_IMAGE022
In a data packet
Figure 583291DEST_PATH_IMAGE015
Channel state information of the subcarriers; then entering the step A2;
step A2, aiming at each personnel reference signal receiver respectively, so as to
Figure 1634DEST_PATH_IMAGE023
Form a signal transmitter, the second
Figure 231758DEST_PATH_IMAGE020
Channel between personal reference signal receivers and between signal transmitter and reference signal receiver
Figure 698643DEST_PATH_IMAGE022
In a data packet
Figure 596192DEST_PATH_IMAGE015
Amplitude entropy of individual subcarriers, constituting
Figure 591961DEST_PATH_IMAGE020
Amplitude entropy characteristics corresponding to personal reference signal receiver
Figure 837610DEST_PATH_IMAGE024
The following were used:
Figure 170502DEST_PATH_IMAGE025
further obtaining amplitude entropy characteristics respectively corresponding to the personnel reference signal receivers; then entering the step A3;
step A3, aiming at each personnel reference signal receiver respectively, so as to
Figure 656978DEST_PATH_IMAGE026
Form a signal transmitter, the second
Figure 682703DEST_PATH_IMAGE020
A signal transmitter and a channel between individual reference signal receivers,Channel correlation between reference signal receivers
Figure 887419DEST_PATH_IMAGE022
In a data packet
Figure 289582DEST_PATH_IMAGE015
Phase difference of sub-carriers, constituting the second
Figure 364985DEST_PATH_IMAGE020
Phase difference characteristics corresponding to personal reference signal receiver
Figure 561611DEST_PATH_IMAGE027
The following were used:
Figure 250694DEST_PATH_IMAGE028
further acquiring phase difference characteristics respectively corresponding to the personnel reference signal receivers; wherein,
Figure 659810DEST_PATH_IMAGE029
to represent
Figure 120878DEST_PATH_IMAGE030
The phase of (a) is determined,
Figure 488406DEST_PATH_IMAGE031
to represent
Figure 261190DEST_PATH_IMAGE032
Then step A4 is entered;
step A4, according to the CSI information of the channels between the signal transmitter and each personnel reference signal receiver
Figure 880521DEST_PATH_IMAGE033
Extracting the distance correlation coefficient characteristics of the amplitudes between the carriers, and constructing the corresponding amplitude distance correlation coefficient characteristics between the signal transmitter and the personnel reference signal receivers
Figure 727254DEST_PATH_IMAGE034
Then, entering the step A5;
step A5, constructing a fingerprint database as follows:
Figure 265683DEST_PATH_IMAGE035
。
as a preferred technical scheme of the invention: the step A4 comprises the following steps A4-1 to A4-3;
step A4-1, according to the data packets respectively including the sequence
Figure 941077DEST_PATH_IMAGE036
The subcarriers are combined by the subcarriers with different sorting positions in pairs to obtain each subcarrier combination, and meanwhile, the subcarrier combinations are obtained according to the
Figure 957574DEST_PATH_IMAGE037
Combining each data packet by two different data packets to obtain each data packet combination, and then entering the step A4-2;
step A4-2, aiming at each personnel reference signal receiver, further aiming at each subcarrier combination, executing the step A4-2-1 to the step A4-2-3, obtaining the distance correlation coefficient between two subcarriers of the personnel reference signal receiver corresponding to the subcarrier combination, and further obtaining the distance correlation coefficient corresponding to each subcarrier combination of each personnel reference signal receiver
Figure 658814DEST_PATH_IMAGE038
(ii) a Then entering the step A4-3;
step A4-2-1, CSI information according to a channel between a signal transmitter and a personnel reference signal receiver
Figure 102565DEST_PATH_IMAGE039
For subcarrier combinations, based on
Figure 662990DEST_PATH_IMAGE040
、
Figure 14337DEST_PATH_IMAGE041
、
Figure 708099DEST_PATH_IMAGE042
With respect to the second located in a subcarrier combination
Figure 588330DEST_PATH_IMAGE043
Subcarrier, number one
Figure 229527DEST_PATH_IMAGE044
Sub-carriers, further combined with each data packet combination, respectively, based on
Figure 587827DEST_PATH_IMAGE045
、
Figure 998080DEST_PATH_IMAGE046
Obtaining the first data packet in the data packet assembly
Figure 49213DEST_PATH_IMAGE047
In a data packet
Figure 912127DEST_PATH_IMAGE043
Sub-carriers and
Figure 74118DEST_PATH_IMAGE048
in a data packet
Figure 476893DEST_PATH_IMAGE043
Euclidean distance between subcarriers
Figure 964506DEST_PATH_IMAGE049
Further, each data packet combination is obtained respectively about the first data packet in the data packet combination
Figure 314716DEST_PATH_IMAGE043
Euclidean distance of subcarriers
Figure 280398DEST_PATH_IMAGE050
And are combined with
Figure 134084DEST_PATH_IMAGE050
As the first in the matrix
Figure 527019DEST_PATH_IMAGE047
Go to the first
Figure 895684DEST_PATH_IMAGE048
Column element, constructing the second in the packet assembly
Figure 537493DEST_PATH_IMAGE043
Matrix corresponding to subcarrier
Figure 776845DEST_PATH_IMAGE051
(ii) a Similarly constructing the second in the data packet combination
Figure 606260DEST_PATH_IMAGE052
Matrix corresponding to subcarrier
Figure 196642DEST_PATH_IMAGE053
(ii) a Then entering the step A4-2-2;
step A4-2-2. Based on the matrix
Figure 379493DEST_PATH_IMAGE051
To do so by
Figure 473350DEST_PATH_IMAGE054
Building a matrix
Figure 216877DEST_PATH_IMAGE051
Corresponding center distance matrix
Figure 28976DEST_PATH_IMAGE055
Wherein
Figure 140151DEST_PATH_IMAGE056
to representMatrix array
Figure 88515DEST_PATH_IMAGE051
To middle
Figure 994155DEST_PATH_IMAGE047
The average value of the rows is then calculated,
Figure 27970DEST_PATH_IMAGE057
representation matrix
Figure 411678DEST_PATH_IMAGE051
To middle
Figure 480128DEST_PATH_IMAGE048
The mean value of the columns is,
Figure 819318DEST_PATH_IMAGE058
representation matrix
Figure 606008DEST_PATH_IMAGE051
The overall mean value of; building matrix by same principle
Figure 934352DEST_PATH_IMAGE053
Corresponding center distance matrix
Figure 122888DEST_PATH_IMAGE059
Then, entering the step A4-2-3;
step A4-2-3, according to the following formula:
Figure 370330DEST_PATH_IMAGE060
calculating to obtain the second position of the human reference signal receiver correspondingly positioned in the subcarrier combination
Figure 378737DEST_PATH_IMAGE043
Subcarrier, number one
Figure 635406DEST_PATH_IMAGE052
Distance-related relation between subcarriersNumber of
Figure 144360DEST_PATH_IMAGE061
(ii) a Wherein,
Figure 562703DEST_PATH_IMAGE062
，
Figure 792828DEST_PATH_IMAGE063
is shown as
Figure 587608DEST_PATH_IMAGE043
Sub-carriers and
Figure 219578DEST_PATH_IMAGE052
the distance covariance between the sub-carriers,
Figure 808822DEST_PATH_IMAGE064
，
Figure 791822DEST_PATH_IMAGE065
is shown as
Figure 655873DEST_PATH_IMAGE043
The distance variance of the subcarriers;
step A4-3, according to the distance correlation coefficient corresponding to each subcarrier combination respectively corresponding to each personnel reference signal receiver
Figure 139419DEST_PATH_IMAGE066
Constructing the characteristics of amplitude distance correlation coefficient between carriers corresponding to the signal transmitter and each personnel reference signal receiver respectively
Figure 571669DEST_PATH_IMAGE067
The following were used:
Figure 307543DEST_PATH_IMAGE068
。
as a preferred technical scheme of the invention: the step B comprises the following steps B1 to B4;
step B1, collecting amplitude entropy characteristics corresponding to the reference signal receivers of each person according to the method of the step A and the preset period duration based on the fact that each person randomly dissociates in the target plant area
Figure 585072DEST_PATH_IMAGE069
Phase difference characteristics
Figure 660475DEST_PATH_IMAGE070
And the corresponding inter-carrier amplitude distance correlation coefficient characteristics between the personnel reference signal receiver and the signal transmitter
