Inversion estimation method for methane emission of open pit coal mine based on unmanned aerial vehicle monitoring data

Abstract

The invention discloses an opencast coal mine methane emission inversion estimation method based on unmanned aerial vehicle monitoring data. The method comprises the steps that S1, flight is planned over a research mining area, and methane concentration monitoring data containing position information and meteorological data are collected; s2, the reverse Lagrangian random diffusion model releases Np virtual particles through the observation points and simulates the reverse motion trail of the random simulation particles in the research mining area by using meteorological data to obtain the footprint sensitivity of the grids of the research mining area corresponding to the observation points, and accumulates the footprint sensitivity of all the grids of the research mining area to obtain the transfer coefficient of the observation points; and S3, calculating the average emission rate of the related local route influenced by the wind direction, and calculating to obtain the total emission rate of the research mining area. According to the method, emission traceability of the research mining area and accurate estimation of the total emission rate and the emission amount are realized, methane emission distribution and emission intensity of the research mining area can be quantitatively evaluated conveniently, and data support is provided for carbon emission accounting and environment monitoring of the research mining area.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of open coal mine methane emission inversion, and particularly relates to an open coal mine methane emission inversion estimation method based on unmanned aerial vehicle monitoring data. BACKGROUND

[0002] The carbon emission of coal mine enterprises mainly comes from methane and carbon dioxide, and the greenhouse effect brought by methane in the past two decades is 84 times that of carbon dioxide, so methane monitoring is the most important in the carbon monitoring of coal mine enterprises in China. Methane itself has relatively small mass and is easy to escape to the upper part of the mine area, and accurate inversion and quantification of methane emission sources are crucial for environmental regulation and climate change research. Traditional ground observation stations and satellite remote sensing have certain limitations in spatial coverage and resolution, and unmanned aerial vehicles equipped with methane sensors can detect methane concentration at low altitude and high resolution, providing a new means for atmospheric pollution source tracing. However, the existing technology either collects methane concentration monitoring data of the location by ground observation stations or obtains the air methane concentration of the open mine area by satellite remote sensing data, and cannot obtain the methane emission source and emission intensity of the open mine area, so it is impossible to quantitatively predict the emission of the open mine area. It is necessary to study a mine area emission source inversion method to effectively invert the emission intensity, which plays an extremely important role in mastering the methane emission characteristics and atmospheric pollution source tracing. SUMMARY

[0003] The present application aims to solve the technical problems pointed out in the background art, and provides an open coal mine methane emission inversion estimation method based on unmanned aerial vehicle monitoring data. The methane concentration monitoring data containing position information and meteorological data above the study mine area are collected, the inverse Lagrangian random diffusion model is used to release virtual particles with the methane concentration monitoring data as the observation point, and the meteorological data are used to simulate the reverse motion trajectory of the particles in the study mine area to simulate the footprint sensitivity corresponding to the observation point j of the study mine area grid or pixel, the methane concentration of the position above the study mine area is reversely tracked to the ground of the study mine area, and the total emission rate of the methane emission of the study mine area is inverted to obtain the daily, monthly and annual emission of the study mine area.

[0004] The purpose of the present application is achieved by the following technical solutions:

[0005] An open coal mine methane emission inversion estimation method based on unmanned aerial vehicle monitoring data, the method comprising:

[0006] S1, using an unmanned aerial vehicle equipped with a methane sensor and a meteorological instrument to plan a flight above the study mine area and collect methane concentration monitoring data containing position information and meteorological data;

[0007] S2, the research mining area is divided into a grid, a reverse Lagrangian random diffusion model is constructed, methane concentration monitoring data containing position information is extracted as an observation point, the reverse Lagrangian random diffusion model releases Np virtual particles from the observation point and randomly simulates the reverse motion trajectory of the particles in the research mining area to simulate the grid (x i , y i ) and the footprint sensitivity f j (x i , y i ) corresponding to the observation point j of the research mining area, and the footprint sensitivity of all grids in the research mining area is accumulated to obtain the transfer coefficient F j of the observation point j;

[0008] S3, the average methane concentration is calculated from the methane concentration monitoring data according to the local flight path k planned by the unmanned aerial vehicle, and the background concentration is subtracted to obtain the concentration difference ΔC k , the transfer coefficients of all observation points of the local flight path k are summarized to obtain F k , and then the average emission rate of the research mining area to the local flight path k is Q k , Δt k is the time interval of the methane sensor collecting methane concentration, the average emission rate of the related local flight path affected by the wind direction is obtained, and the average value Q0 is obtained; the total emission rate Q of the research mining area is obtained according to the following formula: total Q total =Q0xA, A is the coverage area of the research mining area.

[0009] In order to better realize the present application, in the method S1, the unmanned aerial vehicle is provided with a GPS positioning system or / and a Beidou positioning system, the unmanned aerial vehicle flies at a height h above the research mining area to plan a flight plane covering the ground area of the research mining area, a flight path covering the flight plane is planned on the flight plane, and the meteorological data contains temperature, humidity, air pressure and wind speed and direction information.

