CN103838210A - Emergency scene poisonous gas remote wireless monitoring system and method - Google Patents

Abstract

The invention provides an emergency scene poisonous gas remote wireless monitoring system which comprises at least one poisonous gas wireless detection node, at least one information aggregation transferring node and at least one portable intelligent handheld terminal. The poisonous gas wireless detection nodes are used for detecting first poisonous gas information of an emergency scene, the information aggregation transferring nodes are used for aggregating second poisonous gas information obtained through the poisonous gas wireless detection nodes and transmitting the aggregating result to the portable intelligent handheld terminals, the portable intelligent handheld terminals comprise third microcontroller modules, and the third microcontroller modules are used for fusing the second poisonous gas information according to classes of poisonous gas to obtain third poisonous gas information. The aims that poisonous gas is rapidly detected in the emergency scene and corresponding and effective measures are taken are accordingly achieved.

Description

The on-the-spot toxic gas remote wireless monitoring system of accident and method

Technical field

The application relates to Detection Technology and Automatic Equipment field, particularly relates to the on-the-spot toxic gas remote wireless monitoring system of accident and method.

Background technology

In colliery, the field such as oil, chemical industry, fire-fighting, nuclear energy, military affairs, due to reasons such as misoperation, rugged surroundings, ageing equipments, easily there is the leakage of poisonous or inflammable gas.Once inflammable gas occurs to be revealed, very easily cause large area fire and cause an explosion accident, and the very big impact that toxic gas leakage event usually causes health, for example: historic India Bhopal toxic gas leakage case can be rated as disaster, there is the Leakage Gas such as chlorine, sulfuretted hydrogen and the event that causes casualties does not also occupy the minority in domestication factory.

In recent years, people give increasing concern for sudden toxic gas leakage, terrorism wildness in world wide simultaneously, poison gas (virus) also may be served as chemistry (biochemistry) weapon that terrorist uses, so the research and development of detection of poison gas technology are imperative.Reliable fashion is all being explored as possible to ensure national people's property life security by many countries of the world today, farthest reduces the threat that accident scene or terrorist incident cause the mankind.

For the measurement that occurs in the toxic gas in above-mentioned condition, the main methods such as analytical chemistry method, spectroscopic analysis methods, portable gas chromatography-mass spectrometer that adopt at present.Wherein, analytical chemistry method generally gathers sample gas in measure field, then takes back laboratory and carries out chemical analysis and obtain result, and the method is subject to the restriction of condition, cannot detect fast toxic gas; Spectroscopic analysis methods generally needs special spectrometer, apparatus expensive, complicated operation, not Portable belt, and sampling analysis speed slow, cannot realize on-line monitoring; Portable gas chromatography-mass spectrometric method detects toxic gas length consuming time, and investigative range is little.

Leak or be subject to the scene of biological attack for toxic gas, while stating existing method in the use, find, related personnel cannot detect fast to the kind of the toxic gas at accident scene, concentration, also cannot take effective counter-measure for the situation at concrete accident scene simultaneously, make the property of country and enterprise be subject to huge loss, the masses' life security is also subject to significant damage.

Summary of the invention

The application provides the on-the-spot toxic gas remote wireless monitoring system of a kind of accident and method, to solve in accident field quick detection to go out toxic gas and take the problem of corresponding effective counter-measure.

In order to address the above problem, the application discloses the on-the-spot toxic gas remote wireless monitoring system of a kind of accident, comprising: at least one toxic gas radio detection node, at least one information converge transit node and at least one portable intelligent handheld terminal; Wherein, described toxic gas radio detection node, for detection of the first toxic gas information at accident scene, adopts variety of way that described toxic gas radio detection node is sent to on-the-spot the first toxic gas information area of accident;

Described toxic gas radio detection node includes poisonous gas sensor array module and the first micro controller module;

Described toxic gas sensor array module, for detection of the first toxic gas information, comprises one-parameter sensor array module and multi-parameter sensor array module;

The first toxic gas information that described the first micro controller module detects for analyzing described toxic gas sensor array module; Described the first micro controller module comprises cluster module and artificial neural network identification module;

Described cluster module is carried out cluster for the first toxic gas information that described toxic gas sensor array module is detected, and obtains cluster result;

Described artificial neural network identification module is used for, and utilizes the toxic gas model of cognition building based on artificial neural network to identify described cluster result, obtains the second toxic gas information;

Information converges transit node, for the second toxic gas information of each toxic gas radio detection node is converged, and is sent to described portable intelligent handheld terminal;

Described portable intelligent handheld terminal comprises the 3rd micro controller module and effusiometer calculation module, and described the 3rd micro controller module, for by described the second toxic gas information, merges by classification under toxic gas, obtains the 3rd toxic gas information; Utilize effusiometer to calculate module and draw fast speed and the scope of on-the-spot the first toxic gas diffusion of information.

Preferably, described toxic gas radio detection node also includes poisonous gas adsorption module, the first locating module and the first wireless communication module; The first toxic gas information that described toxic gas adsorption module detects for adsorbing toxic gas sensor array module; Described wireless communication module converges transit node for the first toxic gas information of the detection of sensor array module is sent to information; Described the first micro controller module is connected with the first locating module; Described the first locating module is for obtaining the locating information of toxic gas radio detection node; Described toxic gas radio detection node also comprises the first locating module being connected with toxic gas sensor array module, toxic gas adsorption module, the first micro controller module, described the first micro controller module and the first power module being connected with the first wireless communication module; Described the first power module is for providing power supply to the modules of toxic gas radio detection node;

Further, described system also comprises: off-line analysis module, for described the first toxic gas information is carried out to off-line analysis, obtains off-line analysis result;

Further, also comprise: optimize module, for utilizing the Output rusults of described off-line analysis modified result artificial neural network identification module.

Preferably, described cluster module comprises:

The first cluster module, be used for adopting Fuzzy c-means clustering algorithm KFCM that the first toxic gas information exchange is crossed to Function Mapping and obtain initial value to carrying out again fuzzy C-means clustering FCM after high-dimensional feature space, recycling particle group optimizing PSO algorithm is optimized iterative computation to obtain best cluster result to described initial value, and the cluster majorized function of its use comprises:

$f\left(x_i\right)=\frac{1}{J_{\mathrm{\α}}(U,V)+1}=\frac{1}{2\underset{i=1}{\overset{C}{\mathrm{\Σ}}}\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\μ}}_{\mathrm{ij}}^m[1-K(x_i,v_j)]+1}---\left(1\right)$

Wherein, C is cluster number, and N is sample number, U=[μ _ij] _{c × N}fuzzy C Matrix dividing, μ _ijfor sample x _icorresponding to the degree of membership value of j cluster, V=[v _j] be the set of C cluster centre composition, m is the index weight that affects degree of membership matrix norm gelatinization degree, J _α(U, V) is the cost function of FCM, K (x _i, v _j) be gaussian kernel function, wherein x _ifor m dimensional vector, v _jwith x _ithat the more new formula of cluster centre and degree of membership is as follows with the vector of dimension:

$v_j=\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\μ}}_{\mathrm{ij}}^{m}K(x_i,v_j)x_i}{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\μ}}_{\mathrm{ij}}^{m}K(x_i,v_j)}---\left(2\right)$

${\mathrm{\μ}}_{\mathrm{ij}}=\frac{{(1-K(x_i,v_j))}^{-1/(m-1)}}{\underset{j=1}{\overset{C}{\mathrm{\Σ}}}{(1-K(x_i,v_j))}^{-1/(m-1)}}---\left(3\right)$

In each iteration, each particle upgrades position and the speed of self by following the tracks of individual extreme value and global extremum, and its computing formula is as follows:

V _iα(t+1)＝ωV _iα(t)+c ₁r ₁[P _iα(t)-X _iα(t)]+c ₂r ₂[P _fα(t)-X _iα(t)] （4）

X _iα(t+1)＝X _iα(t)+V _iα(t+1) （5）

In formula, V _{i α}(t) be the speed of particle previous moment, V _{i α}(t+1) be the current speed of particle, X _{i α}(t) be the current location of particle, X _{i α}(t+1) be the reposition that particle produces, P _{i α}(t) be individual extreme value, P _{f α}(t) be global extremum, c ₁, c ₂for the study factor (for being greater than zero positive integer), r ₁, r ₂for the random number between [0,1], i=1,2 ..., N, ω is inertial factor.