Figure 537995DEST_PATH_IMAGE071
Meanwhile, fluctuation information of the sum of the numerical values of the pressure sensors in the two work boots respectively worn by each person under the condition of corresponding cycle duration is periodically acquired
Figure 370952DEST_PATH_IMAGE072
Obtaining the acquisition time of each period
Figure 576806DEST_PATH_IMAGE073
System state of
Figure 569033DEST_PATH_IMAGE074
The following were used:
Figure 811926DEST_PATH_IMAGE075
wherein,
Figure 253884DEST_PATH_IMAGE076
or
Figure 732270DEST_PATH_IMAGE077
，
Figure 47845DEST_PATH_IMAGE078
Indicating the time of acquisition of the cycle
Figure 586274DEST_PATH_IMAGE079
First to
Figure 987299DEST_PATH_IMAGE080
The fluctuation range of the sum of the values of the pressure sensors in the two work boots of the individual exceeds the preset pressure threshold range,
Figure 534955DEST_PATH_IMAGE081
indicating the time of acquisition of the cycle
Figure 377140DEST_PATH_IMAGE079
First to
Figure 83541DEST_PATH_IMAGE080
The fluctuation range of the sum of the numerical values of the pressure sensors in the two work boots of the individual member does not exceed the range of the preset pressure threshold value; then entering step B2;
step B2, aiming at each person and each period, acquiring time
Figure 237442DEST_PATH_IMAGE079
System state of
Figure 57630DEST_PATH_IMAGE074
If the personnel correspond to the continuous collection time of each period
Figure 347797DEST_PATH_IMAGE082
Then the system state at the first time in the continuous cycle collection time is reserved
Figure 962449DEST_PATH_IMAGE074
The record corresponding to the person in (1) deletes the system state at each of the rest of the continuous periodic acquisition moments
Figure 338067DEST_PATH_IMAGE074
The corresponding record of the personnel in (1) realizes the acquisition of the time of each period
Figure 961946DEST_PATH_IMAGE079
System state of
Figure 106620DEST_PATH_IMAGE074
Then step B3 is entered;
step B3, aiming at each period of acquisition time respectively
Figure 154823DEST_PATH_IMAGE079
System state of
Figure 752157DEST_PATH_IMAGE074
Obtaining the system state
Figure 179728DEST_PATH_IMAGE074
The combination of the personnel involved in the system and the working positions respectively forms the system state
Figure 444487DEST_PATH_IMAGE074
The corresponding personnel position set is then entered into step B4;
step B4, collecting time based on each period
Figure 932100DEST_PATH_IMAGE079
System state of
Figure 16731DEST_PATH_IMAGE074
Corresponding personnel position set, combined with fingerprint library, and reward function under ordered pairs of each personnel corresponding to a working position combination
Figure 982413DEST_PATH_IMAGE083
The following were used:
Figure 367258DEST_PATH_IMAGE084
training the deep reinforcement learning algorithm model until convergence, and obtaining amplitude entropy characteristics corresponding to the personnel reference signal receiver
Figure 768982DEST_PATH_IMAGE085
Phase difference characteristics
Figure 747433DEST_PATH_IMAGE086
And the corresponding inter-carrier amplitude distance correlation coefficient characteristics between the personnel reference signal receiver and the signal transmitter
Figure 516806DEST_PATH_IMAGE087
For inputting, the working position of the person is the output working position judgment model of the person; wherein,
Figure 490578DEST_PATH_IMAGE088
representation matrix
Figure 54415DEST_PATH_IMAGE089
And
Figure 113638DEST_PATH_IMAGE090
the euclidean distance between them,
Figure 421122DEST_PATH_IMAGE091
representing the euclidean distance between matrix a and matrix B.
As a preferred technical scheme of the invention: the preset pressure threshold range is 90N to 120N.
As a preferred technical scheme of the invention: and (e) acquiring the working position of the person to be monitored based on the execution of the step i, if the person to be monitored is not at any working position, judging that the person to be monitored is in a rest state at the moment, further carrying out periodic detection on the person to be monitored according to the step i within the working time of one day, counting the rest time of the person to be monitored, judging whether the rest time exceeds the preset rest time range, if so, judging that the person to be monitored does not reach the workload requirement of one day, and if not, judging that the person to be monitored reaches the workload requirement of one day.
As a preferred technical scheme of the invention: the receiving antenna embedded in the working clothes is embedded in the tail part of the back of the working clothes, and the pressure sensor embedded in the working boot is a FlexPresure pressure sensor.
Compared with the prior art, the factory personnel supervision method based on the multi-feature channel state information edge calculation has the following technical effects by adopting the technical scheme:
according to the factory personnel supervision method based on multi-feature channel state information edge calculation, the reference point position information is represented by the distance correlation coefficient of amplitude entropy, phase difference and amplitude between carriers in CSI, three feature data are fused and input to a DQN algorithm to perform operation of worker region judgment, and the robustness and accuracy of the DQN algorithm are improved; considering that the dimension of the decision space in the DQN algorithm is large, the burden of an edge server is increased in actual operation, and the operation delay is increased, the invention adopts a cross-field method, and the dimension of the decision space is timely adjusted through the pressure values reflected by the FlexPresure pressure sensors, so that the problem of large dimension of the decision space in the DQN algorithm is solved; the implementation of the method needs to be carried out in a continuous time period, and the method is more suitable for the method considering that the time delay performance of the DQN algorithm in deep reinforcement learning is not high.
Drawings
FIG. 1 is a schematic diagram of an application flow of a plant personnel supervision method based on multi-feature channel state information edge calculation according to the present invention;
FIG. 2 is a diagram of a plant environment model illustrating an embodiment of the present invention;
FIG. 3 is a DQN diagram for identifying areas to which personnel in a plant belong in an embodiment of the present invention;
FIG. 4 is a characteristic diagram of amplitude entropy data of individual reference points in an embodiment of the present invention
FIG. 5 is a diagram showing the convergence of DQN algorithm in an embodiment of the present invention;
fig. 6 is an accumulated error distribution diagram of the DQN algorithm and the KNN and WKNN classification algorithms in an embodiment of the present invention.
Detailed Description
The following description will explain embodiments of the present invention in further detail with reference to the accompanying drawings.
