CN105640573A - Electric power maintainer safety risk monitoring system based on Google glasses - Google Patents

Abstract

The invention relates to an electric power maintainer safety risk monitoring system based on Google glasses. The electric power maintainer safety risk monitoring system comprises a pair of Google glasses, a wearable pulse and body-temperature sensor, a portable intelligent tablet computer, an individual physiological feature data collection and modeling system and an on-site work state monitoring system, wherein the Google glasses can detect blink frequency, blink time and eye-closing total-continuing time of the maintainer; the wearable pulse and body-temperature sensor can measure body temperature and pulse of the maintainer; the individual physiological feature data collection and modeling system can build an abnormal-work state monitoring model for each maintainer during a practical training phase of the maintainer based on the collection data; and the on-site work state monitoring system can give early warning for abnormal states of tension, panic and fatigue of the maintainer on the on-site based on the work state monitoring model. Monitoring level for dangerous work state of electric power maintainers can be improved, so maintenance safety risk can be reduced.

Description

A kind of electric power overhaul personal security risk monitoring and control system based on Google's glasses

Technical field

The present invention relates to a kind of supervisory system towards electric power overhaul personal security risk

Background technology

Electric power repair security risk refers to the risk occurred in electric power repair. From forming, mainly comprise personnel safety risk, power grid security risk and device security risk. People is factor active, the most positive in service work. Above security risk will be formed by the working order of maintenance personnel directly to be affected: if maintenance personnel state is good, it is possible to reduce the security risk of electrical network and equipment; If under maintenance personnel are in the state such as fatigue, anxiety, not only can increase the security risk of self, also can increase the security risk of electrical network, equipment. Therefore, how the working order of maintenance personnel is continued to monitor, and carry out early warning when potentially dangerous occurs, be one of important channel reducing electric power repair security risk.

Along with the development of human sensing device technology, by physical signs such as the blood pressure of monitoring people, pulse, body temperature, brain waves, physiology and psychological condition that work is mediated can be monitored at present. With regard to current correlative study, based on brain wave instrument brain wave monitoring should be the most accurately human body psychology with physiological status monitoring means. But realizing from concrete technology, owing to brain wave detection is contact measurement, and test set is heavy, is not suitable at present being used in electric power repair environment. Therefore field is monitored in electric power overhaul person works's state, mainly adopt the physical signs such as detection blood pressure, pulse, body temperature at present, the method of detected person's state is analyzed again by analytical model research and application data, such as China Patent Publication No. CN104013394A, the name of innovation and creation is called " power engineering working aloft personnel's individual state monitoring device ", and this application case discloses the scheme monitoring electric power working aloft personnel's individual state based on pulse and temperature check.

There is a series of problem in the application in such method, mainly show as: 1) index such as blood pressure, pulse is relatively better for the Detection results of the psychological condition such as nervous, panic, but not obvious for state-detection effects such as fatigues, and fatigue is the class major reason causing security risk; 2) the current leading thinking when setting up analytical model sets up a model towards all groups, but the indexs such as blood pressure, pulse, body temperature have bigger individual difference, therefore very difficult in the determination of the off-note threshold value of model, thus cause the early warning mortality height of final system.

For overcoming the above problems, it is to increase to the early warning success ratio of maintenance personnel's error state (ERST), it is necessary to introduce new maintenance person works's status flag detection method. Simultaneously in the foundation of analytical model, it is also desirable to have new thinking to introduce.

Summary of the invention

In order to overcome the shortcoming that above-mentioned prior art exists, it is an object of the invention to provide a kind of electric power overhaul personal security risk monitoring and control system based on Google's glasses: this system utilizes Google's glasses can detect the ability of wearer's eye activity, monitoring maintenance personnel blink the eye activity datas such as frequency, wink time, eye closing total duration, and monitor pulse and body temperature simultaneously; In specific maintenance personnel's practical training project, overhaul the above physiological characteristic data of personnel based on above equipment collection, form personal characteristics pattern database, and set up abnormal operation pattern recognition model; Field monitoring stage feature based pattern database, the real-time characteristic physiological data collected is carried out off-note detection, early warning is carried out when the error state (ERST) such as nervous, panic, tired occur in working process in maintenance personnel, effectively improve the level to maintenance personnel hazard's Working Status Monitoring, reduce maintenance safety risk.

The present invention is by the following technical solutions: a kind of electric power overhaul personal security risk monitoring and control system based on Google's glasses, comprises Google's glasses, Wearable pulse and body temperature sensor, portable intelligent panel computer, individual physiological characteristic data collection and modeling system, work on the spot condition monitoring system.

Described Google glasses, for measuring frequency nictation of maintenance personnel, wink time, eye closing total duration.