Preferably, described portable intelligent handheld terminal also comprises:

Wake control module up, be suitable for the distribution situation according to described locating information, portable intelligent handheld terminal wakes toxic gas radio detection node for subsequent use that can guarded region up;

Described portable intelligent handheld terminal also comprises the 3rd wireless communication module, the 3rd micro controller module, the 3rd supply module and shows load module; Described the 3rd micro controller module is used for controlling the 3rd wireless communication module and load module; Described the 3rd wireless communication module converges for receiving described information the second toxic gas information that transit node transmits, and by the 3rd micro controller module, the second toxic gas information is carried out to Data Fusion, the 3rd toxic gas information after Data Fusion is sent to described demonstration load module; Described demonstration load module is for duty and the relevant data of displaying scene node; Described the 3rd supply module is with the 3rd wireless communication module, the 3rd micro controller module and show that load module is connected; Described the 3rd supply module is for providing power supply to the modules of portable intelligent handheld terminal.

Preferably, described information converges transit node and comprises intra-node Temperature Humidity Sensor module, the second micro controller module and the second wireless communication module; Described the second wireless communication module is for receiving the second toxic gas information that toxic gas radio detection node obtains; Described intra-node Temperature Humidity Sensor module integration converges on transit node in described information, converges the temperature and humidity in transit node for detection of described information; Described the second wireless communication module is for being sent to portable intelligent handheld terminal by locating information, temperature and humidity information; Described information converges transit node and also comprises the second locating module being connected with described the second micro controller module; Described the second locating module converges the locating information of transit node for obtaining described information; Described information converges transit node and also comprises the second locating module being connected with intra-node Temperature Humidity Sensor module, the second wireless communication module, the second micro controller module, described the second micro controller module and the second supply module being connected with described the second wireless communication module; Described the second supply module provides power supply for the modules that converges transit node to described information.

Preferably, described artificial neural network identification module comprises: N artificial neural network recognin module, described artificial neural network recognin module is for identifying the cluster result of corresponding classification.

Preferably, also comprise:

Fusion Module, for described the 3rd toxic gas information is merged, realize fusion treatment by following formula:

$\left\{\begin{array}{c}\stackrel{\‾}{x}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i{z}_i\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i=1(0\≤w_i\≤1)\end{array}\right.---\left(6\right)$

Wherein, x is true value, z _ifor the measured value of cluster result, w _ifor the weighting factor of cluster result,

for the data after data fusion;

Wherein, be the bias free that ensures to estimate,

, its error variance expression formula is

${\mathrm{\σ}}^2=E\left[{(x-\stackrel{\‾}{x})}^2\right]=E\left[\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i^2{(x-\stackrel{\‾}{x})}^2\right]=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i^2{\mathrm{\σ}}_i^2---\left(7\right)$

Wherein, weighting factor w _i, $w_i=\frac{{\mathrm{\σ}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^{-2}}---\left(8\right)$

(7) and (8) substitutions (6) are obtained to data fusion result is

$\stackrel{\‾}{x}=\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{z}_i{\mathrm{\σ}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^{-2}}---\left(9\right)$

In order to address the above problem, the application discloses the on-the-spot toxic gas long-range wireless monitoring method of a kind of accident, comprising:

Toxic gas radio detection node is sent to accident scene;

Utilize the first toxic gas information of the on-the-spot toxic gas radio detection of the toxic gas sensor array module collection accident node in toxic gas radio detection node;

Utilize the cluster module in toxic gas radio detection node that the first toxic gas information collecting is carried out to cluster, obtain cluster result; Adopt the artificial neural network identification module in toxic gas radio detection node, utilize the toxic gas model of cognition building based on artificial neural network that the cluster result obtaining is identified, and obtain the second toxic gas information;

Utilize information to converge transit node described the second toxic gas information is converged, and be sent to portable intelligent handheld terminal;

Utilize the 3rd micro controller module in portable intelligent handheld terminal by the second toxic gas information, merge by classification under toxic gas, obtain the 3rd toxic gas information, and utilize effusiometer calculation module to determine speed and the scope of on-the-spot the first toxic gas diffusion of information.

Preferably, described method, also comprise that adopting Fuzzy c-means clustering algorithm KFCM that the first toxic gas information exchange is crossed to Function Mapping obtains initial value to carrying out FCM cluster after high-dimensional feature space again, recycling particle group optimizing PSO algorithm is optimized iterative computation to obtain best cluster result to described initial value, its use cluster majorized function comprise:

$f\left(x_i\right)=\frac{1}{J_{\mathrm{\α}}(U,V)+1}=\frac{1}{2\underset{i=1}{\overset{C}{\mathrm{\Σ}}}\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\μ}}_{\mathrm{ij}}^m[1-K(x_i,v_j)]+1}---\left(1\right)$

Wherein, C is cluster number, and N is sample number, U=[μ _ij] _{c × N}fuzzy C Matrix dividing, μ _ijfor sample x _icorresponding to the degree of membership value of j cluster, V=[v _j] be the set of C cluster centre composition, m is the index weight that affects degree of membership matrix norm gelatinization degree, J _α(U, V) is the cost function of FCM, K (x _i, v _j) be gaussian kernel function, wherein x _ifor m dimensional vector, v _jwith x _ithat the more new formula of cluster centre and degree of membership is as follows with the vector of dimension:

$v_j=\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\μ}}_{\mathrm{ij}}^{m}K(x_i,v_j)x_i}{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\μ}}_{\mathrm{ij}}^{m}K(x_i,v_j)}---\left(2\right)$

${\mathrm{\μ}}_{\mathrm{ij}}=\frac{{(1-K(x_i,v_j))}^{-1/(m-1)}}{\underset{j=1}{\overset{C}{\mathrm{\Σ}}}{(1-K(x_i,v_j))}^{-1/(m-1)}}---\left(3\right)$

In each iteration, each particle upgrades position and the speed of self by following the tracks of individual extreme value and global extremum, and its computing formula is as follows:

V _iα(t+1)＝ωV _iα(t)+c ₁r ₁[P _iα(t)-X _iα(t)]+c ₂r ₂[P _fα(t)-X _iα(t)] （ ⁴）

X _iα(t+1)＝X _iα(t)+V _iα(t+1) （5）

In formula, V _{i α}(t) be the speed of particle previous moment, V _{i α}(t+1) be the current speed of particle, X _{i α}(t) be the current location of particle, X _{i α}(t+1) be the reposition that particle produces, P _{i α}(t) be individual extreme value, P _{f α}(t) be global extremum, c ₁, c ₂for the study factor (for being greater than zero positive integer), r ₁, r ₂for the random number between [0,1], i=1,2 ..., N, ω is inertial factor.

Preferably, also comprise the second toxic gas information of utilizing the toxic gas model of cognition building based on artificial neural network to obtain carried out to data fusion, realize Data Fusion by following formula:

$\left\{\begin{array}{c}\stackrel{\‾}{x}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i{z}_i\\ \underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i=1(0\≤w_i\≤1)\end{array}\right.---\left(6\right)$

Wherein, x is true value, z _ifor the measured value of cluster result, w _ifor the weighting factor of cluster result, x is the data after data fusion;

Wherein, be the bias free that ensures to estimate,

, its error variance expression formula is:

${\mathrm{\σ}}^2=E\left[{(x-\stackrel{\‾}{x})}^2\right]=E\left[\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i^2{(x-\stackrel{\‾}{x})}^2\right]=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{w}_i^2{\mathrm{\σ}}_i^2---\left(7\right)$

Wherein, weighting factor w _i, $w_i=\frac{{\mathrm{\σ}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^{-2}}---\left(8\right)$

(7) and (8) substitutions (6) are obtained to data fusion result is:

$\stackrel{\‾}{x}=\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{z}_i{\mathrm{\σ}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{\mathrm{\σ}}_i^{-2}}---\left(9\right)$

Compared with prior art, the application comprises following advantage:

First, the application can be sent to accident scene by toxic gas radio detection node by transmitter or aircraft or robot, detects fast the first toxic gas information at accident scene by the one-parameter sensor array module in toxic gas radio detection node and multi-parameter sensor array module.