[0010] Preferably, the virtual particles released by the observation point in the reverse Lagrangian random diffusion model are used for reverse motion trajectory simulation, and the reverse Lagrangian random diffusion model is constructed as follows:

[0011] The calculation basis of the particle random diffusion model is the generalized Langevin equation, which is based on the following assumptions: the particle position x and the velocity u evolve together as a Markov process:

[0012] du i =a i (x, u, t)dt + b i，j (x, u, t)dξ j ,

[0013] dx i =ui dt (1)

[0014] where a i and b i，j are functions of (x, u, t), and dξ j is a random increment drawn from a Gaussian distribution with mean 0 and variance dt, and dξ i is independent of dξ j when i≠j;

[0015] In the backward time frame of the inverse Lagrangian stochastic diffusion model, using the time coordinate t' = T0-t, where T0is an arbitrary shift constant, assuming T0= 0, the particle velocity according to O' is:

[0016]

[0017] In the (x, u', t') coordinate system, based on the generalized Langevin equation, the expression of the particle velocity evolution trajectory is as follows:

[0018] du' i = a' i (x, u', t') dt' + b i,j (x, u', t') dξ j ,

[0019] dx i = u' i dt' (5)

[0020] A fundamental constraint on the Langevin equation coefficients is that the Eulerian velocity probability density function g' a (x, u', t') must satisfy the Fokker-Planck equation corresponding to equation (1):

[0021]

[0022] In the (x, u, t) coordinate system, it is noted that g' a (x, -u', -t') = g a (x, u, t), u' = -u, and

[0023]

[0024] where a' i and B' i，j take values at (x, -u', -t') = (x, u, t); requiring a' i and B' i，j to satisfy this Fokker-Planck equation, a backward model with complete mixing is obtained;

[0025] where B i，j represents b i，k j，k / 2; in g' a is Gaussian and B i，j is independent of u:

[0026]

[0027] Here u i is the instantaneous velocity of the particle, U i is the mean velocity of the background flow; since the amplitude of the particle's random velocity variation is the same in both time frames, we get:

[0028]

[0029] where is the Eulerian vertical velocity variance, and τ i is the Lagrangian decorrelation time scale of w;

[0030] Equations (4), (5) and (7) constitute the inverse Lagrangian stochastic dispersion model.

[0031] Preferably, in the method S3, the total emission rate Q total is obtained as follows:

[0032] The ensemble-averaged methane concentration at position x and time t due to an arbitrary emission source S (kg m -3 s -1 ) is given by:

[0033]

[0034] where P f (x, t | x0, t0) is the transition probability density, defined as p f (x, t | x0, t0) dx is the probability of finding a particle initially at (x0, t0) in the monitoring domain dx centered at x at time t; the calculation of P f (x, t | x0, t0) is a function of the Lagrangian model;

[0035] The particle is released at (x, t) and collected at (x0, t0), giving the estimate of the backward-in-time conditional probability density P b (x0, t0 | x, t):

[0036] P b (x0, t0 | x, t) = P f (x, t | x0, t0)

[0037] ​P b is used in concentration calculations f ;

[0038] For a continuous emission source, whose emission rate is uniform over a plane, the concentration C v (x) can be determined by calculating the ensemble average "residence time" of particles released at x0: that is, the average "residence time" of the set of particles released at x0that are within the monitoring domain V sens centered at x

[0039]

[0040] The time spent by particles released from V sens in the source volume V src centered at x0

[0041]

[0042] where t' = -t; for a stationary atmosphere, T b and T f have:

[0043]

[0044] In practice, C v (x) will be calculated from a sample average of individual particle residence times, yielding a volume average concentration C v (x) expressed as:

[0045]

[0046] where N particles are released uniformly from the monitoring domain V sens , denotes the residence time of an individual particle;

[0047] For a planar source with emission rate Q, consider it as a thin volume source composed of an infinitesimal height dz above the ground, and having an equivalent volume emission rate S = Q / dz. Then the contribution to T b from a particle that impacts the ground within the source boundary at vertical ground speed w0and passes through dz both upwards and downwards in one time step is simply 2dz / |w0|; the concentration can then be expressed as:

[0048]

[0049] where the summation is over all ground impacts within the source region; thus the concentration calculation uses the backward Lagrangian model to release particles uniformly within the monitoring domain V sens and sums the w0reciprocals over all ground impacts within the source region;

[0050] The sensitivity of the monitoring point to an arbitrary grid cell (x i , y j ) of the source region is defined as a footprint function, f

[0051]

[0052] This parameter can be obtained from the previously constructed backward trajectory model given the atmospheric stability, friction velocity u * , surface roughness length z0, boundary layer height H, number of virtual particles N p , time step Δt, grid resolution of the source region Δx, Δy, etc., then the emission rate Q has:

[0053] Q = C v (x) x f i，j (x i , y j ) (14)

[0054] In the flight path monitoring by the unmanned aerial vehicle, the line average concentration C L of the flight path is taken as the basis for the inversion of the emission rate Q; the footprint function is generalized: C L is simulated as the average value of the concentrations of P points distributed at equal intervals along the monitoring path; and a group of backward trajectories is calculated from each of the points, and the total sensitivity F of the monitoring flight path to all grid cells of the source region is obtained by averaging the point footprint functions:

[0055]

[0056] N is the total number of particles released from each measurement point, and the inner summation only includes landing events within the source;

[0057] At this time, the generalization of formula (14) is performed, and the emission rate Q has:

[0058] Q = C v (x) x F, C v (x) = C L - C0 (15); wherein C0 is the background concentration value.

[0059] Preferably, the random differential equation expression of the virtual particles in the inverse Lagrangian stochastic diffusion model with respect to the inverse time t is as follows:

[0060] du′ i = a′ i (x, u′, t′) dt′ + b i，j (x, u′, t′) dξ j , wherein a′​i and b i，j are functions of (x, u, t) and dξ j are random increments drawn from a Gaussian distribution with mean 0 and variance dt, and dξ i are independent of each other. j

[0061] Preferably, the footprints sensitivity f j (x i , y i ) in the footprints function simulated by the backward Lagrangian stochastic dispersion model is expressed as follows:

[0062] where N is the total number of particles released from each measurement point, and the inner summation only includes landing events within the source of the study mine area.

[0063] Preferably, the transfer coefficient F of the observation point j is expressed as follows:

[0064] where j is the label of the flight monitoring point, and P is the total number of monitoring points.