Described Wearable pulse and body temperature sensor, comprise pulse transducer module, body temperature sensor, bluetooth communication module, master control unit, for measuring pulse and the body temperature of maintenance personnel. Described pulse transducer adopts photosensitive Cleaning Principle, and schematic circuit diagram are as shown in Figure 1. Described body temperature sensor, schematic circuit diagram are as shown in Figure 2. Described bluetooth communication module, is connected with master control unit, is responsible for sending detection data to portable intelligent panel computer.

Described portable intelligent panel computer has 3G data corresponding function, for being connected with body temperature sensor with Google glasses, Wearable pulse by bluetooth communications protocol at Overhaul site, collect human body feature detection data, and collection result is sent back to the work on the spot condition monitoring system at rear.

Described individual physiological characteristic data collection and modeling system, comprise hardware system and software system, at the individual physiological characteristic data collection phase of maintenance personnel, gathering physiological characteristic data, and set up analytical model based on image data. described hardware system, comprises the computer equipment of Ambulatory EEG figure instrument and operating software system. described software system, comprise data acquisition module and data modeling module, in the specific examination link of training, Google's glasses will be worn for maintenance personnel, Wearable pulse and body temperature sensor and Ambulatory EEG figure instrument, data acquisition module collects the detection data of above equipment, wherein blink frequency, wink time, eye closing total duration, pulse and body temperature are for generating personal characteristics pattern database, brain wave data is used for the working order of assistant analysis trouble hunt ing personnel, based on the method that subjectiveness and objectiveness combines, personal characteristics mode data stores respectively according to normal operation state and abnormal operation, data modeling module, for normal operation status flag data and abnormal work data characteristics data, builds the normal operation stateful pattern recognition model and the abnormal operation pattern recognition model that are applicable to this maintenance personnel respectively, and algorithm steps is as follows:

Step 1: data prediction

For the data set X={x obtained₁, x₂..., x_n, wherein x is for comprising nictation frequency, wink time, eye closing total duration, and 5 dimensional vectors of pulse and body temperature, do normalized on each dimension degree, obtains normalization data collection

Step 2: calculate initial cluster center and initially degree of being subordinate to

Calculate according to the following formulaThe density at place

$D_i^{\left(0\right)}=\underset{j=1}{\overset{n}{\mathrm{\Σ}}}\frac{1}{1+2{\left({\mathrm{SD}}_{\mathrm{ij}}\right)}^2/d^2}$

Wherein SD_ijFor dataWithBetween statistical distance, be defined as follows:

${\mathrm{SD}}_{\mathrm{ij}}=\sqrt{{({\stackrel{\‾}{x}}_i-{\stackrel{\‾}{x}}_j)}^T{\mathrm{\Σ}}^{-1}({\stackrel{\‾}{x}}_i-{\stackrel{\‾}{x}}_j)}$

�� is data setCovariance matrix

D is data setThe average statistical distance of middle data, is defined as follows:

$d=\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}\underset{j=1}{\overset{n}{\mathrm{\Σ}}}{\left({\mathrm{SD}}_{\mathrm{ij}}\right)}^2}{n(n-1)}}$

Then withCorresponding data pointIt is the 1st cluster centre, it is designated as

When calculating the q time iteration according to the following formulaThe density at place

$D_i^{\left(q\right)}=(1-\frac{1}{1+2{\left|\right|{\stackrel{\‾}{x}}_i-v_{q-1}^{\left(0\right)}\left|\right|}^2/d^2})D_i^{(q-1)}$

Then withCorresponding data pointIt is the q cluster centre, it is designated as

WhenDuring 0 < �� < 1, above calculating terminates, gained resultAs initial cluster center.

Data setIn data be U=[u for the subordinated-degree matrix of initial cluster center_iq]_n��Q, calculate initially degree of being subordinate to according to following formula

$u_{\mathrm{iq}}=\left\{\begin{array}{cc}0& {\stackrel{\&OverBar;}{x}}_i\&Element;V^{\left(0\right)}\\ 1& {\stackrel{\&OverBar;}{x}}_i\&NotElement;V^{\left(0\right)}\end{array}\right.$

Step 3: recalculate cluster centre V^(k), method of calculation are as follows

$v_q^{\left(k\right)}=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{u}_{\mathrm{iq}}^m{\stackrel{\&OverBar;}{x}}_i}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{u}_{\mathrm{iq}}^m},q=\mathrm{1,2},...,Q$

Wherein m (m > 1) is fuzzy index

Step 4: calculate degree of being subordinate to

It is calculated as follows relative to the degree of being subordinate to of new cluster centre

$u_{\mathrm{iq}}^{\left(k\right)}=\frac{{\left[\frac{1}{{\left|\right|{\stackrel{\&OverBar;}{x}}_i-v_q^{\left(k\right)}\left|\right|}^2}\right]}^{\frac{1}{m-1}}}{\underset{q=1}{\overset{Q}{\mathrm{\&Sigma;}}}{\left[\frac{1}{{\left|\right|{\stackrel{\&OverBar;}{x}}_i-v_q^{\left(k\right)}\left|\right|}^2}\right]}^{\frac{1}{m-1}}}$

Step 5: ifThen stopping iteration, otherwise make k=k+1, turn to step 3, wherein �� is little positive number given in advance.