Secondly, in the application, be combined with by cluster module and artificial neural network identification module, can effectively determine type and the concentration of the second toxic gas, and by experiment with rig-site utilization Data correction Network Recognition module, to improve constantly the accuracy of detection of on-the-spot toxic gas.

Again, portable intelligent handheld terminal in the application can monitor in real time the duty of the on-the-spot toxic gas radio detection of accident node and wake up can guarded region toxic gas radio detection node for subsequent use, and provide toxic gas in time with the diffusion concentration field of spatial variations.

Brief description of the drawings

Fig. 1 is the on-the-spot toxic gas remote wireless monitoring system of accident structural drawing described in the embodiment of the present application;

Fig. 1-A shows the on-the-spot described toxic gas radio detection node section structural representation of accident described in the embodiment of the present application;

Fig. 1-B shows the on-the-spot described portable intelligent handheld terminal part-structure schematic diagram of accident described in the embodiment of the present application;

Fig. 2 is the transmission mode schematic diagram of the on-the-spot toxic gas radio detection of accident node described in the embodiment of the present application;

Fig. 3 is the structural drawing of the on-the-spot toxic gas radio detection of accident node described in the embodiment of the present application;

Fig. 4 is the structural drawing that accident field data converges transit node described in the embodiment of the present application;

Fig. 5 is the structural drawing of the on-the-spot portable intelligent handheld terminal of accident described in the embodiment of the present application;

Fig. 6 is the online test method of the on-the-spot toxic gas of accident described in the embodiment of the present application;

Fig. 7 is the on-the-spot toxic gas detection of accident and recognition principle described in the embodiment of the present application;

Fig. 8 is the information transmission mode of the on-the-spot toxic gas remote wireless monitoring system of accident (in the time breaking down) described in the embodiment of the present application.

Embodiment

For the above-mentioned purpose, the feature and advantage that make the application can become apparent more, below in conjunction with the drawings and specific embodiments, the application is described in further detail.

With reference to Fig. 1, show the on-the-spot toxic gas remote wireless monitoring system of accident structural drawing described in the embodiment of the present application.Wherein, comprising: at least one toxic gas radio detection node (A101-A10n), at least one information converge transit node (A201-A20m) and at least one portable intelligent handheld terminal (A301-A30p).Wherein, described toxic gas radio detection node, for detection of the first toxic gas information at accident scene, adopts variety of way described toxic gas radio detection node to be sent to the first toxic gas information area at accident scene; Information converges transit node, converges, and be sent to described portable intelligent handheld terminal for the second toxic gas information that each toxic gas radio detection node is obtained.

The first toxic gas information comprise common are that poisonous gas and chemical and biological weapons discharge contain lifetime killer as the gas of bacterium, virus, toxin.Wherein common are poisonous gas kind more, as: the gases such as carbon monoxide, nitrogen monoxide, sulfuretted hydrogen, sulphuric dioxide, chlorine, ammonia, hydrogen fluoride.

Show the on-the-spot described toxic gas radio detection node section structural representation of accident described in the embodiment of the present application with reference to Fig. 1-A.Wherein, toxic gas radio detection node includes poisonous gas sensor array module (101) and the first micro controller module (102); Described toxic gas sensor array module (101), for detection of the first toxic gas information, can comprise one-parameter sensor array module and multi-parameter sensor array module; In embodiments of the present invention, sensor array module comprises chemical sensor array module and biosensor array module.

The first toxic gas information that described the first micro controller module (102) detects for analyzing described toxic gas sensor array module (101); Described the first micro controller module (102) comprises cluster module (103) and artificial neural network identification module (104); Described cluster module (103) is carried out cluster for the first toxic gas information that described toxic gas sensor array module (101) is detected, and obtains cluster result; Described artificial neural network identification module (104) for, utilize based on artificial neural network build toxic gas model of cognition identify described cluster result, obtain the second toxic gas information.

One-parameter sensor array module comprises one-parameter biosensor array module and one-parameter chemical sensor array module, and wherein, one-parameter biosensor array module is for detection of the virus or the bacterium etc. that are discharged by chemical and biological weapons; One-parameter chemical sensor array module is for detection of toxic gases such as common carbon monoxide or nitrogen monoxides.Multi-parameter sensor array module comprises multiparameter biosensor array module and multiparameter chemical sensor array module; Wherein, multiparameter chemical sensor array module is for detection of the mixed gas such as carbon monoxide and nitric oxide production mixed gas or other gas mixing; Multiparameter biosensor array module is for detection of the virus or bacterium or the mould poison etc. that mix.

Fig. 1-B shows the on-the-spot described portable intelligent handheld terminal part-structure schematic diagram of accident described in the embodiment of the present application.Wherein, described portable intelligent handheld terminal comprises the 3rd micro controller module 105, and described the 3rd micro controller module 105, for by described the second toxic gas information, merges by classification under toxic gas, obtains the 3rd toxic gas information.

Effusiometer is calculated module (504), for speed and the scope of on-the-spot the first toxic gas diffusion of information of quick obtaining.As in the time that the on-the-spot generation of accident toxic gas leaks, not only require kind and the concentration of fast detecting the first toxic gas information, also need speed and the scope of on-the-spot the first toxic gas diffusion of information of quick obtaining.According to the Fick diffusion differential equation, have

$\frac{\∂C}{\∂t}+u_x\frac{\∂C}{\∂x}+u_y+\frac{\∂C}{\∂y}+u\frac{\∂C}{\∂z}=D_x\frac{{\∂}^{2}C}{{\∂y}^2}+D_y\frac{{\∂}^{2}C}{{\∂y}^2}+D_z\frac{{\∂}^{2}C}{{\∂z}^2}---\left(10\right)$

In formula, C is gaseous mass concentration, mg/m ³; u _x, u _y, u _zfor x, y, the mean wind speed m/s in z direction; D _x, D _y, D _zfor x, y, the coefficient of diffusion in z direction, m ²/ s.

If only consider the situation of plane wind, u _z=0, establishing wind angle is α, and wind speed is f,

u _x＝fcosα，u _y＝fsinα （ ¹¹）

For the ease of solving, establishing coefficient of diffusion is constant, i.e. D _x=D _y=D _z=D, diffusion equation can be reduced to

$\frac{\∂C}{\∂t}+f\cos \mathrm{\α}\frac{\∂C}{\∂x}+f\sin \mathrm{\α}\frac{\∂C}{\∂y}=D[\frac{{\∂}^{2}C}{{\∂x}^2}+\frac{{\∂}^{2}C}{{\∂y}^2}+\frac{{\∂}^{2}C}{{\∂z}^2}]---\left(12\right)$

The high H of being of storage tank that is provided with poisonous gas, radius is R, initial concentration C=C in tank ₀, concentration C=0 in tank.Starting condition is

$C=\left\{\begin{array}{cc}{C}_0,& \sqrt{x^2+y^2}\≤R,\left|z\right|\≤H,t=0\\ 0,& \sqrt{x^2+y^2}\≥R,t=0\\ 0,& \sqrt{x^2+y^2}=\±\∞\\ 0,& z=\∞\end{array}\right.---\left(13\right)$

Boundary condition is

$\frac{\∂C}{\∂z}=0,z=0---\left(14\right)$

By starting condition and Boundary Condition for Solving diffusion equation (10), can obtain on-the-spot the first toxic gas information concentration process and on-the-spot the first toxic gas diffusion of information radius process over time over time, to take as early as possible effective counter-measure.

Said system is mainly for the at a distance detection of the first toxic gas information at accident scene, for in-plant accident scene can adopt toxic gas radio detection node (A101-A10n), directly and portable intelligent handheld terminal communicate.

In the application, be combined with by cluster module and artificial neural network identification module, can effectively determine type and the concentration of the second toxic gas.

Preferably, the on-the-spot prior detection node that all do not have of accident, once accident occurs, in the situation that can not determine which kind of toxic gas accident scene be, can adopt the mode such as transmitter or aircraft or robot in the application described toxic gas radio detection node to be sent to the toxic gas region at accident scene, can fast detecting go out the first toxic gas information at accident scene.Show the transmission mode schematic diagram of the on-the-spot toxic gas radio detection of accident node described in the embodiment of the present application with reference to Fig. 2.Wherein, toxic gas radio detection node can provide n radio detection node.In the time that toxic gas radio detection node is relatively smooth to the road between the first toxic gas information area (201) at accident scene, can adopt robot (202) to carry the mode of poisonous gas radio detection node, toxic gas radio detection node is sent to on-the-spot the first toxic gas information area (201) of accident.