[0065] Preferably, the average value Q0 of the average emission rate of the relevant local flight line affected by the wind direction is expressed as follows: K is the total number of the relevant local flight lines affected by the wind direction.

[0066] Preferably, in the method S3, the total emission rate Q total represents the total emission of the study mine area per unit time, and the total emission rate Q total is in units of μg / s, and the total emission rate Q totol is used to calculate the monthly emission and the annual emission of the study area, respectively.

[0067] ​The application discloses an open-pit coal mine methane emission amount inversion estimation system based on unmanned aerial vehicle monitoring data, which comprises an unmanned aerial vehicle detection and collection system, a reverse Lagrange random diffusion model and a methane emission amount estimation module, wherein the unmanned aerial vehicle detection and collection system comprises an unmanned aerial vehicle loaded with a methane sensor and a meteorological instrument; a GPS positioning system or a Beidou positioning system is installed on the unmanned aerial vehicle and used for positioning position information; the unmanned aerial vehicle of the unmanned aerial vehicle detection and collection system plans a flight above a research mine area; the methane sensor is used for collecting methane concentration monitoring data; the meteorological instrument is used for collecting meteorological data; the unmanned aerial vehicle detection and collection system communicates and transmits the methane concentration monitoring data containing position information and the meteorological data to the reverse Lagrange random diffusion model; the reverse Lagrange random diffusion model is internally provided with a research mine area which is divided into grids and has position information; the methane concentration monitoring data containing position information is extracted as an observation point; the reverse Lagrange random diffusion model releases Np virtual particles from the observation point and simulates the reverse motion track of the particles in the research mine area by using the meteorological data to simulate the footprint sensitivity f(x,y) corresponding to the observation point j of the grid (x,y) of the research mine area, and accumulates the footprint sensitivities of all the grids of the research mine area to obtain the transfer coefficient F of the observation point j; the methane emission amount estimation module divides a flight route planned by the unmanned aerial vehicle into a plurality of local routes, calculates the average methane concentration of a local route k, and obtains a concentration difference value Delta Ck by subtracting a set background concentration; the transfer coefficients of all the observation points of the local route k are summarized to obtain F; and the average emission rate of the research mine area to the local route k is Q, wherein Delta Ck is the concentration difference value, F is the transfer coefficient, A is the coverage area of the research mine area, and Delta t is the time interval of the methane concentration frequency collected by the methane sensor. i i j i i j k k k total total

[0068] Compared with the prior art, the application has the following advantages and beneficial effects:

[0069] ​​​​​​​​​​​​(1) The present application collects and researches the methane concentration monitoring data containing position information and meteorological data above the mining area, and the inverse Lagrangian random diffusion model releases virtual particles as observation points with the methane concentration monitoring data and simulates the reverse motion trajectory of the particles in the research mining area by using the meteorological data to simulate the footprint sensitivity corresponding to the observation point j of the grid or pixel in the research mining area, realizes the reverse tracking of the position of the methane concentration above the research mining area to the ground of the research mining area, and inversely obtains the total emission rate of the methane emission of the research mining area, and can calculate the daily emission, monthly emission and annual emission of the research mining area.

[0070] (2) The present application realizes the accurate estimation of the total emission rate and emission amount of the research mining area by the inverse Lagrangian random diffusion model with the methane concentration monitoring data as the observation point for the reverse simulation in the research mining area, and is convenient for quantitatively evaluating the methane emission distribution and emission intensity of the research mining area, and provides data support for the carbon emission accounting and environmental monitoring of the research mining area.

[0071] (3) The present application fills the technical blank of the coal mine methane emission source tracing and emission amount quantitative estimation, can accurately inverse the methane emission distribution and emission intensity of the mining area, and improves the accuracy of methane detection and carbon monitoring activities. BRIEF DESCRIPTION OF DRAWINGS

[0072] Figure 1 It is a method flow diagram of the present application open coal mine methane emission inversion estimation method;

[0073] Figure 2 It is a schematic diagram of the equipment carried by the unmanned aerial vehicle of the embodiment;

[0074] Figure 3 It is a schematic diagram of the planned flight route of the embodiment example in the research mining area.

[0075] Among them, the name corresponding to the reference sign in the drawing is:

[0076] 1- unmanned aerial vehicle, 2- methane gas extraction module, 3- methane sensor, 4- meteorological instrument, 5- planned flight route, 6- compass, 7- wind vane, 8- methane emission source. DETAILED DESCRIPTION

[0077] The present application will be further described in detail in combination with the embodiments:

[0078] Embodiment

[0079] As shown in the figure, an open coal mine methane emission inversion estimation method based on unmanned aerial vehicle monitoring data, the method comprises: Figure 1