The method is designed to judge whether a certain worker is in a working state, whether the worker is in a reasonable working area needs to be judged, and the CSI is considered to be more stable and finer in granularity, so that the method is widely applied to the field of personnel detection and positioning. The worker area distinguishing method related by the invention essentially adopts the fingerprint matching technology, and in order to ensure the feasibility of the fingerprint matching technology, different CSI (channel state information) features need to be extracted to represent the positions of different reference points, so that the feature data in a fingerprint database needs to meet the following characteristics: 1) The characteristic data values corresponding to different reference points in the fingerprint database have considerable differences; 2) The characteristic data values corresponding to the same reference point in the fingerprint database at different times have relative stability. Through a plurality of groups of data acquisition tests, the amplitude entropy and the distance correlation coefficient of the amplitude between the carriers are shown to meet the characteristics, the data difference of the phase difference at different reference points is not considerable enough, so that the weight of the phase difference at the time of data matching needs to be properly reduced, and the distance correlation coefficient of the amplitude between the carriers can reflect the state whether a link is invaded or not, so that the phase difference can represent the environmental information of a factory.
Based on the design idea, the invention designs a plant personnel supervision method based on multi-feature channel state information edge calculation, as shown in FIG. 2, for aiming at the system comprising
Figure 512051DEST_PATH_IMAGE092
The target factory area of each working position realizes the aim of the target factory area
Figure 981209DEST_PATH_IMAGE092
Monitoring the position of each person, wherein each person wears a work suit with an embedded receiving antenna and a work boot with a built-in pressure sensor; in the application, the receiving antenna embedded in the working clothes is embedded in the tail part of the back of the working clothes, and the pressure sensor embedded in the working boot is a FlexPresure pressure sensor.
In practical application, as shown in fig. 1, the method for supervising plant personnel includes the following steps a to B, and a personnel working position determination model is obtained.
Step A, based on that each person respectively corresponds to each working position in the target factory area one by one, continuously broadcasting by using a transmitting antenna which covers the target factory area by signals and is used as a signal transmitter
Figure 793307DEST_PATH_IMAGE093
Each data packet including a sequence
Figure 373324DEST_PATH_IMAGE094
And B, respectively and sequentially receiving each data packet from the signal transmitter by a receiving antenna which is positioned in the target plant area and used as a reference signal receiver and a receiving antenna which is positioned in the work clothes of each person and used as a reference signal receiver, building a fingerprint database, and then entering the step B.
In practical application, the step a is specifically designed and executed as the following steps A1 to A5, and a fingerprint database is built.
Step A1, respectively aiming at each subcarrier in each data packet broadcasted by a signal transmitter, acquiring channel state information of the subcarriers between the signal transmitter and a reference signal receiver, and constructing CSI information of a channel between the signal transmitter and the reference signal receiver
Figure 321689DEST_PATH_IMAGE009
The following were used:
Figure 758486DEST_PATH_IMAGE010
wherein,
Figure 57881DEST_PATH_IMAGE095
，
Figure 845184DEST_PATH_IMAGE096
，
Figure 913634DEST_PATH_IMAGE013
indicating channel-off between signal transmitter and reference signal receiverIn the first place
Figure 255753DEST_PATH_IMAGE022
In a data packet
Figure 511285DEST_PATH_IMAGE015
Channel state information of the subcarriers.
Meanwhile, aiming at each subcarrier in each data packet broadcasted by a signal transmitter, channel state information about the subcarrier between the signal transmitter and each personnel reference signal receiver is collected, and CSI information of a channel between the signal transmitter and each personnel reference signal receiver is constructed
Figure 229843DEST_PATH_IMAGE016
The following were used:
Figure 418379DEST_PATH_IMAGE017
wherein,
Figure 665820DEST_PATH_IMAGE097
，
Figure 408648DEST_PATH_IMAGE016
table representation signal transmitter and its
Figure 396808DEST_PATH_IMAGE020
CSI information of channels between individual personal reference signal receivers,
Figure 174272DEST_PATH_IMAGE021
indicating signal transmitters and
Figure 592615DEST_PATH_IMAGE020
channel between personal reference signal receivers with respect to
Figure 822739DEST_PATH_IMAGE014
In a data packet
Figure 351940DEST_PATH_IMAGE015
Channel state information of the subcarriers; then step A2 is entered.
Step A2, aiming at each personnel reference signal receiver respectively, so as to
Figure 515068DEST_PATH_IMAGE023
Form a signal transmitter, the second
Figure 245258DEST_PATH_IMAGE020
Channel between personal reference signal receivers and between signal transmitter and reference signal receiver
Figure 237047DEST_PATH_IMAGE022
In a data packet
Figure 569939DEST_PATH_IMAGE015
Amplitude entropy of individual subcarriers, constituting
Figure 321994DEST_PATH_IMAGE098
Amplitude entropy characteristics corresponding to personal reference signal receiver
Figure 347719DEST_PATH_IMAGE024
The following were used:
Figure 818015DEST_PATH_IMAGE025
further obtaining amplitude entropy characteristics respectively corresponding to the personnel reference signal receivers; then step A3 is entered.
Step A3, aiming at each personnel reference signal receiver respectively, so as to
Figure 95543DEST_PATH_IMAGE026
Form a signal transmitter, the second
Figure 233264DEST_PATH_IMAGE020
Channel and signal transmitter between personal reference signal receiversThe channel between the reference signal receivers is related to
Figure 426960DEST_PATH_IMAGE022
In a data packet
Figure 118973DEST_PATH_IMAGE015
Phase difference of sub-carriers constituting the second
Figure 59247DEST_PATH_IMAGE098
Phase difference characteristics corresponding to personal reference signal receiver
Figure 520315DEST_PATH_IMAGE027
The following were used:
Figure 356684DEST_PATH_IMAGE028
further acquiring phase difference characteristics respectively corresponding to the personnel reference signal receivers; wherein,
Figure 801572DEST_PATH_IMAGE029
to represent
Figure 14379DEST_PATH_IMAGE030
The phase of (a) is determined,
Figure 329953DEST_PATH_IMAGE031
to represent
Figure 599873DEST_PATH_IMAGE032
Then step A4 is entered.
Step A4, according to the CSI information of the channels between the signal transmitter and each personnel reference signal receiver
Figure 266478DEST_PATH_IMAGE033
Extracting the distance correlation coefficient characteristics of the amplitudes between the carriers, and constructing the amplitude distance correlation coefficient characteristics between the carriers corresponding to the signal transmitter and the personnel reference signal receivers respectively
Figure 282975DEST_PATH_IMAGE034
Then, step A5 is entered.
In practical applications, the step A4 is specifically designed to perform the following steps A4-1 to A4-3.
Step A4-1, according to the data packets respectively including the sequence
Figure 453057DEST_PATH_IMAGE099
The subcarriers are combined by the subcarriers with different sorting positions in pairs to obtain each subcarrier combination, and meanwhile, the subcarrier combinations are obtained according to the
Figure 162387DEST_PATH_IMAGE100
And (4) combining the data packets with two different data packets to obtain each data packet combination, and then entering the step A4-2.
Step A4-2, aiming at each personnel reference signal receiver, further aiming at each subcarrier combination, executing the step A4-2-1 to the step A4-2-3, obtaining the distance correlation coefficient between two subcarriers corresponding to the subcarrier combination by the personnel reference signal receiver, and further obtaining the distance correlation coefficient corresponding to each subcarrier combination by each personnel reference signal receiver
Figure 785129DEST_PATH_IMAGE038
(ii) a Then step A4-3 is entered.
Step A4-2-1, CSI information according to a channel between a signal transmitter and a personnel reference signal receiver