In the time that toxic gas radio detection node is uneven to the road between the first toxic gas information area (201) at accident scene, can adopt aircraft (203) to carry the mode of poisonous gas radio detection node, toxic gas radio detection node is sent to on-the-spot the first toxic gas information area (201) of accident.

In the time that toxic gas radio detection node is more severe to the environment between the first toxic gas information area (201) at accident scene, the situations such as strong wind, typhoon or against the wind, road is rugged and rough, the mode that can adopt transmitter (204) transmitting toxic gas radio detection node, sends to on-the-spot the first toxic gas information area (201) of accident by toxic gas radio detection node.

Above three kinds of modes are optimum way that toxic gas radio detection node is sent to the first toxic gas information area at accident scene, those skilled in the art can adopt other modes according to concrete applied environment, and the application is without restriction to the mode of concrete employing.

With reference to Fig. 3, show the structural drawing of the on-the-spot toxic gas radio detection of accident node described in the embodiment of the present application.Wherein, by including poisonous gas chemical sensor array module (301) and toxic gas biosensor array module (302) forms toxic gas sensor array module of the present invention, toxic gas chemical sensor array module (301) and toxic gas biosensor array module (302) all can by one-parameter sensor array module and or multi-parameter sensor array module form, the first toxic gas information that its detection obtains is sent to the first micro controller module (102).Toxic gas radio detection node also includes poisonous gas adsorption module (303), the first locating module (305) and the first wireless communication module (304);

Also include poisonous gas adsorption module (303) for adsorbing the toxic gas of toxic gas radio detection node position, site of the accident annex.Be appreciated that also and gather by the original poison gas in the scene of the accident, be sealed in toxic gas adsorption module, in order to the follow-up off-line analysis that carries out in laboratory;

Wherein, the first wireless communication module (304) is sent to information for the second toxic gas information that the first micro controller module (102) analysis is obtained and converges transit node;

Wherein the first micro controller module (102) is connected with the first locating module (305); Described the first locating module (305) is for obtaining the locating information of toxic gas radio detection node;

Described toxic gas radio detection node also comprises the first supply module (306), for providing power supply to the modules of toxic gas radio detection node.

Described the first micro controller module (102) can also be used for the adsorbed state of control chart 3 toxic gas adsorption modules (303).

Further, described system also comprises: off-line analysis module, for described the first toxic gas information is carried out to off-line analysis, obtains off-line analysis result;

After reclaiming by the toxic gas radio detection node of the scene of the accident, in laboratory, utilize off-line analysis module (such as large-scale poison gas analytical instrument etc.) to carry out accurate special analysis the toxic gas in the toxic gas adsorption module (303) of each toxic gas radio detection node, obtain accurate poison gas analysis result, such as the concentration of poison gas type and this poison gas type.

Further, also comprise: optimize module, for utilizing the Output rusults of described off-line analysis modified result artificial neural network identification module.

Be appreciated that, in laboratory, each toxic gas radio detection node reclaiming is carried out to the result that off-line analysis obtains with professional equipment, may be more accurate, the application can be according to off-line analysis result (also comprising poison gas type and concentration) so, compare with the second toxic gas information (comprising poison gas type and concentration) of toxic gas radio detection node itself, if error exceedes threshold value, illustrate that in the toxic gas model of cognition building based on artificial neural network, parameter exists deviation, the application utilizes off-line analysis result to revise above-mentioned parameter.

Show accident field data described in the embodiment of the present application and converge the structural drawing of transit node referring to Fig. 4.Wherein, information converges transit node and comprises intra-node Temperature Humidity Sensor module (403), the second micro controller module (402) and the second wireless communication module (401); The second toxic gas information that described the second wireless communication module (401) obtains for receiving toxic gas radio detection node; Described intra-node Temperature Humidity Sensor module (403) is integrated in described information and converges on transit node, converges the temperature and humidity in transit node for detection of described information, and is sent to the second micro controller module; Described information converges transit node and also comprises the second locating module (404) being connected with described the second micro controller module (402); Described the second locating module (404) converges locating information and the relevant information of transit node for obtaining described information; Described the second wireless communication module (401) is for receiving locating information, the temperature and humidity information of the on-the-spot toxic gas radio detection of accident node, and locating information, temperature and humidity information exchange are crossed to (certainly information can also be preserved) after the processing of the second micro controller module, locating information after treatment, temperature and humidity information are sent to portable intelligent handheld terminal; Described information converges transit node and also comprises the second supply module (405), provides power supply for the modules that converges transit node to described information.

Show the structural drawing of the on-the-spot portable intelligent handheld terminal of accident described in the embodiment of the present application referring to Fig. 5.Wherein, described portable intelligent handheld terminal also comprises the 3rd wireless communication module (501), the 3rd micro controller module (105), demonstration load module (502), data fusion module (503), effusiometer calculation module (504), wakes control module (505) and the 3rd supply module (506) up; Described the 3rd micro controller module (105) is for controlling the 3rd wireless communication module (501), demonstration load module (502), body diffusion computing module (504), waking control module (505), data fusion module (503) up; Described the 3rd wireless communication module (501) converges for receiving described information the second toxic gas information that transit node detects, and described the second toxic gas information exchange is crossed to the 3rd microcontroller mould (105) and be sent to described data fusion module (503) piece and carry out Data Fusion.The 3rd toxic gas information after data fusion module (503) Data Fusion is sent to described demonstration load module (502); Described effusiometer is calculated speed and the scope of module (504) for the diffusion of fast detecting toxic gas, and testing result is sent to the 3rd micro controller module (105), the 3rd micro controller module (105) sends to demonstration load module (502) to show the speed of described toxic gas diffusion and scope; Described demonstration load module (502) is for the duty of displaying scene node and the various detection data of toxic gas; Described the 3rd supply module (506) is for providing power supply to the modules of portable intelligent handheld terminal.

Described portable intelligent handheld terminal also comprises: wake control module (505) up, be suitable for the distribution situation according to described locating information, portable intelligent handheld terminal wakes toxic gas radio detection node for subsequent use that can guarded region up, by the 3rd micro controller module (105) and the 3rd wireless communication module (501) wake up can guarded region toxic gas radio detection node for subsequent use.

In throwing in toxic gas radio detection node, sometimes may there is detection node situation pockety, now, can be according to the distribution situation of locating information, in the region that can monitor, wake the toxic gas radio detection node for subsequent use in the region that can monitor by the wake module in portable intelligent handheld terminal up, toxic gas radio detection node is evenly distributed in the region that can monitor.In the time that in accident scene, certain toxic gas radio detection node breaks down, portable intelligent handheld terminal can also utilize the first locating module in toxic gas radio detection node to obtain the locating information of the first toxic gas radio detection node, according to the distribution situation of locating information, in the region that can monitor, adjust the delivering path of the first toxic gas information.For example: in the time monitoring a certain toxic gas radio detection node at portable handheld terminal and break down, can adjust online original by the path of this node transmission information, do not reach this node by the first original toxic gas information, reach the non-malfunctioning node nearest apart from this malfunctioning node (being toxic gas radio detection node) and change, the detection information of guaranteeing thus whole toxic gases reaches reliably information and converges transit node.

In like manner, when information converges indivedual nodes in transit node or portable intelligent handheld terminal while breaking down, the method while all adopting toxic gas radio detection node to break down is carried out communication.Be that information converges transit node or portable intelligent handheld terminal is selected to carry out communication apart from the nearest non-malfunctioning node of malfunctioning node.

Described portable intelligent handheld terminal also has the duty of observable Site Detection node; Change and detect information network delivering path; Manually wake redundant detection node up; Find the function of work on the spot detection node and malfunctioning node.Can also maintenance data integration technology, the measuring accuracy of the on-the-spot toxic gas of raising accident and the diffusion tendency of analysis toxic gas.