[0080] ​S1, using the unmanned aerial vehicle 1 equipped with a methane sensor 3 and a meteorological instrument 4 to plan flight over the study area and collect methane concentration monitoring data and meteorological data containing position information. In some embodiments, the unmanned aerial vehicle is equipped with a GPS positioning system or / and a Beidou positioning system (to obtain the latitude and longitude information of the device), as shown in Figure 2 The unmanned aerial vehicle 1 is equipped with a methane gas extraction module 2, a methane sensor 3 (or methane gas analyzer), and a meteorological instrument 4 (or meteorological module). The unmanned aerial vehicle 1 is a DJI Weiyu 350RTK model with a maximum take-off weight of 9 kg and a maximum wind speed of 12 m / s, and is equipped with a fixed frame that can carry a methane sensor and a meteorological instrument. The methane gas extraction module of the methane gas extraction module 2 is integrated with the methane sensor 3 and realizes air extraction and detection of methane concentration in the air. It uses tunable diode laser absorption spectroscopy technology with a range of 0-10000 ppm, a resolution of 1 ppm, a detection limit of 1 ppm, a response time of 1 s, and is connected to the meteorological instrument and the unmanned aerial vehicle device through a power line data transmission line. The unmanned aerial vehicle 1 flies at a height h above the methane emission source 8 (emission source inside the study area, or open coal mine working surface) to plan a flight plane covering the ground area of the study area. The flight plane covers the flight route, and the meteorological data includes temperature, humidity, air pressure, and wind speed and direction information. As shown in Figure 3 The flight height h is preferably 50-80 meters above the methane emission source 8 (i.e. open coal mine working surface), and the flight plane is preferably at least 100 meters outwardly expanded from the methane emission source 8 (or open coal mine working surface) (the specific distance depends on the situation of the study area), and the planned flight area is larger than the methane emission source 8 or open coal mine working surface, i.e. the planned flight area is located in the peripheral area vertically above the methane emission source 8, which facilitates the inverse motion trajectory simulation of the inverse Lagrangian random diffusion model in the study area. Referring to Figure 3 As shown in the figure, the embodiment example constructs a flight route for the unmanned aerial vehicle 1 in the area vertically above the methane emission source 8 or open coal mine working surface expanded outwardly by 100 meters, and the example flight route is a rectangular flight route as shown in Figure 3 The starting and ending points are the northeast corner of the rectangle, and regardless of the wind direction, the compass 3, wind vane 7, planned flight route 5, and methane emission source 8 are required to follow Figure 3The example shows that the flight path is aligned with the four cardinal directions (east, west, south, and north). This facilitates the determination of flight paths affected by wind direction. For example, if the wind is from the southwest, the affected flight paths are east and north. In the reverse motion trajectory simulation of the inverse Lagrange random diffusion model, the methane sensor 3 of the UAV within the east and north flight paths will receive methane concentration monitoring data. The south and west flight paths are in the upwind area, and the methane sensor 3 of the UAV within the south and west flight paths receives very little methane concentration monitoring data, so it is not considered in this embodiment. Only the flight paths affected by wind direction are considered. The flight altitude is 50 meters relative to the working surface in the example. The UAV conducts the experiment according to the planned flight path. During the execution, the methane sensor 3 extracts the surrounding air every 1 second to monitor the methane content (ppm); the meteorological instrument 4 obtains the current temperature, humidity, air pressure, wind speed, and wind direction data; the GPS synchronously obtains the current latitude and longitude information; the UAV 1 is also equipped with a 4G module, which transmits data in real time.

[0081] S2. Divide the study mining area into grids, construct a reverse Lagrange random diffusion model, extract methane concentration monitoring data containing location information as observation points, release Np virtual particles from the observation points using the reverse Lagrange random diffusion model, and use meteorological data to randomly simulate the reverse motion trajectory of the particles within the study mining area to obtain the grid (x) of the study mining area. i y i The footprint sensitivity f corresponding to observation point j j (x i y i The transfer coefficient F of observation point j is obtained by accumulating the sensitivity of all grid footprints in the research mining area. j .

[0082] In some embodiments, the footprint sensitivity f in the footprint function simulated by the inverse Lagrange random diffusion model j (x i y i The expression is as follows:

[0083] Where N is the total number of particles released from each measuring point, and the internal summation only includes particles corresponding to landing events or wind-affected flight paths within the research mining area. The wind direction influence determination is as follows: The weather instrument 4 of UAV 1 monitors meteorological data, including temperature, humidity, air pressure, wind speed, and wind direction information. As the UAV flies along the flight path, it obtains wind direction information and determines whether it affects the flight path. If the flight path is affected by wind direction, for example, if the wind is northerly (blowing from the north), then the southbound flight path is affected by the wind direction. Figure 3In the route example, the route direction is aligned with the four directions of east, west, north and south (so as to facilitate the route judgment affected by the wind direction, such as the wind direction is southwest, then the route affected by the wind direction is the east route and the north route, in the reverse trajectory simulation of the inverse Lagrangian random diffusion model, the methane sensor 3 of the unmanned aerial vehicle in the east route and the north route range will receive the methane concentration monitoring data, the south route and the west route are in the upwind area, the methane sensor 3 in the range of the south route and the west route receives little methane concentration monitoring data, which is not considered in this embodiment, and only the wind direction related route is considered.

[0084] In some embodiments, the transfer coefficient F of the observation point j is expressed as follows:

[0085] Where j is the route monitoring point mark, and P is the total number of monitoring points.

[0086] In some embodiments, the virtual particles released by the observation points in the inverse Lagrangian random diffusion model are used to simulate the reverse trajectory, and the inverse Lagrangian random diffusion model is constructed as follows:

[0087] The calculation basis of the particle random diffusion model is the generalized Langevin equation, which is based on the following assumptions: the particle position x and the velocity u evolve together as a Markov process:

[0088] du i =a i (x,u,t)dt+b i，j (x,u,t)dξ j ,

[0089] dx i =u i dt (1)

[0090] Where a i and b i，j are functions of (x, u, t), and dξ j is a random increment selected from a Gaussian distribution with mean 0 and variance dt, and when i≠j, dξ i is independent of dξ j ; this embodiment discretizes formula (1) and uses it to calculate a large number of particle trajectories emitted from the emission source, and then determines the volume average concentration according to the residence time of the particles in the monitoring point V sens (range for a continuous emission source).