Figure 870897DEST_PATH_IMAGE039
For subcarrier combinations, based on
Figure 161064DEST_PATH_IMAGE101
、
Figure 303945DEST_PATH_IMAGE041
、
Figure 679562DEST_PATH_IMAGE042
With respect to the second located in a combination of sub-carriers
Figure 303442DEST_PATH_IMAGE043
Subcarrier, number one
Figure 448115DEST_PATH_IMAGE102
Sub-carriers, further combined with each data packet combination, respectively, based on
Figure 499248DEST_PATH_IMAGE045
、
Figure 627741DEST_PATH_IMAGE046
Obtaining the first data packet in the data packet assembly
Figure 789732DEST_PATH_IMAGE047
In a data packet
Figure 54491DEST_PATH_IMAGE043
Sub-carriers and
Figure 19735DEST_PATH_IMAGE048
in a data packet
Figure 369945DEST_PATH_IMAGE043
Euclidean distance between subcarriers
Figure 476572DEST_PATH_IMAGE049
Further, each data packet combination is obtained respectively about the first data packet in the data packet combination
Figure 861417DEST_PATH_IMAGE043
Euclidean distance of subcarriers
Figure 519932DEST_PATH_IMAGE050
And are combined with
Figure 91858DEST_PATH_IMAGE050
As the first in the matrix
Figure 861231DEST_PATH_IMAGE047
Go to the first
Figure 97653DEST_PATH_IMAGE048
Elements of a column, constituting the second in the packet assembly
Figure 802435DEST_PATH_IMAGE043
Matrix corresponding to subcarrier
Figure 127237DEST_PATH_IMAGE051
(ii) a Similarly constructing the second in the data packet combination
Figure 434722DEST_PATH_IMAGE052
Matrix corresponding to subcarrier
Figure 528580DEST_PATH_IMAGE053
(ii) a And then step A4-2-2 is entered.
Step A4-2-2. Based on the matrix
Figure 528897DEST_PATH_IMAGE051
To do so by
Figure 809836DEST_PATH_IMAGE054
Building a matrix
Figure 652503DEST_PATH_IMAGE051
Corresponding center distance matrix
Figure 600867DEST_PATH_IMAGE055
Wherein
Figure 506507DEST_PATH_IMAGE056
representation matrix
Figure 540322DEST_PATH_IMAGE051
To middle
Figure 189609DEST_PATH_IMAGE047
All of the rowsValue of,
Figure 258059DEST_PATH_IMAGE057
representation matrix
Figure 334599DEST_PATH_IMAGE051
To middle
Figure 121290DEST_PATH_IMAGE048
The mean value of the columns is,
Figure 571338DEST_PATH_IMAGE058
representation matrix
Figure 369661DEST_PATH_IMAGE051
The overall mean value of; constructing a matrix by the same principle
Figure 617103DEST_PATH_IMAGE053
Corresponding center distance matrix
Figure 625510DEST_PATH_IMAGE059
And then proceeds to step A4-2-3.
Step A4-2-3, according to the following formula:
Figure 616600DEST_PATH_IMAGE060
calculating to obtain the second position of the human reference signal receiver correspondingly positioned in the subcarrier combination
Figure 659642DEST_PATH_IMAGE043
Subcarrier, number one
Figure 609144DEST_PATH_IMAGE052
Distance correlation coefficient between subcarriers
Figure 113636DEST_PATH_IMAGE061
(ii) a Wherein,
Figure 642838DEST_PATH_IMAGE062
，
Figure 274807DEST_PATH_IMAGE063
is shown as
Figure 129631DEST_PATH_IMAGE043
Sub-carriers and
Figure 847051DEST_PATH_IMAGE052
the distance covariance between the sub-carriers,
Figure 586468DEST_PATH_IMAGE064
，
Figure 869682DEST_PATH_IMAGE065
is shown as
Figure 767843DEST_PATH_IMAGE043
The distance variance of the subcarriers.
Step A4-3, according to the distance correlation coefficient corresponding to each subcarrier combination respectively corresponding to each personnel reference signal receiver
Figure 238139DEST_PATH_IMAGE066
Constructing the characteristics of amplitude distance correlation coefficient between carriers corresponding to the signal transmitter and each personnel reference signal receiver respectively
Figure 640301DEST_PATH_IMAGE067
The following were used:
Figure 246863DEST_PATH_IMAGE103
。
step A5, constructing a fingerprint database as follows:
Figure 177910DEST_PATH_IMAGE035
。
step B, based on the fact that each person randomly dissociates in the target plant area, according to the preset period duration, the period is step by stepThe method of step A, collecting the corresponding amplitude entropy characteristics of each personnel reference signal receiver
Figure 807606DEST_PATH_IMAGE004
Phase difference characteristics
Figure 744950DEST_PATH_IMAGE005
And the correlation coefficient characteristic of the amplitude distance between the corresponding carriers between the personnel reference signal receiver and the signal transmitter
Figure 206018DEST_PATH_IMAGE006
Meanwhile, fluctuation information of the sum of the numerical values of the pressure sensors in the two work boots respectively worn by each person under the condition of corresponding cycle duration is periodically acquired
Figure 307967DEST_PATH_IMAGE007
And training a depth reinforcement learning algorithm model DQN by combining a fingerprint library to obtain a personnel working position judgment model.
In practical applications, the step B is specifically designed to perform the following steps B1 to B4.
Step B1, collecting amplitude entropy characteristics corresponding to the reference signal receivers of each person according to the method of the step A and the preset period duration based on the fact that each person randomly dissociates in the target plant area
Figure 487275DEST_PATH_IMAGE069
Phase difference characteristics
Figure 700082DEST_PATH_IMAGE070
And the corresponding inter-carrier amplitude distance correlation coefficient characteristics between the personnel reference signal receiver and the signal transmitter
Figure 546815DEST_PATH_IMAGE071
Meanwhile, fluctuation information of the sum of the numerical values of the pressure sensors in the two work boots respectively worn by each person under the condition of corresponding cycle duration is periodically acquired
Figure 554085DEST_PATH_IMAGE072
Obtaining the acquisition time of each period
Figure 220690DEST_PATH_IMAGE073
System state of
Figure 234258DEST_PATH_IMAGE104
The following were used:
Figure 76443DEST_PATH_IMAGE075
wherein,
Figure 785773DEST_PATH_IMAGE076
or
Figure 939674DEST_PATH_IMAGE077
，
Figure 759862DEST_PATH_IMAGE078
Indicating the time of acquisition of the cycle
Figure 315609DEST_PATH_IMAGE079
First to
Figure 930261DEST_PATH_IMAGE105
The fluctuation range of the sum of the values of the pressure sensors in the two work boots of the individual exceeds the preset pressure threshold range,
Figure 314668DEST_PATH_IMAGE081
indicating the time of acquisition of the cycle
Figure 938547DEST_PATH_IMAGE079
First to
Figure 348800DEST_PATH_IMAGE080
The fluctuation range of the sum of the numerical values of the pressure sensors in the two work boots of the individual member does not exceed the range of the preset pressure threshold value; then step B2 is entered.
In practical applications, the preset pressure threshold range varies with the weight of each worker, and the specific design herein uses 90N to 120N, and the threshold value should not exceed the weight of the corresponding worker.
Step B2, aiming at each person and each period, acquiring time
Figure 665511DEST_PATH_IMAGE079
System state of
Figure 528425DEST_PATH_IMAGE074
If the personnel correspond to the continuous collection time of each period
Figure 424837DEST_PATH_IMAGE082
Then the system state at the first time in the continuous collection time of each period is reserved
Figure 424017DEST_PATH_IMAGE074
The record corresponding to the person in (1) deletes the system state at each of the rest of the continuous periodic acquisition moments
Figure 646051DEST_PATH_IMAGE074
The corresponding record of the personnel in (1) realizes the acquisition of the time of each period
Figure 993331DEST_PATH_IMAGE079
System state of
Figure 959013DEST_PATH_IMAGE074
Then step B3 is entered.
Step B3, respectively aiming at each period of acquisition time
Figure 78279DEST_PATH_IMAGE079
System state of
Figure 205635DEST_PATH_IMAGE074
Obtaining the system state
Figure 43141DEST_PATH_IMAGE074
ZhongshiInvolving the combination of persons with working positions, respectively, to form the system state
Figure 953459DEST_PATH_IMAGE074