Described cluster module comprises: the first cluster module, be used for adopting Fuzzy c-means clustering algorithm KFCM that the first toxic gas information exchange is crossed to Function Mapping and obtain initial value to carrying out again FCM cluster after high-dimensional feature space, recycling particle group optimizing PSO algorithm is optimized to obtain best cluster result to described initial value, and the cluster majorized function of its use comprises:

$f\left(x_i\right)=\frac{1}{J_a(U,V)+1}=\frac{1}{2\underset{i=1}{\overset{C}{\mathrm{\Σ}}}\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\μ}}_{\mathrm{ij}}^m[1-K(x_i,v_j)]+1}---\left(1\right)$

Wherein, C is cluster number, and N is sample number, U=[μ _ij] _{c × N}fuzzy C Matrix dividing, μ _ijfor sample x _icorresponding to the degree of membership value of j cluster, V=[v _j] be the set of C cluster centre composition, m is the index weight that affects degree of membership matrix norm gelatinization degree, J _α(U, V) is the cost function of FCM, K (x _i, v _j) be gaussian kernel function, wherein x _ifor m dimensional vector, v _jwith x _iit is the vector with dimension.The more new formula of cluster centre and degree of membership is as follows:

$v_j=\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\μ}}_{\mathrm{ij}}^{m}K(x_i,v_j)x_i}{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\μ}}_{\mathrm{ij}}^{m}K(x_i,v_j)}---\left(2\right)$

${\mathrm{\μ}}_{\mathrm{ij}}=\frac{{(1-K(x_i,v_j))}^{-1/(m-1)}}{\underset{j=1}{\overset{C}{\mathrm{\Σ}}}{(1-K(x_i,v_j))}^{-1/(m-1)}}---\left(3\right)$

PSO algorithm is the fitness based on population and particle, and each particle has position and two features of speed.In each iteration, each particle upgrades position and the speed of self by following the tracks of individual extreme value and global extremum, and its computing formula is as follows:

V _iα(t+1)＝ωV _iα(t)+c ₁r ₁[P _iα(t)-X _iα(t)]+c ₂r ₂[P _fα(t)-X _iα(t)] （4）

X _iα(t+1)＝X _iα(t)+V _iα(t+1) （5）

In formula, V _{i α}(t) be the speed of particle previous moment, V _{i α}(t+1) be the current speed of particle, X _{i α}(t) be the current location of particle, X _{i α}(t+1) be the reposition that particle produces, P _{i α}(t) be individual extreme value, P _{f α}(t) be global extremum, c ₁, c ₂for the study factor (for being greater than zero positive integer), r ₁, r ₂for the random number between [0,1], i=1,2 ..., N, ω is inertial factor.

Cluster is a kind of important data analysis technique, and by cluster, people can identify intensive and sparse region, thereby find overall distribution pattern, and mutual relationship between data attribute.Fuzzy C-Means Cluster Algorithm is the popularizing form of k means clustering algorithm, both can not ensure to find the optimum solution of problem, all likely converges to local extremum.Therefore, the initial cluster center of application particle group optimizing (PSO) algorithm optimization Fuzzy c-means clustering algorithm (KFCM) is realized data clusters, to realize higher cluster accuracy.

The numerical value of exporting the first toxic gas information according to the toxic gas sensor array module in the on-the-spot toxic gas radio detection of accident node, the number C(that determines the first toxic gas information cluster is C kind mixed gas).N is within definite time period, by the sum (being that N is total sample number) of toxic gas sensor array module sampling output the first toxic gas information value.Utilize the computing formula of fuzzy C-means clustering (FCM) and core fuzzy C-means clustering (KFCM) to carry out cluster to the first toxic gas information, concrete fuzzy C-means clustering (FCM) and the computing formula of core fuzzy C-means clustering (KFCM) can adopt computing formula of the prior art, and the application is not limited this.

Further, in the first cluster module, the KFCM algorithm steps based on PSO can briefly be described below:

1. the number C(C that determines the first toxic gas information cluster can be the number of the detectable poison gas type of toxic gas sensor array module), set permissible error ε, and establish iterations t=1, set total sample number N, set Inertia Weight ω, study factor c1 and c2 and index weight, by speed and the position of above calculation of parameter particle (particle can be understood as each detected value of toxic gas sensor array module transmission).The concrete calculating speed of particle and the formula of position can adopt computing formula of the prior art, and the application is not limited this.

2. initialization population v ₁, v ₂..., v _j..., v _c, from sample set (each poison gas information of surveying for toxic gas sensor array module), appoint and get C vector and carry out initialization v _j, establish sample space X=[x1, x2 ..., x _n], calculate gaussian kernel function K(x _i, v _j);

Wherein, v _jfor cluster centre.

3. for each sample calculation degree of membership matrix U, then calculate cluster majorized function $f\left(x_i\right)=\frac{1}{J_{\mathrm{\α}}(U,V)+1}=\frac{1}{2\underset{i=1}{\overset{C}{\mathrm{\Σ}}}\underset{j=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\μ}}_{\mathrm{ij}}^m[1-K(x_i,v_j)]+1}---\left(1\right)$ Revise particle rapidity and position according to formula (4) formula (5) again, revise individual extreme value and global extremum according to cluster majorized function value, to produce particle of future generation;

4. judgement: whether current iteration number of times reaches predefined maximum times, if reached, stops iteration, obtains the optimum solution of cluster centre; Otherwise forward step to 3.;

5. upgrade the degree of membership of calculating particle colony, upgrade the cluster centre that calculates colony; Upgrade by formula (3), (2).

6. calculate the difference E of adjacent generations degree of membership matrix, if E< is ε, iterative computation stops; Otherwise forward to 5..

Described artificial neural network identification module comprises: N artificial neural network recognin module, described artificial neural network recognin module is for identifying the cluster result of corresponding classification.

Build artificial neural network submodule by historical data and experimental data, and it is trained, then just can be applied to scene.Therefore the number of sensors (comprising: one-parameter chemical sensor+one-parameter biology sensor+multiparameter chemical sensor+multiparameter biology sensor) and the related experiment data that, are integrated on toxic gas radio detection node have been determined artificial neural network submodule quantity.

Further, described system also comprises: Fusion Module, for described the 3rd toxic gas information is merged, realize fusion treatment by following formula:

$\left\{\begin{array}{c}\stackrel{\&OverBar;}{x}=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i{z}_i\\ \underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i=1(0\&le;w_i\&le;1)\end{array}\right.---\left(6\right)$

Wherein, x is true value, z _ifor the measured value of cluster result, w _ifor the weighting factor of cluster result, x is the data after data fusion;

Wherein, be the bias free that ensures to estimate,

, its error variance expression formula is

${\mathrm{\&sigma;}}^2=E\left[{(x-\stackrel{\&OverBar;}{x})}^2\right]=E\left[\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i^2{(x-\stackrel{\&OverBar;}{x})}^2\right]=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i^2{\mathrm{\&sigma;}}_i^2---\left(7\right)$

Wherein, weighting factor w _i,

$w_i=\frac{{\mathrm{\&sigma;}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&sigma;}}_i^{-2}}---\left(8\right)$

By formula (7) and formula (8) substitution formula (6) acquisition data fusion result be

$\stackrel{\&OverBar;}{x}=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{z}_i{\mathrm{\&sigma;}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&sigma;}}_i^{-2}}---\left(9\right)$

Described portable intelligent handheld terminal also comprises:

For said system embodiment; the present invention also provides a kind of accident on-the-spot toxic gas long-range wireless monitoring method; the online test method that shows the on-the-spot toxic gas of accident described in the embodiment of the present application with reference to figure 6, comprising: toxic gas radio detection node is sent to accident scene; Utilize the first toxic gas information of the on-the-spot toxic gas radio detection of the toxic gas sensor array module collection accident node in toxic gas radio detection node.Toxic gas sensor array comprises one-parameter sensor array module and multi-parameter sensor array module.One-parameter sensor array module comprises one-parameter biosensor array module and one-parameter chemical sensor array module; Multi-parameter sensor array module comprises multiparameter biosensor array module and multiparameter chemical sensor array module.Utilize the cluster module in toxic gas radio detection node that the first toxic gas information collecting is carried out to cluster, obtain cluster result; As shown in Figure 6, x ₁-x _nrepresent the output of one-parameter chemical sensor array module (601), x ₁₁-x _mmbe expressed as the output of multiparameter chemical sensor array module (602), y ₁-y _nbe expressed as the output of one-parameter biosensor array module (603), y ₁₁-y _nnbe expressed as the output of multiparameter biosensor array module (604).The output data of each sensor array module all adopt Fuzzy c-means clustering algorithm KFCM that the first toxic gas information exchange is crossed to Function Mapping and obtain the initial value after cluster to carrying out FCM cluster after high-dimensional feature space again, the object of cluster is that identical gas is combined, the initial value after the cluster obtaining.For reaching best cluster result, recycling particle group optimizing PSO algorithm is optimized iterative computation to described initial value, and the cluster majorized function of its use comprises:

$f\left(x_i\right)=\frac{1}{J_{\mathrm{\&alpha;}}(U,V)+1}=\frac{1}{2\underset{i=1}{\overset{C}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ij}}^m[1-K(x_i,v_j)]+1}---\left(1\right)$

Wherein, C is cluster number, and N is sample number, U=[μ _ij] _{c × N}fuzzy C Matrix dividing, μ _ijfor sample x _icorresponding to the degree of membership value of j cluster, V=[v _j] be the set of C cluster centre composition, m is the index weight that affects degree of membership matrix norm gelatinization degree, J _α(U, V) is the cost function of FCM, K (x _i, v _j) be gaussian kernel function, wherein x _ifor m dimensional vector, v _jwith x _iit is the vector with dimension.The more new formula of cluster centre and degree of membership is as follows:

$v_j=\frac{\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ij}}^{m}K(x_i,v_j)x_i}{\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ij}}^{m}K(x_i,v_j)}---\left(2\right)$

${\mathrm{\&mu;}}_{\mathrm{ij}}=\frac{{(1-K(x_i,v_j))}^{-1/(m-1)}}{\underset{j=1}{\overset{C}{\mathrm{\&Sigma;}}}{(1-K(x_i,v_j))}^{-1/(m-1)}}---\left(3\right)$

PSO algorithm is the fitness based on population and particle, and each particle has position and two features of speed.In each iteration, each particle upgrades position and the speed of self by following the tracks of individual extreme value and global extremum, and its computing formula is as follows:

V _iα(t+1)＝ωV _iα(t)+c ₁r ₁[P _iα(t)-X _iα(t)]+c ₂r ₂[P _fα(t)-X _iα(t)] （4）

X _iα(t+1)＝X _iα(t)+V _iα(t+1) （5）

In formula, V _{i α}(t) be the speed of particle previous moment, V _{i α}(t+1) be the current speed of particle, X _{i α}(t) be the current location of particle, X _{i α}(t+1) be the reposition that particle produces, P _{i α}(t) be individual extreme value, P _{f α}(t) be global extremum, c ₁, c ₂for the study factor (for being greater than zero positive integer), r ₁, r ₂for the random number between [0,1], i=1,2 ..., N, ω is inertial factor.

Adopt the artificial neural network identification module in toxic gas radio detection node, utilize the toxic gas model of cognition building based on artificial neural network that the initial value of the cluster result obtaining is identified, and obtain the second toxic gas information; Utilize information to converge transit node described the second toxic gas information is converged, and be sent to portable intelligent handheld terminal; Utilize the 3rd micro controller module in portable intelligent handheld terminal by the second toxic gas information, merge by classification under toxic gas, obtain the 3rd toxic gas information, thereby determined type and the concentration of the 3rd final toxic gas information.

Preferably, described method also comprises: by the first toxic gas information at toxic gas adsorption module (606) the absorption accident scene in toxic gas radio detection node, to get described the first toxic gas information and adopt dedicated analysis instrument (607) to carry out off-line analysis, obtain off-line analysis result; Utilize described off-line analysis result to revise the recognition result of artificial neural network identification module.

Preferably, Parameter Map 7 shows the on-the-spot toxic gas detection of accident and the concrete application example of recognition principle described in the embodiment of the present application.In the time that the on-the-spot generation of accident is single or be mixed with poisonous gas leakage, by various transmission modes in the application, n toxic gas radio detection node sent to accident scene.Because the poison gas at accident scene may be pure gas, also may be mixed gas (in figure 701), cause the detection data that the radio detection node being distributed in zones of different is passed back to have certain difference, particularly more accurately type and the concentration of each gas in detection and Identification mixed gas, has larger difficulty.For this reason, the method that adopts particle group optimizing (PSO) algorithm and Fuzzy c-means clustering algorithm (KFCM) to combine, i.e. PSO+KFCM algorithm.Then effectively classify (702) to detecting data based on PSO+KFCM algorithm, obtain the data class 1-data class m(703 after cluster and optimization).Again by m artificial neural network, i.e. ANN1-ANNm(704) data class 1-data class m is identified one by one after, obtain gas 1-gas m after treatment.Finally obtain determining type and the concentration of gas 1-gas m.Further, after each accident treatment completes, the toxic gas of the toxic gas adsorption module absorption to each toxic gas detection node carries out off-line analysis, according to this toxic gas detection module of analysis result correction, for revising weights (the i.e. figure 705 of the toxic gas model of cognition knowledge building based on artificial neural network, based on off-line analysis data, revise neural network weight), to realize the object that improves constantly measuring accuracy.

In the time being sent to certain toxic gas radio detection node at accident scene and breaking down, portable intelligent handheld terminal can utilize the first locating module in toxic gas radio detection node to obtain the locating information of the first toxic gas radio detection node, according to the distribution situation of locating information, in the region that can monitor, adjust the delivering path of the first toxic gas information.Detection method when the on-the-spot toxic gas detection node section of accident sends fault described in the embodiment of the present application with reference to Fig. 8.Portable handheld terminal can detect the state of malfunctioning node, and can adjust easily radio detection node in accident scene (801) and the information in transit node set (802) of converging converges the information transmission path of transit node.

The toxic gas radio detection node (2) of supposing to be sent in accident scene (801) breaks down, it is toxic gas radio detection node (3) that the first toxic gas information that toxic gas radio detection node (1) detects is sent to apart from the nearest non-malfunctioning node of malfunctioning node, then converges transit node (1) by toxic gas radio detection node (3) information of being sent to.

Suppose that the information information in transit node set (802) of converging converges transit node (1) while breaking down, toxic gas radio detection node (2) detects the second toxic gas information obtaining and is transferred to and converges the nearest non-fault transit node information of transit node (1) from fault transit node information and converge in transit node (2), then information converges transit node (2) and reaches portable handheld terminal (2);

While supposing that portable handheld terminal (1) in portable handheld terminal set (803) breaks down, information converges the second toxic gas information that transit node (1) converges, and to be transferred to from the nearest non-fault portable handheld terminal of fault transit node be portable handheld terminal (2).

Described method, also comprises the second toxic gas information of utilizing the toxic gas model of cognition building based on artificial neural network to obtain is carried out to data fusion, realizes Data Fusion by following formula:

$\left\{\begin{array}{c}\stackrel{\&OverBar;}{x}=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i{z}_i\\ \underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i=1(0\&le;w_i\&le;1)\end{array}\right.---\left(6\right)$

Wherein, x is true value, z _ifor the measured value of cluster result, w _ifor the weighting factor of cluster result, x is the data after data fusion;

According to the historical data of each toxic gas radio detection node, the variance that calculates n toxic gas radio detection node in measuring system is respectively in advance

if the true value that certain measurement moment need be estimated is x, each toxic gas radio detection node is respectively z at the measured value in this moment ₁, z ₂..., z _n, each measurement result is independent each other, and the weighting factor of each toxic gas radio detection node is respectively w ₁, w ₂..., w _n, after fusion treatment

value and weighting factor should meet formula (4).After time crash analysis is complete, the data of each toxic gas radio detection node when in time accident are added to described historical data, again estimate corresponding variance.