[0091] One of the essential constraints of the Langevin equation coefficient is the Euler velocity probability density function g a(x, u, t) must satisfy the Fokker-Planck equation (FPE) corresponding to equation (1):

[0092]

[0093] where B i,j represents b i，k (x, u, t) = -B j，k / 2; in g a (x, u, t) = -B i，j (x, u, t) = -B

[0094]

[0095] u i is the instantaneous velocity of a particle, U i is the mean velocity of the background flow

[0096]

[0097] In the backward time frame of the inverse Lagrangian stochastic diffusion model, the time coordinate t' = T0- t is used, where T0is an arbitrary shift constant. Assuming T0= 0, the particle velocity in terms of O' is:

[0098]

[0099] In the (x, u', t') coordinate system, the expression of the particle velocity evolution trajectory is given based on the generalized Langevin equation as follows:

[0100] du' i = a' i (x, u', t') dt' + b i,j (x, u', t') dξ j ,

[0101] dx i = u' i dt' (5)

[0102] A fundamental constraint on the Langevin equation coefficients is that the Eulerian velocity probability density function g' a (x, u', t') must satisfy the Fokker-Planck equation corresponding to equation (1):

[0103]

[0104] In the (x, u, t) coordinate system, it is noted that g' a (x, -u', -t') = g a (x, u, t), u' = -u, and

[0105]

[0106] where a′ i and B′ i，j take values at (x, -u′, -t′) = (x, u, t); require a′ i and B′ i，j satisfy this Fokker-Planck equation, a backward model of sufficient mixing is obtained;

[0107] where B i，j denotes b i，k b j，k / 2; the special solution under the condition that g′ a is Gaussian distribution and B i，j is independent of u is:

[0108]

[0109] Here u i is the instantaneous velocity of the particle, U t is the mean velocity of the background flow; since the amplitude of the random velocity variation of the particle is the same under the two time scales, the following formula is obtained:

[0110]

[0111] where is the Eulerian vertical velocity variance, and τ l is the Lagrangian decorrelation time scale of w;

[0112] The inverse Lagrangian stochastic dispersion model is constituted by the joint of formulas (4), (5) and (7).

[0113] S3, calculating the average methane concentration from the methane concentration monitoring data of the local flight path k planned by the unmanned aerial vehicle, and subtracting the set background concentration to obtain a concentration difference ΔC k , collecting the transfer coefficients of all observation points of the local flight path k to obtain F k , and then the average emission rate of the study mine area to the local flight path k is Q k , Δt k is the time interval of the methane sensor collecting the methane concentration frequency, the average emission rate of the relevant local flight path affected by the wind direction is obtained, and the average value Q0 is obtained. In some embodiments, the unmanned aerial vehicle plans a flight path over the study mine area to obtain a plurality of local flight paths, the average value Q0 of the average emission rate of the relevant local flight path affected by the wind direction (the wind direction affected can use the wind direction affected judgment method, or can use the inverse Lagrangian stochastic dispersion model to judge the associated flight path corresponding to the particle landing event) is expressed as follows: K is the total number of local routes affected by wind direction. The judgment of whether the route is affected by the wind direction is as follows: the meteorological instrument 4 of the unmanned aerial vehicle monitors meteorological data, which contains temperature, humidity, air pressure, wind speed and wind direction information. When the unmanned aerial vehicle flies along the route, the wind direction information is obtained to judge whether the route is affected. If the route is affected by the wind direction, for example, the wind direction is north wind (blowing from the north), then the south route is affected by the wind direction. In Figure 3 In the route example, the route direction is aligned with the four directions of east, west, south and north (so as to facilitate the judgment of the route affected by the wind direction, for example, the wind direction is southwest wind, and the routes affected by the wind direction are the east route and the north route. In the reverse motion trajectory simulation of the inverse Lagrangian stochastic diffusion model, the methane sensor 3 of the unmanned aerial vehicle in the east route and the north route range will receive the methane concentration monitoring data. The south route and the west route are in the upwind area, and the methane sensor 3 of the unmanned aerial vehicle in the south route and the west route range receives little methane concentration monitoring data. In this embodiment, only the routes affected by the wind direction are considered.

[0114] In some embodiments, in method S3, the total emission rate Q total The method is as follows:

[0115] Since any emission source S (kg m -3 s -1 The ensemble average methane concentration expression at the x position and time t is as follows:

[0116]

[0117] Where P f (x,t|x0,t0) is the transition probability density, defined as P f (x,t|x0,t0)dx is the probability of finding a particle initially at (x0,t0) in the monitoring domain dx centered at x at time t; P f (x,t|x0,t0) is a function of the Lagrangian model;

[0118] The particle is released at (x,t) and collected at (x0,t0), which gives the estimate of the backward-in-time conditional probability density P b (x0,t0|x,t);

[0119] P b (x0,t0|x,t) = P f (x,t|x0,t0)

[0120] P b is used to replace P f in concentration calculation;

[0121] For a continuous emission source, its emission rate is uniform in the plane, and the concentration Cv (x) can be determined by calculating the ensemble average "residence time" of particles released at xo: i.e. the average "residence time" of the ensemble of particles released at xo that end up in the monitoring domain V centered at x sens

[0122]

[0123] from V sens released at xo that end up in the source volume V centered at xo src

[0124]

[0125] where t' = -t; for a stationary atmosphere, T b and T f have:

[0126]

[0127] In practice, C v (x) is calculated from the sample average of individual particle residence times, C v (x) is expressed as:

[0128]

[0129] where N particles are released uniformly from the monitoring domain V sens represents the residence time of an individual particle;

[0130] For a source with emission rate Q, consider it as a thin volume source consisting of an infinitesimal height dz above the ground, with equivalent volume emission rate S = Q / dz. Then the contribution to T b from a particle that impacts the ground within the source boundary at vertical ground speed w0 and passes through dz both upwards and downwards in one time step is 2dz / |w0|; the concentration can then be expressed as:

[0131]

[0132] where the summation is over all ground impacts within the source region (simulated by backtracking trajectories in a reverse Lagrangian stochastic dispersion model, where the flight path of a ground impact particle is the relevant flight path affected by the wind direction); the concentration calculation therefore uses the backtracking Lagrangian model to release particles uniformly from the monitoring domain V sens and sum the inverse w0 over all ground impacts within the source region;