The corresponding personnel position set is then entered into step B4;
step B4, collecting time based on each period
Figure 192811DEST_PATH_IMAGE106
System state of
Figure 753718DEST_PATH_IMAGE074
Corresponding personnel position set, combined with fingerprint library, and reward function under ordered pairs of each personnel corresponding to a working position combination
Figure 78520DEST_PATH_IMAGE107
The following were used:
Figure 386004DEST_PATH_IMAGE108
training the deep reinforcement learning algorithm model until convergence, and obtaining amplitude entropy characteristics corresponding to the personnel reference signal receiver
Figure 214283DEST_PATH_IMAGE085
Phase difference characteristics
Figure 214600DEST_PATH_IMAGE086
And the correlation coefficient characteristic of the amplitude distance between the corresponding carriers between the personnel reference signal receiver and the signal transmitter
Figure 761119DEST_PATH_IMAGE087
For inputting, the working position of the person is the output working position judgment model of the person; wherein,
Figure 872294DEST_PATH_IMAGE088
representation matrix
Figure 820659DEST_PATH_IMAGE089
And
Figure 723368DEST_PATH_IMAGE090
the euclidean distance between them,
Figure 757183DEST_PATH_IMAGE091
representing the euclidean distance between matrix a and matrix B.
The structure diagram of the DQN algorithm is shown in FIG. 3;
step B41, constructing two neural networks with completely similar structures and completely identical parameters, respectively recording the neural networks as an evaluation network and a target network, and recording the set of all parameters of the evaluation network as an evaluation network
Figure 406471DEST_PATH_IMAGE109
The set of all parameters of the target network is recorded as
Figure 209341DEST_PATH_IMAGE110
；
Step B42 intWhen the CSI and the pressure value data are collected for the second time, the current state of the system is baseds t And evaluating Q predicted values reflected by each action decision in a corresponding decision space of the network, and combining the Q predicted values
Figure 285882DEST_PATH_IMAGE111
Greedy selection policy determination of action decisionsa t Further receive a rewardr t And enters the next state of the systems t+1 Simultaneously recording the records t , a t , r t , s t+1 Storing the data in an experience playback pool;
step B43, starting from the 1 st acquisition of CSI and pressure data, repeatedly executing the step S452 until the experience playback pool is filled up;
step B44, randomly extracting N samples from the experience playback pool, training the evaluation network by adopting the N samples, and recording one sample as as t , a t , r t , s t+1 }；
Step B45, mixings t 、a t Inputting the Q predicted value to an evaluation network
Figure 72572DEST_PATH_IMAGE112
(ii) a Will be provided withs t+1 Inputting to a target network, obtainings t+1 Corresponding maximum Q target value
Figure 400917DEST_PATH_IMAGE113
Whereina'Represents according to the states t+1 The action decision taken when the target network obtains the maximum Q value;
step B46, determinings t , a t , r t , s t+1 Whether or notNThe last sample of the samples, if so, order
Figure 332662DEST_PATH_IMAGE114
(ii) a If not, order
Figure 580104DEST_PATH_IMAGE115
；
Step B47, for
Figure 854091DEST_PATH_IMAGE116
And
Figure 845180DEST_PATH_IMAGE117
difference between them, using gradient descent method to pair parameter sets
Figure 622643DEST_PATH_IMAGE118
Updating is carried out;
step B48, updating the parameter set in the target network regularly
Figure 306566DEST_PATH_IMAGE119
。
Based on the obtained personnel working position judgment model, in practical application, the personnel working position judgment model is further applied to realize position monitoring of the personnel to be monitored.
Step i, collecting amplitude entropy characteristics corresponding to a reference signal receiver of a person to be monitored
Figure 943214DEST_PATH_IMAGE004
Phase difference characteristics
Figure 266224DEST_PATH_IMAGE120
And the correlation coefficient characteristic of the amplitude distance between corresponding carriers between the reference signal receiver and the signal transmitter of the personnel to be monitored
Figure 898194DEST_PATH_IMAGE006
Simultaneously acquiring fluctuation information of the sum of the numerical values of the pressure sensors in two work boots worn by the person to be monitored under the condition of corresponding cycle duration
Figure 487438DEST_PATH_IMAGE007
And applying a working position judgment model of the personnel to obtain the working position of the personnel to be monitored.
In practical application, based on the execution of the step i, the working position where the person to be monitored is located is obtained, if the person to be monitored is not located at any working position, the person to be monitored is judged to be in a rest state at the moment, further, the person to be monitored is periodically detected according to the step i within one-day working time, the rest time of the person to be monitored is counted, whether the rest time exceeds a preset rest time range is judged, if yes, the person to be monitored is judged not to reach the one-day workload requirement, and if not, the person to be monitored is judged to reach the one-day workload requirement.
The method for detecting the working state of the workers in the workshop based on the CSI-FlexPresure edge calculation is applied to practice, and fig. 4 shows the relation between the amplitude entropy corresponding to individual reference points under the same subcarrier and the sampling times. As shown in fig. 5, which is a diagram representing the convergence of the DQN algorithm applied in the present invention, since the parameters of the DQN network model are random in the initial stage of training, the probability of action decision taken at this time is not optimal, and as the training times increase, the DQN algorithm continuously optimizes the parameters of the network model, so that the action decision tends to be optimized, the reward function value tends to be maximized, and a steady state is reached. As can be seen from the figure, after 80 rounds of training, the DQN algorithm applied by the present invention tends to converge. Fig. 6 is an error accumulation distribution diagram of the KNN and WKNN classification algorithms and the DQN algorithm in this embodiment, and an experimental result shows that the algorithm provided in this embodiment can further improve the accuracy of the person region identification.
According to the factory personnel supervision method based on multi-feature channel state information edge calculation, the position information of the reference point is represented by the distance correlation coefficient of amplitude entropy, phase difference and amplitude between carriers in CSI, three feature data are fused and input to a DQN algorithm to perform operation of judging areas of workers, and the robustness and accuracy of the DQN algorithm are improved; considering that the dimension of the decision space in the DQN algorithm is large, the burden of an edge server is increased in actual operation, and the operation delay is increased, the invention adopts a cross-field method, and the dimension of the decision space is timely adjusted through the pressure values reflected by the FlexPresure pressure sensors, so that the problem of large dimension of the decision space in the DQN algorithm is solved; the implementation of the method needs to be carried out in a continuous time period, and the method is more suitable for the method considering that the time delay performance of the DQN algorithm in deep reinforcement learning is not high.
The embodiments of the present invention have been described in detail with reference to the drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the gist of the present invention.