Wherein, be the bias free that ensures to estimate,

, its error variance expression formula is:

${\mathrm{\&sigma;}}^2=E\left[{(x-\stackrel{\&OverBar;}{x})}^2\right]=E\left[\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i^2{(x-\stackrel{\&OverBar;}{x})}^2\right]=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i^2{\mathrm{\&sigma;}}_i^2---\left(7\right)$

Wherein, weighting factor w _i, $w_i=\frac{{\mathrm{\&sigma;}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&sigma;}}_i^{-2}}---\left(8\right)$

By formula (7) and formula (8) substitution formula (6) acquisition data fusion result be:

$\stackrel{\&OverBar;}{x}=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{z}_i{\mathrm{\&sigma;}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&sigma;}}_i^{-2}}---\left(9\right)$

By being distributed in the toxic gas sensor array module in the on-the-spot toxic gas radio detection of accident node, detection is this toxic gas radio detection node toxic gas around, complete basic processing and identification, comprising pre-service, cluster analysis, pattern-recognition, to draw kind and the concentration of the toxic gas that node was detected separately.Then can cross portable intelligent handheld terminal, maintenance data integration technology, the final associating inferred results (being kind and the concentration of on-the-spot toxic gas) that obtains.

The application can be sent to accident scene by toxic gas radio detection node by transmitter or aircraft or robot, detects fast the first toxic gas information at accident scene by the one-parameter sensor array module in toxic gas radio detection node and multi-parameter sensor array module.Secondly, in the application, be combined with by cluster module and artificial neural network identification module, can effectively determine type and the concentration of the second toxic gas.Again, the portable intelligent handheld terminal in the application can monitor in real time the duty of the on-the-spot toxic gas radio detection of accident node and wake up can guarded region toxic gas radio detection node for subsequent use.The application's main innovate point comprises: (1) detection system innovation: the difficult point that fully takes into account toxic gas Site Detection, carry out overall improvement and framework from system, can throw in several ways the toxic gas detection node of (mode such as robot input, aircraft input, transmitter input), converge transit node and can be away from the handheld terminal of poison gas field monitor for the information of data relay, with under unknown emergency case, under the personnel's of guarantee safety, can measure relatively accurately poison gas type and concentration; (2) detection algorithm integrated innovation: the environmental consideration that the present invention is based on toxic gas scene, for the accurate toxic gas that detects, propose first the detection data of poison gas to be carried out to the KFCM cluster based on PSO, then adopt the toxic gas model of cognition based on artificial neuron model construction to identify, finally handheld terminal carry out all data fusion this combine especially, obtained the toxic gas recognition result of pinpoint accuracy.And can utilize described off-line analysis result to revise the recognition result of artificial neural network identification module, further to improve the accuracy of detection of the on-the-spot toxic gas long distance wireless of accident; (3) intelligent principle of handheld terminal innovation: handheld terminal of the present invention can be determined poison gas range of scatter and the speed at poison gas scene fast, can receive the data of all toxic gas detection nodes, and it is carried out to fusion treatment.And can adjust online according to on-the-spot actual conditions the transmission path of detection information, and make whole scheme overall architecture simply effective, facilitate technician to make follow-up decision.

The on-the-spot toxic gas remote wireless monitoring system of a kind of accident and the method that above the application are provided, be described in detail, applied principle and the embodiment of specific case to the application herein and set forth, the explanation of above embodiment is just for helping to understand the application's method and core concept thereof; , for one of ordinary skill in the art, according to the application's thought, all will change in specific embodiments and applications, in sum, this description should not be construed as the restriction to the application meanwhile.

Claims (10)

1. the on-the-spot toxic gas remote wireless monitoring system of accident, is characterized in that, comprising: at least one toxic gas radio detection node, at least one information converge transit node and at least one portable intelligent handheld terminal; Wherein, described toxic gas radio detection node, for detection of the first toxic gas information at accident scene, adopts variety of way that described toxic gas radio detection node is sent to on-the-spot the first toxic gas information area of accident;

Described toxic gas radio detection node includes poisonous gas sensor array module and the first micro controller module;

Described toxic gas sensor array module, for detection of the first toxic gas information, comprises one-parameter sensor array module and multi-parameter sensor array module;

The first toxic gas information that described the first micro controller module detects for analyzing described toxic gas sensor array module; Described the first micro controller module comprises cluster module and artificial neural network identification module;

Described cluster module is carried out cluster for the first toxic gas information that described toxic gas sensor array module is detected, and obtains cluster result;

Described artificial neural network identification module is used for, and utilizes the toxic gas model of cognition building based on artificial neural network to identify described cluster result, obtains the second toxic gas information;

Information converges transit node, for the second toxic gas information of each toxic gas radio detection node is converged, and is sent to described portable intelligent handheld terminal;

Described portable intelligent handheld terminal comprises the 3rd micro controller module and effusiometer calculation module, and described the 3rd micro controller module, for by described the second toxic gas information, merges by classification under toxic gas, obtains the 3rd toxic gas information; Utilize effusiometer to calculate module and draw fast speed and the scope of on-the-spot the first toxic gas diffusion of information.

2. system according to claim 1, is characterized in that, described toxic gas radio detection node also includes poisonous gas adsorption module, the first locating module and the first wireless communication module; The first toxic gas information that described toxic gas adsorption module detects for adsorbing toxic gas sensor array module; Described wireless communication module converges transit node for the first toxic gas information of the detection of sensor array module is sent to information; Described the first micro controller module is connected with the first locating module; Described the first locating module is for obtaining the locating information of toxic gas radio detection node; Described toxic gas radio detection node also comprises the first locating module being connected with toxic gas sensor array module, toxic gas adsorption module, the first micro controller module, described the first micro controller module and the first power module being connected with the first wireless communication module; Described the first power module is for providing power supply to the modules of toxic gas radio detection node;

Further, described system also comprises: off-line analysis module, for described the first toxic gas information is carried out to off-line analysis, obtains off-line analysis result;

Further, also comprise: optimize module, for utilizing the Output rusults of described off-line analysis modified result artificial neural network identification module.

3. system according to claim 1, is characterized in that, described cluster module comprises:

The first cluster module, be used for adopting Fuzzy c-means clustering algorithm KFCM that the first toxic gas information exchange is crossed to Function Mapping and obtain initial value to carrying out again fuzzy C-means clustering FCM after high-dimensional feature space, recycling particle group optimizing PSO algorithm is optimized iterative computation to obtain best cluster result to described initial value, and the cluster majorized function of its use comprises:

$f\left(x_i\right)=\frac{1}{J_a(U,V)+1}=\frac{1}{2\underset{i=1}{\overset{C}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ij}}^m[1-K(x_i,v_j)]+1}---\left(1\right)$

Wherein, C is cluster number, and N is sample number, U=[μ _ij] _{c × N}fuzzy C Matrix dividing, μ _ijfor sample x _icorresponding to the degree of membership value of j cluster, V=[v _j] be the set of C cluster centre composition, m is the index weight that affects degree of membership matrix norm gelatinization degree, J _α(U, V) is the cost function of FCM, K (x _i, v _j) be gaussian kernel function, wherein x _ifor m dimensional vector, v _jwith x _ithat the more new formula of cluster centre and degree of membership is as follows with the vector of dimension:

$v_j=\frac{\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ij}}^{m}K(x_i,v_j)x_i}{\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ij}}^{m}K(x_i,v_j)}---\left(2\right)$

${\mathrm{\&mu;}}_{\mathrm{ij}}=\frac{{(1-K(x_i,v_j))}^{-1/(m-1)}}{\underset{j=1}{\overset{C}{\mathrm{\&Sigma;}}}{(1-K(x_i,v_j))}^{-1/(m-1)}}---\left(3\right)$

In each iteration, each particle upgrades position and the speed of self by following the tracks of individual extreme value and global extremum, and its computing formula is as follows:

V _iα(t+1)＝ωV _iα(t)+c ₁r ₁[P _iα(t)-X _iα(t)]+c ₂r ₂[P _fα(t)-X _iα(t)] （4）

X _iα(t+1)＝X _iα(t)+V _iα(t+1) （5）

In formula, V _{i α}(t) be the speed of particle previous moment, V _{i α}(t+1) be the current speed of particle, X _{i α}(t) be the current location of particle, X _{i α}(t+1) be the reposition that particle produces, P _{i α}(t) be individual extreme value, P _{f α}(t) be global extremum, c ₁, c ₂for the study factor (for being greater than zero positive integer), r ₁, r ₂for the random number between [0,1], i=1,2 ..., N, ω is inertial factor.