[0133] define C as the concentration at the monitoring point x​​​i , y j The sensitivity of the monitor point to all grid cells in the source region is the footprint function, F(x, y), then the total sensitivity of the monitor point to all grid cells in the source region is:

[0134]

[0135] This parameter can be given by the atmospheric stability, friction velocity u * , surface roughness length z0, boundary layer height H, number of virtual particles N p , time step At, source grid resolution Ax, Ay, etc. Given by the previously constructed backward trajectory model, then the emission rate Q is:

[0136] Q = C v (x) x F(x, y) (14) i，j i , y j

[0137] In the flight path monitoring by the UAV, the line average concentration C L is taken as the basis for the inversion of the emission rate Q; the footprint function is generalized: C L is simulated as the average of P point concentrations distributed equidistantly along the monitoring path; and a group of backward trajectories is calculated from each of the points, and the total sensitivity F of the monitoring flight path to all grid cells in the source region is obtained by averaging the point footprint functions:

[0138]

[0139] N is the total number of particles released from each measurement point, and the inner summation only includes landing events within the source;

[0140] At this time, the formula (14) is generalized, and the emission rate Q (i.e., the total emission rate Q total ) is:

[0141] Q = C v (x) x F(x, y) (14) v (x) = C L - C0 (15); wherein c0 is the background concentration value.

[0142] In some embodiments, the stochastic differential equation expression of the virtual particles in the inverse Lagrangian stochastic dispersion model with respect to the inverse time t is as follows:

[0143] du′ i = a′ i (x, u', t') dt' + b i，j (x, u', t') dξ j , wherein a′ i and b i，j ​​is a function of (x, u, t) and dξ j is a random increment drawn from a Gaussian distribution with mean 0 and variance dt, and dξ i and dξ j are independent of each other.

[0144] The total emission rate Q of the study mining area is obtained according to the following formula total : Q total = Q0 x A, A is the coverage area of the study mining area. In some embodiments, the total emission rate Q total represents the total emission of the study mining area per unit time, and the total emission rate Q total is μg / s, and the monthly emission and annual emission of the study area are calculated using the total emission rate Q total respectively. In this embodiment, the monitoring frequency of the methane sensor 3 on the unmanned aerial vehicle is 1 second. The monthly emission (tons / month) is expressed as follows:

[0145] In the formula, Q [μg / s] of Q is Q total , and the unit is μg / s.

[0146] The annual emission (tons / month) is expressed as follows:

[0147] In the formula, Q [μg / s] of Q is Q totol , and the unit is μg / s.

[0148] An open-pit coal mine methane emission inversion estimation system based on unmanned aerial vehicle monitoring data, comprising an unmanned aerial vehicle detection and collection system, a reverse Lagrangian random diffusion model and a methane emission estimation module. The unmanned aerial vehicle detection and collection system comprises an unmanned aerial vehicle carrying a methane sensor and a meteorological instrument. A GPS positioning system or a Beidou positioning system is installed on the unmanned aerial vehicle for positioning position information. The unmanned aerial vehicle of the unmanned aerial vehicle detection and collection system plans to fly over the study mining area. The methane sensor is used to collect methane concentration monitoring data, and the meteorological instrument is used to collect meteorological data. The unmanned aerial vehicle detection and collection system communicates and transmits the methane concentration monitoring data containing position information and the meteorological data to the reverse Lagrangian random diffusion model. The reverse Lagrangian random diffusion model has a grid division and a study mining area with position information. The methane concentration monitoring data containing position information is extracted as an observation point. The reverse Lagrangian random diffusion model releases Np virtual particles from the observation point and simulates the reverse motion trajectory of the particles in the study mining area using the meteorological data to simulate the footprint sensitivity f i (x i , y j ) corresponding to the observation point j of the study mining area grid (x i , y iThe transfer coefficient F of observation point j is obtained by accumulating the sensitivity of all grid footprints in the research mining area. j The methane emission estimation module divides the planned flight path of the UAV into several local flight paths and calculates the average methane concentration of local flight path k. The difference between this average and the set background concentration is then used to obtain the concentration difference ΔC. k The transfer coefficients of all observation points on the local route k are summed to obtain F. k The average emission rate of the mining area to local flight path k is Q. k , Δt k The time interval for the methane sensor to collect methane concentration frequencies is used to obtain the average emission rate of relevant local flight paths affected by wind direction, and the average value Q0 is calculated; the methane emission estimation module obtains the total emission rate Q of the study mining area according to the following formula. total Q total =Q0×A, where A is the area covered by the research mining area.

[0149] In this example, the uniform emission source A in an open-pit coal mine in Inner Mongolia is an irregular plane. Photogrammetry determined its area to be approximately A = 5.602 × 10⁵ m². 2 Drone flight path planning: (See attached...) Figure 2 Flight path planning was conducted using a specific method. The drone was tested along the planned flight path: the drone took off and flew along the set route. During the flight, a methane sensor extracted ambient air every 1 second to monitor methane content (ppm); a meteorological instrument acquired current temperature, humidity, air pressure, wind speed, and wind direction data; on the day of the experiment, the wind direction was measured to be northeast, and the wind speed was U = 4.5 m / s. The average methane concentration measured in the field for the four flight paths was C... N =1.9334ppm, C S =1.9353ppm,C W =1.9489ppm, C E =1.9610ppm. Background concentration C0 =1.909ppm. The relevant parameters of the inverse Lagrange stochastic diffusion model are set to commonly used default values ​​to establish the parameter system for turbulent diffusion and inverse simulation, namely: atmospheric stability classification is neutral (Monin-Obukhov stability type D), friction velocity u... * =0.3m / s, surface roughness length z0 = 0.1m, boundary layer height H = 1000m, virtual particle number N p =20000, time step Δt = 1s, source region grid resolution Δx, Δy = 10m × 10m. Since the wind direction was northeast on the day of the experiment, data from the main methane emission directions (south and west routes) were selected for inversion; the average emission rate of the south route is as follows: Q S =ΔC S F S= 10.99 pgm -2 s -1 ; the average emission rate of the west route is as follows: Q W = AC W F W = 11.02 pgm -2 s -1 .