Claims (6)

1. A factory personnel supervision method based on multi-feature channel state information edge calculation is characterized by comprising the following steps: the system is used for monitoring the positions of Z personnel in a target plant area aiming at the target plant area containing Z working positions, and each personnel wears a work clothes embedded with a receiving antenna and a work boot embedded with a pressure sensor; the method for supervising the factory personnel comprises the following steps of A to B, and a personnel working position judgment model is obtained; according to the step i, the position monitoring of the personnel to be monitored is realized by applying a personnel working position judgment model;
a, based on that each person is respectively positioned at each working position in a target factory area in a one-to-one correspondence manner, continuously broadcasting M data packets by using a transmitting antenna which covers the target factory area as a signal transmitter through signals, wherein each data packet respectively comprises N subcarriers in sequence, and each data packet from the signal transmitter is respectively received in sequence by a receiving antenna which is positioned in the target factory area as a reference signal receiver and a receiving antenna which is positioned in a working clothes where each person is positioned as a reference signal receiver, a fingerprint database is built, and then the step B is carried out;
in the step A, a fingerprint database is built according to the following steps A1 to A5;
step A1, respectively aiming at each subcarrier in each data packet broadcasted by a signal transmitter, acquiring channel state information of the subcarriers between the signal transmitter and a reference signal receiver, and constructing CSI information H of a channel between the signal transmitter and the reference signal receiver s The following were used:
Figure FDA0003916399720000011
wherein M is more than or equal to 1 and less than or equal to M, N is more than or equal to 1 and less than or equal to N,
Figure FDA0003916399720000012
channel state information indicating a channel between the signal transmitter and the reference signal receiver with respect to an nth subcarrier in an mth data packet;
meanwhile, aiming at each subcarrier in each data packet broadcasted by the signal transmitter,collecting channel state information about subcarriers between a signal transmitter and each personnel reference signal receiver respectively, and constructing CSI information H of channels between the signal transmitter and each personnel reference signal receiver respectively rz The following were used:
Figure FDA0003916399720000021
wherein Z is more than or equal to 1 and less than or equal to Z, H rz CSI information representing a channel between the signal transmitter and the z-th personal reference signal receiver,
Figure FDA0003916399720000022
channel state information indicating a channel between the signal transmitter and the z-th personal reference signal receiver with respect to an nth subcarrier of the mth data packet; then entering the step A2;
step A2. Aiming at each personnel reference signal receiver respectively, so as to
Figure FDA0003916399720000023
Forming the amplitude entropy of the channel between the signal transmitter and the z-th personnel reference signal receiver and the channel between the signal transmitter and the reference signal receiver with respect to the nth subcarrier in the m-th data packet, and forming the amplitude entropy characteristic div | H corresponding to the z-th personnel reference signal receiver z L is as follows:
Figure FDA0003916399720000024
further obtaining amplitude entropy characteristics respectively corresponding to the personnel reference signal receivers; then entering the step A3;
step A3, aiming at the reference signal receiver of each person respectively, so as to
Figure FDA0003916399720000025
Forming a channel between a signal transmitter and a z-th personnel reference signal receiver and transmitting signalsThe phase difference of the channel between the machine and the reference signal receiver about the nth subcarrier in the mth data packet forms the phase difference characteristic dif & lt H corresponding to the z th individual member reference signal receiver z The following were used:
Figure FDA0003916399720000031
further acquiring phase difference characteristics respectively corresponding to the personnel reference signal receivers; wherein,
Figure FDA0003916399720000032
to represent
Figure FDA0003916399720000033
The phase of (a) is determined,
Figure FDA0003916399720000034
to represent
Figure FDA0003916399720000035
Then step A4 is entered;
step A4, according to CSI information H of channels between the signal transmitter and each personnel reference signal receiver respectively rz Extracting the distance correlation coefficient characteristics of the amplitudes between the carriers, and constructing corresponding inter-carrier amplitude distance correlation coefficient characteristics dCorH between the signal transmitter and each personnel reference signal receiver rz Then, entering the step A5;
step A5, constructing a fingerprint database as follows:
Figure FDA0003916399720000036
b, collecting amplitude entropy characteristics div | H corresponding to the reference signal receiver of each person according to the method of the step A and the period according to the preset period duration based on the fact that each person randomly dissociates in the target plant area z | t Phase differenceSign dif & lt H z t And corresponding inter-carrier amplitude distance correlation coefficient characteristic dCorH between personnel reference signal receiver and signal transmitter rz t Meanwhile, fluctuation information of the sum of the numerical values of the pressure sensors in the two work boots respectively worn by each person under the condition of corresponding cycle duration is periodically acquired
Figure FDA0003916399720000038
Training a specified depth reinforcement learning algorithm model by combining a fingerprint library to obtain a personnel working position judgment model; step i, collecting amplitude entropy characteristics div | H corresponding to a reference signal receiver of a person to be monitored z | t Phase difference characteristic dif & lt H z t And corresponding inter-carrier amplitude distance correlation coefficient characteristic dCorH between reference signal receiver and signal transmitter of personnel to be monitored rz t Meanwhile, fluctuation information of the sum of the numerical values of the pressure sensors in two working boots worn by a person to be monitored under the condition of corresponding period duration is collected
Figure FDA0003916399720000037
And applying a worker working position judgment model to obtain the working position of the worker to be monitored.
2. The plant personnel supervision method based on multi-feature channel state information edge calculation according to claim 1, characterized by: the step A4 comprises the following steps A4-1 to A4-3;
step A4-1, according to the data packets respectively including N subcarriers in sequence, combining the subcarriers with different sorting positions in pairs to obtain each subcarrier combination, simultaneously, according to the M data packets, combining the different data packets in pairs to obtain each data packet combination, and then entering the step A4-2;
step A4-2, aiming at each personnel reference signal receiver, further aiming at each subcarrier combination, executing the steps A4-2-1 to A4-2-3 to obtain a distance correlation coefficient between two subcarriers in the subcarrier combination corresponding to the personnel reference signal receiver, and further obtaining the distance correlation coefficientObtaining the distance correlation coefficient corresponding to each subcarrier combination respectively corresponding to each personnel reference signal receiver
Figure FDA00039163997200000411
Then entering the step A4-3;