Described work on the spot condition monitoring system, maintenance personnel's production status for the production of scene is monitored, comprise working order pattern recognition model and the module that communicates, described communication module receives the individual physiological feature detection data that field portable Intelligent flat computer sends back, described working order pattern recognition model is based on frequency nictation, wink time, eye closing total duration, pulse and body temperature real time data, calculate normal operation state data in itself and individual physiological property data base, degree of the being subordinate to relation of abnormal operation data, when the degree of being subordinate to of real time data and abnormal operation data, higher than during with the degree of being subordinate to of normal operation state data, maintenance personnel's potentially dangerous early warning will be sent to the scene personnel of guarding.

Accompanying drawing explanation

Fig. 1 is pulse detection schematic circuit diagram

Fig. 2 is temperature check schematic circuit diagram

Embodiment

Below in conjunction with accompanying drawing and embodiment, the present invention is described in further detail.

In FIG, that LED/light source adopts is the IR204-A of hundred million photoelectrons, and the SFH203P of the Ou Silang that the photodiode receiving LED light adopts, that operational amplifier adopts is the OPA177 of Texas Instrument, and its Output rusults is sent to the input terminus of micro-chip.

In fig. 2, detection body temperature sensor used is DS18B20, and it exports the input terminus being sent to micro-chip.

For the personal characteristics detection data obtained in the specific examination link of training, adopt method that objective observation combines with subjective observation to judge its data whether belonging to error state (ERST) or standard state. Subjective observation adopts coach's observation to carry out with to the mode overhauling personnel's survey. Maintenance personnel state in real training examination process is carried out record to train subjective point by coach's observation. Survey content, mainly for the state overhauling personnel in examination process, after examination terminates, please the personnel of overhauling will be filled in.

Objective observation, according to the working order of brain wave data trouble hunt ing personnel, mainly gathers the slow �� wave datum relevant especially to tired state, and the �� ripple relevant especially to Stress. The phenomenon that analysis personnel the slow �� ripple of weight analysis based on brain wave analysis software and �� wave energy obviously increases, and compare of analysis subjective observation result, to determine whether maintenance personnel have entered fatigue doze state and Stress. If relevant precarious position is confirmed, the corresponding time period will be extracted by the data of Google's glasses and Wearable pulse and body temperature sensor collection, form abnormal work feature mode database. Other data removed outside off-note data will become normal operation feature mode database.

Claims (3)

1. the electric power overhaul personal security risk monitoring and control system based on Google's glasses, it is characterized in that, comprise Google's glasses, Wearable pulse and body temperature sensor, portable intelligent panel computer, individual physiological characteristic data collection and modeling system, work on the spot condition monitoring system

Described Google glasses, for measuring frequency nictation of maintenance personnel, wink time, eye closing total duration;

Described Wearable pulse and body temperature sensor, comprise pulse transducer module, body temperature sensor assembly, bluetooth communication module, master control unit, pulse transducer module adopts photosensitive pulse detection principle, detection maintenance personnel's pulse, body temperature sensor assembly, detection maintenance personnel's body temperature, bluetooth communication module, it is connected with master control unit, sends detection data to portable intelligent panel computer;

Described portable intelligent panel computer has 3G data corresponding function, it is connected with body temperature sensor with Google glasses, Wearable pulse by bluetooth communications protocol at Overhaul site, collect human body feature detection data, and collection result is sent back to the work on the spot condition monitoring system at rear;

Described individual physiological characteristic data collection and modeling system, comprise hardware system and software system, at the individual physiological characteristic data collection phase of maintenance personnel, gather physiological characteristic data and form individual physiological property data base, and set up working order pattern recognition model based on this database;

Described work on the spot condition monitoring system, comprise working order pattern recognition model and the module that communicates, for the production of the maintenance personnel state monitoring at scene, the module that wherein communicates receives the individual physiological feature detection data that field portable Intelligent flat computer sends back, working order pattern recognition model is based on frequency nictation, wink time, eye closing total duration, pulse and body temperature real time data, calculate normal operation state data in itself and individual physiological property data base, degree of the being subordinate to relation of abnormal operation data, when the degree of being subordinate to of real time data and abnormal operation data, higher than during with the degree of being subordinate to of normal operation state data, the personnel that guard to scene send maintenance personnel's potentially dangerous early warning.

2. the hardware system of individual physiological characteristic data collection according to claim 1 and modeling system, is characterized in that, comprise the computer equipment of Ambulatory EEG figure instrument and operating software system.