[0150] The emission rates of the two directions are integrated with the route length proportion as the weight:

[0151] L S and L W respectively represent the emission weight coefficients of the south region of the study mine and the west region of the study mine.

[0152] The total emission rate of the study mine: Q totld = Q0x A = 6.18x 10 6 pg / s.

[0153] The monthly emission amount (tons / month) is as follows:

[0154]

[0155] The annual emission amount (tons / month) is as follows:

[0156]

[0157] The above merely describes the preferred embodiments of the present application and is not used to limit the present application, and any modification, equivalent replacement, and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An open-pit coal mine methane emission inversion estimation method based on unmanned aerial vehicle monitoring data, characterized in that: The method comprises: S1, using a UAV equipped with a methane sensor and a meteorological instrument to plan flight over the study mine area and collect methane concentration monitoring data and meteorological data containing position information; S2, the research mining area is divided into grid, the inverse Lagrangian random diffusion model is constructed, the methane concentration monitoring data containing position information is extracted as observation point, the inverse Lagrangian random diffusion model releases Np virtual particles at observation point and utilizes meteorological data to simulate the reverse motion trajectory of particles in the research mining area, the grid (x i , y i ) of the research mining area corresponding to the footprint sensitivity f j (x i , y i ) of observation point j is simulated, and the transfer coefficient F j of observation point j is accumulated by the footprint sensitivity of all grids in the research mining area. S3、from methane concentration monitoring data, the average methane concentration of the local flight route k planned by the unmanned aerial vehicle is calculated, and the concentration difference ΔC is obtained by subtracting the set background concentration k , the transfer coefficient of all observation points of the local flight route k is summarized to obtain F k , the average emission rate of the study mine area to the local flight route k is Q k , Δt k is the time interval of the methane concentration collected by the methane sensor, the average emission rate of the relevant local flight route affected by the wind direction is obtained, and the average value Q0 is obtained; the total emission rate Q of the study mine area is obtained according to the following formula total : Q total = Q0×A, A is the coverage area of the study mine area.

2. The method for estimating open-pit coal mine methane emission volume based on unmanned aerial vehicle monitoring data according to claim 1, characterized in that: In method S1, the UAV is equipped with a GPS positioning system or / and a Beidou positioning system, the UAV plans a flight plane covering the ground area of the study mine area at a flight height h over the study mine area, plans a route covering the flight plane, and the meteorological data contains temperature, humidity, air pressure and wind speed and direction information.

3. The method of claim 1, wherein the method further comprises: In the reverse Lagrangian stochastic dispersion model, virtual particles are released from observation points to simulate reverse motion trajectories, and the reverse Lagrangian stochastic dispersion model is constructed as follows: The calculation basis of the particle random diffusion model is the generalized Langevin equation, which is based on the following assumptions: the particle position x and velocity u evolve together as a Markov process: du i = a i (x, u, t)dt + b i，j (x, u, t)d\x j , dx i = u i dt (1) where a i and b i，j are functions of (x, u, t) and dξ j is a random increment drawn from a Gaussian distribution with mean 0 and variance dt, and dξ i is independent of dξ j when i≠j. In the backward time frame of the reverse Lagrangian stochastic dispersion model, the time coordinate t' = T0-t is used, where T0 is an arbitrary transformation constant, and if T0 = 0, then according to the particle velocity of O' is: In the (x, u', t') coordinate system, based on the generalized Langevin equation, the expression of the particle velocity evolution trajectory is as follows: du' i = a' i (x, u', t') dt' + b i,j (x, u', t') dξ j , dx i = u' i dt' (5) One essential constraint on the coefficients of the Langevin equation is that the probability density function g' of the Euler velocity must be positive a (x, u', t') must satisfy the Fokker-Planck equation corresponding to equation (1): In the (x, u, t) coordinate system, note that g' (x, u, t) = g(x, -u, -t) a (x, -u', -t') = g a (x, u, t), u' = -u, and where a' = a - b i and B' = B + b i，j at (x, -u', -t') = (x, u, t); requiring a' > 0 i and B' > 0 i，j satisfying this Fokker-Planck equation, a perfectly mixing backward model is obtained; where B i，j represents b i，k b j，k / 2; in g′ a is Gaussian and B i，j independent of u: Here u i is the instantaneous velocity of the particle, U i is the average velocity of the background flow; since the amplitude of the particle's random velocity variation is the same under both time frames, we get the following equation: where is the Eulerian vertical velocity variance, while τl is the Lagrangian decorrelation timescale of w. Formulas (4), (5) and (7) constitute the reverse Lagrangian stochastic dispersion model.