step A4-2-1, according to CSI information H of a channel between a signal transmitter and a personnel reference signal receiver rz For subcarrier combinations, based on i e [1,N]、j∈[1,N]And i is not equal to j, and the ith subcarrier and the jth subcarrier in the subcarrier combination are further combined with each data packet combination respectively based on x belongs to [1,M ]]、y∈[1,M]Obtaining the Euclidean distance between the ith subcarrier in the x data packet and the ith subcarrier in the y data packet in the data packet combination
Figure FDA0003916399720000041
Further, the Euclidean distance of each data packet combination respectively related to the ith subcarrier positioned in the data packet combination is obtained
Figure FDA0003916399720000042
And are provided with
Figure FDA0003916399720000043
As the element of the x row and y column in the matrix, constructing the matrix A corresponding to the ith subcarrier in the data packet combination * (ii) a Similarly, a matrix B corresponding to the jth subcarrier in the data packet combination is constructed * (ii) a Then entering the step A4-2-2;
step A4-2-2. Based on matrix A * To in order to
Figure FDA0003916399720000044
Constructing matrix A * The corresponding center distance matrix a, wherein,
Figure FDA0003916399720000045
representation matrix A * The average value of the x-th row in (1),
Figure FDA0003916399720000046
representation matrix A * The mean value of the y-th column in (c),
Figure FDA0003916399720000047
representation matrix A * The overall mean of (a); constructing matrix B by the same method * The corresponding center distance matrix B is entered into the step A4-2-3;
step A4-2-3, according to the following formula:
Figure FDA0003916399720000048
calculating to obtain the distance correlation coefficient between the ith subcarrier and the jth subcarrier of the personnel reference signal receiver correspondingly positioned in the subcarrier combination
Figure FDA0003916399720000049
Wherein,
Figure FDA00039163997200000410
represents the distance covariance between the ith and jth subcarriers,
Figure FDA0003916399720000051
Figure FDA0003916399720000052
represents the distance variance of the ith subcarrier;
step A4-3, according to the distance correlation coefficient corresponding to each subcarrier combination respectively corresponding to each personnel reference signal receiver
Figure FDA0003916399720000053
Constructing inter-carrier amplitude distance correlation coefficient characteristics dCorH corresponding to the signal transmitter and each personnel reference signal receiver respectively rz The following were used:
Figure FDA0003916399720000054
3. the plant personnel supervision method based on multi-feature channel state information edge calculation according to claim 1, characterized by: the step B comprises the following steps B1 to B4;
b1, collecting amplitude entropy characteristics div | H corresponding to the reference signal receivers of each person according to the preset period duration and the method of the step A periodically based on the fact that each person randomly dissociates in the target plant area z | t Phase difference characteristic dif & lt H z t And corresponding inter-carrier amplitude distance correlation coefficient characteristic dCorH between personnel reference signal receiver and signal transmitter rz t Meanwhile, fluctuation information of the sum of the numerical values of the pressure sensors in the two work boots respectively worn by each person under the condition of corresponding cycle duration is periodically acquired
Figure FDA0003916399720000055
Obtaining the system state s at the acquisition time t of each period t The following were used:
Figure FDA0003916399720000056
wherein,
Figure FDA0003916399720000057
or a combination of the values of 0,
Figure FDA0003916399720000058
the fluctuation range representing the sum of the values of the pressure sensors in the two work shoes of the z-th individual member at the periodic acquisition time t exceeds the preset pressure threshold range,
Figure FDA0003916399720000059
representing the z-th person at the periodic acquisition instant tThe fluctuation range of the sum of the numerical values of the pressure sensors in the two working shoes does not exceed the range of a preset pressure threshold value; then entering step B2;
b2, respectively aiming at each person and system state s at each period acquisition time t t If the person is corresponding to the continuous collection time of each period
Figure FDA00039163997200000510
The system state s at the first time of the acquisition time of each continuous period is reserved t The record corresponding to the person in (1) deletes the system state s at the rest of the acquisition moments in each continuous period t The record corresponding to the personnel realizes the system state s under the acquisition time t of each period t Then step B3 is entered;
step B3, respectively aiming at the system state s under each period acquisition time t t Obtaining a system state s t The combination of the persons involved in (a) and the working positions respectively constitutes the system state s t The corresponding personnel position set is then entered into step B4;
step B4. is based on system state s at each cycle acquisition time t t Corresponding personnel position set, combined with fingerprint library, and reward function r under ordered pairs of which each personnel corresponds to a working position combination t The following were used:
Figure FDA0003916399720000061
training the appointed depth reinforcement learning algorithm model until convergence, and obtaining amplitude entropy characteristic div | H corresponding to the personnel reference signal receiver z | t Phase difference characteristic dif & lt H z t And corresponding inter-carrier amplitude distance correlation coefficient characteristic dCorH between personnel reference signal receiver and signal transmitter rz t For input, the working position of the personnel is an output personnel working position judgment model; wherein d (div | H) z | t ,div|H z I) tableDisplay matrix div | H z | t And div | H z The euclidean distance between | s.
4. The method for supervising plant personnel based on multi-feature channel state information edge calculation recited in claim 3, wherein: the preset pressure threshold range is 90N to 120N.
5. The plant personnel supervision method based on multi-feature channel state information edge calculation according to claim 1, characterized by: and (e) acquiring the working position of the person to be monitored based on the execution of the step i, if the person to be monitored is not at any working position, judging that the person to be monitored is in a rest state at the moment, further carrying out periodic detection on the person to be monitored according to the step i within the working time of one day, counting the rest time of the person to be monitored, judging whether the rest time exceeds the preset rest time range, if so, judging that the person to be monitored does not reach the workload requirement of one day, and if not, judging that the person to be monitored reaches the workload requirement of one day.
6. The plant personnel supervision method based on multi-feature channel state information edge calculation according to any one of claims 1 to 5, characterized by: the working boots are characterized in that each person wears a working suit with an embedded receiving antenna, the receiving antenna is embedded into the tail part of the back of the working suit, and the pressure sensor arranged in each working boot is a FlexPresure pressure sensor.
CN202211075973.2A 2022-09-05 2022-09-05 Factory personnel supervision method based on multi-feature channel state information edge calculation Active CN115166636B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211075973.2A CN115166636B (en) 2022-09-05 2022-09-05 Factory personnel supervision method based on multi-feature channel state information edge calculation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211075973.2A CN115166636B (en) 2022-09-05 2022-09-05 Factory personnel supervision method based on multi-feature channel state information edge calculation