4. system according to claim 1, is characterized in that, described portable intelligent handheld terminal also comprises:

Wake control module up, be suitable for the distribution situation according to described locating information, portable intelligent handheld terminal wakes toxic gas radio detection node for subsequent use that can guarded region up;

Described portable intelligent handheld terminal also comprises the 3rd wireless communication module, the 3rd micro controller module, the 3rd supply module and shows load module; Described the 3rd micro controller module is used for controlling the 3rd wireless communication module and load module; Described the 3rd wireless communication module converges for receiving described information the second toxic gas information that transit node transmits, and by the 3rd micro controller module, the second toxic gas information is carried out to Data Fusion, the 3rd toxic gas information after Data Fusion is sent to described demonstration load module; Described demonstration load module is for duty and the relevant data of displaying scene node; Described the 3rd supply module is with the 3rd wireless communication module, the 3rd micro controller module and show that load module is connected; Described the 3rd supply module is for providing power supply to the modules of portable intelligent handheld terminal.

5. system according to claim 1, is characterized in that, described information converges transit node and comprises intra-node Temperature Humidity Sensor module, the second micro controller module and the second wireless communication module; Described the second wireless communication module is for receiving the second toxic gas information that toxic gas radio detection node obtains; Described intra-node Temperature Humidity Sensor module integration converges on transit node in described information, converges the temperature and humidity in transit node for detection of described information; Described the second wireless communication module is for being sent to portable intelligent handheld terminal by locating information, temperature and humidity information; Described information converges transit node and also comprises the second locating module being connected with described the second micro controller module; Described the second locating module converges the locating information of transit node for obtaining described information; Described information converges transit node and also comprises the second locating module being connected with intra-node Temperature Humidity Sensor module, the second wireless communication module, the second micro controller module, described the second micro controller module and the second supply module being connected with described the second wireless communication module; Described the second supply module provides power supply for the modules that converges transit node to described information.

6. system according to claim 1, is characterized in that, described artificial neural network identification module comprises: N artificial neural network recognin module, described artificial neural network recognin module is for identifying the cluster result of corresponding classification.

7. system according to claim 1, is characterized in that, also comprises:

Fusion Module, for described the 3rd toxic gas information is merged, realize fusion treatment by following formula:

$\left\{\begin{array}{c}\stackrel{\&OverBar;}{x}=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i{z}_i\\ \underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i=1(0\&le;w_i\&le;1)\end{array}\right.---\left(6\right)$

Wherein, x is true value, z _ifor the measured value of cluster result, w _ifor the weighting factor of cluster result, x is the data after data fusion;

Wherein, be the bias free that ensures to estimate,

, its error variance expression formula is

${\mathrm{\&sigma;}}^2=E\left[{(x-\stackrel{\&OverBar;}{x})}^2\right]=E\left[\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i^2{(x-\stackrel{\&OverBar;}{x})}^2\right]=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i^2{\mathrm{\&sigma;}}_i^2---\left(7\right)$

Wherein, weighting factor w _i, $w_i=\frac{{\mathrm{\&sigma;}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&sigma;}}_i^{-2}}---\left(8\right)$

(7) and (8) substitutions (6) are obtained to data fusion result is

$\stackrel{\&OverBar;}{x}=\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{z}_i{\mathrm{\&sigma;}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&sigma;}}_i^{-2}}---\left(9\right)$

8. the on-the-spot toxic gas long-range wireless monitoring method of accident, is characterized in that, comprising:

Toxic gas radio detection node is sent to accident scene;

Utilize the first toxic gas information of the on-the-spot toxic gas radio detection of the toxic gas sensor array module collection accident node in toxic gas radio detection node;

Utilize the cluster module in toxic gas radio detection node that the first toxic gas information collecting is carried out to cluster, obtain cluster result; Adopt the artificial neural network identification module in toxic gas radio detection node, utilize the toxic gas model of cognition building based on artificial neural network that the cluster result obtaining is identified, and obtain the second toxic gas information;

Utilize information to converge transit node described the second toxic gas information is converged, and be sent to portable intelligent handheld terminal;

Utilize the 3rd micro controller module in portable intelligent handheld terminal by the second toxic gas information, merge by classification under toxic gas, obtain the 3rd toxic gas information, and utilize effusiometer calculation module to determine speed and the scope of on-the-spot the first toxic gas diffusion of information.

9. method according to claim 8, it is characterized in that, described method, also comprise that adopting Fuzzy c-means clustering algorithm KFCM that the first toxic gas information exchange is crossed to Function Mapping obtains initial value to carrying out FCM cluster after high-dimensional feature space again, recycling particle group optimizing PSO algorithm is optimized iterative computation to obtain best cluster result to described initial value, and the cluster majorized function of its use comprises:

$f\left(x_i\right)=\frac{1}{J_{\mathrm{\&alpha;}}(U,V)+1}=\frac{1}{2\underset{i=1}{\overset{C}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ij}}^m[1-K(x_i,v_j)]+1}---\left(1\right)$

Wherein, C is cluster number, and N is sample number, U=[μ _ij] _{c × N}fuzzy C Matrix dividing, μ _ijfor sample x _icorresponding to the degree of membership value of j cluster, V=[v _j] be the set of C cluster centre composition, m is the index weight that affects degree of membership matrix norm gelatinization degree, J _α(U, V) is the cost function of FCM, K (x _i, v _j) be gaussian kernel function, wherein x _ifor m dimensional vector, v _jwith x _ithat the more new formula of cluster centre and degree of membership is as follows with the vector of dimension:

$v_j=\frac{\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ij}}^{m}K(x_i,v_j)x_i}{\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&mu;}}_{\mathrm{ij}}^{m}K(x_i,v_j)}---\left(2\right)$

${\mathrm{\&mu;}}_{\mathrm{ij}}=\frac{{(1-K(x_i,v_j))}^{-1/(m-1)}}{\underset{j=1}{\overset{C}{\mathrm{\&Sigma;}}}{(1-K(x_i,v_j))}^{-1/(m-1)}}---\left(3\right)$

In each iteration, each particle upgrades position and the speed of self by following the tracks of individual extreme value and global extremum, and its computing formula is as follows:

V _iα(t+1)＝ωV _iα(t)+c ₁r ₁[P _iα(t)-X _iα(t)]+c ₂r ₂[P _fα(t)-X _iα(t)] （4）

X _iα(t+1)＝X _iα(t)+V _iα(t+1) （5）

In formula, V _{i α}(t) be the speed of particle previous moment, V _{i α}(t+1) be the current speed of particle, X _{i α}(t) be the current location of particle, X _{i α}(t+1) be the reposition that particle produces, P _{i α}(t) be individual extreme value, P _{f α}(t) be global extremum, c ₁, c ₂for the study factor (for being greater than zero positive integer), r ₁, r ₂for the random number between [0,1], i=1,2 ..., N, ω is inertial factor.

10. method according to claim 8, is characterized in that, also comprises the second toxic gas information of utilizing the toxic gas model of cognition building based on artificial neural network to obtain is carried out to data fusion, realizes Data Fusion by following formula:

$\left\{\begin{array}{c}\stackrel{\&OverBar;}{x}=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i{z}_i\\ \underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i=1(0\&le;w_i\&le;1)\end{array}\right.---\left(6\right)$

Wherein, x is true value, z _ifor the measured value of cluster result, w _ifor the weighting factor of cluster result, x is the data after data fusion;

Wherein, be the bias free that ensures to estimate,

, its error variance expression formula is:

${\mathrm{\&sigma;}}^2=E\left[{(x-\stackrel{\&OverBar;}{x})}^2\right]=E\left[\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i^2{(x-\stackrel{\&OverBar;}{x})}^2\right]=\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{w}_i^2{\mathrm{\&sigma;}}_i^2---\left(7\right)$

Wherein, weighting factor w _i, $w_i=\frac{{\mathrm{\&sigma;}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&sigma;}}_i^{-2}}---\left(8\right)$

(7) and (8) substitutions (6) are obtained to data fusion result is:

$\stackrel{\&OverBar;}{x}\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{z}_i{\mathrm{\&sigma;}}_i^{-2}}{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{\mathrm{\&sigma;}}_i^{-2}}---\left(9\right)$

Images