3. the software system of individual physiological characteristic data collection according to claim 1 and modeling system, it is characterized in that, comprise data acquisition module and data modeling module, in the specific examination link of training, Google's glasses will be worn for maintenance personnel, Wearable pulse and body temperature sensor and Ambulatory EEG figure instrument, data acquisition module collects the detection data of above equipment, wherein blink frequency, wink time, eye closing total duration, pulse and body temperature are for generating personal characteristics pattern database, brain wave data is used for the working order of assistant analysis trouble hunt ing personnel, based on the method that subjectiveness and objectiveness combines, personal characteristics mode data stores respectively according to normal operation state and abnormal operation, data modeling module is for normal operation status flag data and abnormal work data characteristics data, build the normal operation stateful pattern recognition model and the abnormal operation pattern recognition model that are applicable to this maintenance personnel respectively, algorithm steps is as follows:

Step 1: data prediction

For the data set X={x obtained₁, x₂..., x_n, wherein x is for comprising nictation frequency, wink time, eye closing total duration, and 5 dimensional vectors of pulse and body temperature, do normalized on each dimension degree, obtains normalization data collection

Step 2: calculate initial cluster center and initially degree of being subordinate to

Calculate according to the following formulaThe density at place

$D_i^{\left(0\right)}=\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}\frac{1}{1+2{\left({\mathrm{SD}}_{\mathrm{ij}}\right)}^2/d^2}$

Wherein SD_ijFor dataWithBetween statistical distance, be defined as follows:

${\mathrm{SD}}_{\mathrm{ij}}=\sqrt{{({\stackrel{\&OverBar;}{x}}_i-{\stackrel{\&OverBar;}{x}}_j)}^T{\mathrm{\&Sigma;}}^{-1}({\stackrel{\&OverBar;}{x}}_i-{\stackrel{\&OverBar;}{x}}_j)}$

�� is data setCovariance matrix

D is data setThe average statistical distance of middle data, is defined as follows:

$d=\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\left({\mathrm{SD}}_{\mathrm{ij}}\right)}^2}{n(n-1)}}$

Then with $D_{\max }^{\left(0\right)}=\max \{D_i^{\left(0\right)},i=\mathrm{1,2},...,n\}$ Corresponding data pointIt is the 1st cluster centre, it is designated as

When calculating the q time iteration according to the following formulaThe density at place

$D_i^{\left(q\right)}=(1-\frac{1}{1+2{\left|\right|{\stackrel{\&OverBar;}{x}}_i-v_{q-1}^{\left(0\right)}\left|\right|}^2/d^2})D_i^{(q-1)}$

Then with $D_{\max }^{\left(q\right)}=\max \{D_i^{\left(q\right)},i=\mathrm{1,2},...,n-q-1\}$ Corresponding data pointIt is the q cluster centre, it is designated as

When $D_{Q+1}<\mathrm{\&alpha;}{D}_{\max }^{\left(0\right)},$ During 0 < �� < 1, above calculating terminates, gained result $V^{\left(0\right)}=\{v_1^{\left(0\right)},v_2^{\left(0\right)},...,v_Q^{\left(0\right)}\}$ As initial cluster center.

Data setIn data be U=[u for the subordinated-degree matrix of initial cluster center_iq]_n��Q, calculate initially degree of being subordinate to according to following formula

$u_{\mathrm{iq}}=\left\{\begin{array}{cc}0& {\stackrel{\&OverBar;}{x}}_i\&Element;V^{\left(0\right)}\\ 1& {\stackrel{\&OverBar;}{x}}_i\&NotElement;V^{\left(0\right)}\end{array}\right.$

Step 3: recalculate cluster centre V^(k), method of calculation are as follows

$v_q^{\left(k\right)}=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{u}_{\mathrm{iq}}^m{\stackrel{\&OverBar;}{x}}_i}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{u}_{\mathrm{iq}}^m},q=\mathrm{1,2},...,Q$

Wherein m (m > 1) is fuzzy index

Step 4: calculate degree of being subordinate to

It is calculated as follows relative to the degree of being subordinate to of new cluster centre

$u_{\mathrm{iq}}^{\left(k\right)}=\frac{{\left[\frac{1}{{\left|\right|{\stackrel{\&OverBar;}{x}}_i-v_q^{\left(k\right)}\left|\right|}^2}\right]}^{\frac{1}{m-1}}}{\underset{q=1}{\overset{Q}{\mathrm{\&Sigma;}}}{\left[\frac{1}{{\left|\right|{\stackrel{\&OverBar;}{x}}_i-v_q^{\left(k\right)}\left|\right|}^2}\right]}^{\frac{1}{m-1}}}$

Step 5: ifThen stopping iteration, otherwise make k=k+1, turn to step 3, wherein �� is little positive number given in advance.