4. The method of claim 1, wherein the method further comprises: In method S3, the total exhaust rate Q total The method is obtained as follows: The ensemble average methane concentration at position x and time t due to any emission source S (kg m -3 s -1 ) is expressed as follows: where P f (x, t \ x0, t0) is the transition probability density defined as P f (x, t \ x0, t0) dx is the probability of finding a particle initially at (x0, t0) in the monitoring domain dx centered at x at time t; P f (x, t \ x0, t0) is a function of the Lagrangian model; The particles are released at (x, t) and collected at (x0, t0), giving the backward-in-time conditional probability density P b Estimation of (x0, t0 | x, t): P b (x0, t0 | x, t) = P f (x, t | x0, t0) P is used in the concentration calculation instead of P b P is used in the concentration calculation instead of P f ; For a continuous emission source, its emission rate is uniform over a plane, and the concentration C v (x) can be determined by calculating the population average "residence time" of particles released at x0that are in the set of monitoring domains V sens with center at x From V sens The particles released stay inside the source volume V src for a time where t' = -t; for a stationary atmosphere, T b and T f have: In practice, C v (x) The volume average concentration C v (x) is calculated from the sample average of the individual particle residence times. wherein the monitored domain V sens uniformly release N particles, denotes the residence time of a single particle; For a source with emission rate Q, consider it as a thin volume source of infinitesimal height dz above the ground and with equivalent volume emission rate S = Q / dz. Then the contribution of a particle to T b when it impacts the ground within the source boundary with a vertical ground speed w0and passes up and down through dz in one time step is 2dz / |w0|; the concentration can be expressed as: where the summation is over all landing events within the source region; thus the concentration calculation uses a backward Lagrangian model with uniform release of particles within the monitoring domain V sens and summing the inverse of w0 over all landing events within the source region; The sensitivity of the monitoring point to an arbitrary grid cell (x i , y j ) of the source region is defined as a footprint function, then the total sensitivity of the monitoring point to all grid cells of the source region is: This parameter can be given in terms of atmospheric stability, friction velocity u * , surface roughness length z0, boundary layer height H, virtual particle number N p , time step Δt, source region grid resolution Δx, Δy, etc. given by the previously constructed back-trajectory model, then the emission rate Q has: Q = C v (x) x f i，j (x i , y j ) (14) In the case of route monitoring by the UAV, the line average concentration C L as a basis for inverting the emission rate Q; Generalization of the footprint function: C L The simulation is the average of P point concentrations distributed equally along the monitoring path; and from each of these points a set of back trajectories is calculated, the footprint function is averaged over the set of points, and the total sensitivity F of the source region to the monitoring flight line is obtained: N is the total number of particles released from each measurement point, and the inner summation only includes landing events within the source; At this time, formula (14) is extended, and the emission rate Q is: Q = C v (x) x F, C v (x) = C L - c0(15); wherein c0is a background concentration value.

5. The method for estimating open-pit coal mine methane emission volume based on unmanned aerial vehicle monitoring data according to claim 1, characterized in that: The stochastic differential equation expression of the virtual particle in the reverse Lagrangian stochastic dispersion model with reverse time t is as follows: du' i = a' i (x, u', t') dt' + b i，j (x, u', t') dξ j where a' i and b i，j are functions of (x, u, t), and dξ j is a random increment taken from a Gaussian distribution with mean 0 and variance dt, and dξ i and dξ j are independent of each other when i≠j.

6. The method of claim 4, wherein the method further comprises: Footprint sensitivity f in the footprint function of the backward Lagrangian stochastic dispersion model simulation j (x i , y i ) is expressed as follows: where N is the total number of particles released from each measurement point, and the inner sum includes only landing events within the source of the study area.

7. The method of estimating open cut coal mine methane emissions from UAV monitoring data according to claim 6, characterized in that: The transfer coefficient F of the observation point j is expressed as follows: where j is the leg monitoring point label and P is the total number of monitoring points.

8. The method for estimating open-pit coal mine methane emission volume based on unmanned aerial vehicle monitoring data according to claim 1 or 2, characterized in that: The UAV divides the flight path planned above the mining area into several local paths, and the average value Q0 of the average emission rate of the relevant local paths affected by the wind direction is expressed as follows: K is the total number of the relevant local paths affected by the wind direction.

9. The method for estimating open-pit coal mine methane emission volume inversion based on unmanned aerial vehicle monitoring data according to claim 1, characterized in that: In method S3, the total emission rate Q total represents the total amount of emissions per unit of time, in seconds, of the study mine, the total emission rate Q total in μg / s, using the total emission rate Q total The monthly and annual emissions of the study area are calculated, respectively.

10. An open-pit coal mine methane emission inversion estimation system based on unmanned aerial vehicle monitoring data, characterized in that: The system comprises a drone detection and acquisition system, a reverse Lagrange random diffusion model, and a methane emission estimation module. The drone detection and acquisition system includes a drone equipped with a methane sensor and a meteorological instrument. The drone is equipped with a GPS positioning system and / or a BeiDou positioning system for location information. The drone plans its flight over the study mining area. The methane sensor collects methane concentration monitoring data, and the meteorological instrument collects meteorological data. The drone detection and acquisition system communicates and transmits the methane concentration monitoring data, including location information, and the meteorological data to the reverse Lagrange random diffusion model. The reverse Lagrange random diffusion model constructs a gridded study mining area with location information. The methane concentration monitoring data containing location information is extracted as observation points. The reverse Lagrange random diffusion model releases Np virtual particles at the observation points and uses meteorological data to randomly simulate the inverse motion trajectory of the particles within the study mining area to obtain the grid (x) of the study mining area. i y i The footprint sensitivity f corresponding to observation point j j (x i y i The transfer coefficient F of observation point j is obtained by accumulating the sensitivity of all grid footprints in the research mining area. j The methane emission estimation module divides the planned flight path of the UAV into several local flight paths and calculates the average methane concentration of local flight path k. The difference between this average and the set background concentration is used to obtain the concentration difference value ΔC. k The transfer coefficients of all observation points on the local route k are summed to obtain F. k The average emission rate of the mining area to local flight path k is Q. k , Δt k The time interval for methane concentration frequency acquisition by the methane sensor is used to obtain the average emission rate of relevant local flight paths affected by wind direction and calculate the average value Q0; the methane emission estimation module obtains the total emission rate Q of the study mining area according to the following formula. total Q total =Q0×A, where A is the area covered by the research mining area.

Images