Publications (2)

Publication Number Publication Date
CN115166636A CN115166636A (en) 2022-10-11
CN115166636B true CN115166636B (en) 2022-12-20

Family

ID=83482023

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211075973.2A Active CN115166636B (en) 2022-09-05 2022-09-05 Factory personnel supervision method based on multi-feature channel state information edge calculation

Country Status (1)

Country Link
CN (1) CN115166636B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104267439A (en) * 2014-08-20 2015-01-07 哈尔滨工程大学 Unsupervised human detecting and positioning method
WO2018119949A1 (en) * 2016-12-29 2018-07-05 深圳天珑无线科技有限公司 Channel state information phase correction method and apparatus
CN108924736A (en) * 2018-06-14 2018-11-30 西北师范大学 A kind of passive indoor occupant condition detection method based on PCA-Kalman
CN109640269A (en) * 2018-12-18 2019-04-16 上海大学 Fingerprint positioning method based on CSI Yu Time Domain Fusion algorithm
CN111918388A (en) * 2020-08-17 2020-11-10 南京邮电大学 CSI fingerprint passive positioning method based on depth separable convolution
CN112153736A (en) * 2020-09-14 2020-12-29 南京邮电大学 Personnel action identification and position estimation method based on channel state information
WO2021160189A1 (en) * 2020-02-14 2021-08-19 重庆邮电大学 Csi method for recognizing human fall in wi-fi interference environment

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2020170221A1 (en) * 2019-02-22 2020-08-27 Aerial Technologies Inc. Handling concept drift in wi-fi-based localization

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104267439A (en) * 2014-08-20 2015-01-07 哈尔滨工程大学 Unsupervised human detecting and positioning method
WO2018119949A1 (en) * 2016-12-29 2018-07-05 深圳天珑无线科技有限公司 Channel state information phase correction method and apparatus
CN108924736A (en) * 2018-06-14 2018-11-30 西北师范大学 A kind of passive indoor occupant condition detection method based on PCA-Kalman
CN109640269A (en) * 2018-12-18 2019-04-16 上海大学 Fingerprint positioning method based on CSI Yu Time Domain Fusion algorithm
WO2021160189A1 (en) * 2020-02-14 2021-08-19 重庆邮电大学 Csi method for recognizing human fall in wi-fi interference environment
CN111918388A (en) * 2020-08-17 2020-11-10 南京邮电大学 CSI fingerprint passive positioning method based on depth separable convolution
CN112153736A (en) * 2020-09-14 2020-12-29 南京邮电大学 Personnel action identification and position estimation method based on channel state information

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
Deep Q-Network-Aided Adaptive mmWave Multi-User NOMA Selection and Detection;K.Satyanarayana等;《ICC 2021 - IEEE International Conference on Communications》;20210806;第1-6页 *
Digital Twin Assisted Task Offloading for Aerial Edge Computing and Networks;Bin Li等;《 IEEE Transactions on Vehicular Technology》;20220614;第1-16页 *
The Wireless IoT Device Identification based on Channel State Information Fingerprinting;Rui Liu等;《2020 IEEE 9th Joint International Information Technology and Artificial Intelligence Conference》;20210203;第534-541页 *
基于CSI相位矫正的室内指纹定位技术研究;刘兆岩等;《无线电工程》;20200205(第02期);第22-27页 *
基于信道状态信息的无源被动定位;吴哲夫等;《传感技术学报》;20150515(第05期);第69-75页 *
基于无线信道状态相位信息优化的定位算法;周明快等;《传感技术学报》;20180615(第06期);第141-146页 *

Also Published As

Publication number Publication date
CN115166636A (en) 2022-10-11

Similar Documents

Publication Publication Date Title
CN108764059B (en) Human behavior recognition method and system based on neural network
CN101516099B (en) Test method for sensor network anomaly
CN110929934A (en) Equipment failure prediction method and device, computer equipment and storage medium
CN105139029B (en) A kind of Activity recognition method and device of prison prisoner
CN106228178A (en) Networks congestion control prognoses system
WO2021043126A1 (en) System and method for event recognition
Harb et al. En-route data filtering technique for maximizing wireless sensor network lifetime
CN107257351A (en) One kind is based on grey LOF Traffic anomaly detections system and its detection method
Kharitonov et al. Comparative analysis of machine learning models for anomaly detection in manufacturing
CN111148142A (en) Dormant cell detection method based on anomaly detection and integrated learning in mobile communication network
CN114705177B (en) Fiber-optic gyroscope attitude measurement data processing method based on error analysis
CN103345552A (en) Method and device for assessing reliability of power ICT communication network
CN111044974B (en) Indoor positioning method and device based on WiFi signal and storage medium
CN109921938A (en) Fault detection method under a kind of cloud computing environment
CN110930541B (en) Method for analyzing working condition state of agricultural machine by using GPS information
CN113571133A (en) Lactic acid bacteria antibacterial peptide prediction method based on graph neural network
CN101237357B (en) Online failure detection method for industrial wireless sensor network
CN117235647A (en) Mineral resource investigation business HSE data management method based on edge calculation
CN115166636B (en) Factory personnel supervision method based on multi-feature channel state information edge calculation
CN110098944A (en) A method of protocol data flow is predicted based on FP-Growth and RNN
JP5401885B2 (en) Model construction method, construction system, and construction program
CN116192520A (en) Secure communication management method and system based on big data
CN113328881B (en) Topology sensing method, device and system for non-cooperative wireless network
CN109600378A (en) The heterogeneous sensor network accident detection method of non-stop layer node
Scheffel et al. WSN data confidence attribution using predictors

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