Wide-area gas tracing method and device based on multistage sensing asynchronous cooperative triggering, medium and product

Abstract

The invention discloses a wide-area gas traceability method and device based on multistage sensing asynchronous cooperative triggering, a medium and a product, and relates to the field of wide-area gas leakage traceability, and the method comprises the steps: carrying out the adaptive three-dimensional grid division of a to-be-detected region; environment background information is detected; smoke plume discovery is carried out by adopting a random walk strategy and introducing a weight avoidance mechanism; determining a threshold value of binary detection according to the mean value of the environment background noise, and carrying out binary calculation on the sampled data; determining a near-field region according to a binary calculation result; performing concentration tracking according to the near-field region, and performing near-field state evaluation in real time; adjusting the size of the near-field region according to an evaluation result; continuously judging whether to exit the concentration detection of the near-field region; when the concentration detection of the near-field area is quitted and switched to the far-field area, a binary detection cycle is entered; determining a concentration peak value according to the concentration detection data of all the near-field regions; according to the invention, efficient, accurate and low-cost pollution source positioning in a complex wide-area environment can be realized.

Description

Technical Field

[0001] This application relates to the field of wide-area gas leak tracing, and in particular to a wide-area gas leak tracing method, equipment, medium and product based on multi-level sensor asynchronous collaborative triggering. Background Technology

[0002] In the field of wide-area gas leak tracing (e.g., industrial zone environmental monitoring, emergency response, pollution source investigation), existing technologies mainly rely on single-type gas sensor systems for detection and location. These systems typically suffer from the following significant problems:

[0003] (1) High energy consumption and cost bottlenecks In order to achieve effective traceability, the system often needs to continuously operate high-precision and high-selectivity gas sensors. These sensors consume a lot of power, which leads to a significant reduction in the flight time of the platform (such as a drone), and high overall system operating and maintenance costs.

[0004] (2) Difficulty in balancing detection accuracy and efficiency: While a single high-precision sensor strategy offers relatively high positioning accuracy, its slow response speed and limited coverage make it difficult to quickly locate pollution sources over a large area, resulting in low efficiency. In a single low-precision sensor strategy, although low-precision (or cross-sensitive) sensors offer fast response, low cost, and ease of large-scale deployment, their poor selectivity and susceptibility to environmental background interference lead to high false alarm and false negative rates, resulting in insufficient positioning accuracy. Commonly used binarized region search methods struggle to distinguish between near and far-field environmental differences.

[0005] (3) Low resource utilization efficiency: Existing methods usually use a fixed grid division strategy (such as uniform grid) for region search, which cannot dynamically adjust the detection resolution according to the size of the area to be tested and the complexity of the environment, resulting in oversampling in open areas or undersampling in complex areas, leading to waste of time and resources.

[0006] (4) Lack of dynamic coordination mechanism: Existing technologies fail to effectively combine the rapid, wide-area scanning capability of low-precision sensors with the precise positioning capability of high-precision sensors. They lack an intelligent linkage mechanism that dynamically triggers and switches the operating modes of different sensors based on environmental conditions (such as plume detection and concentration gradient changes). Especially in mobile platform applications (such as drones), achieving rapid, accurate, and robust wide-area gas source tracing and positioning while ensuring energy efficiency has become a critical technological bottleneck that urgently needs to be overcome.

[0007] Therefore, there is an urgent need for a new gas source tracing method that can effectively solve the problems of high energy consumption, low accuracy, resource waste and lack of intelligent collaborative mechanism, and achieve efficient, accurate and low-cost pollution source location in complex and wide-area environments. Summary of the Invention

[0008] The purpose of this application is to provide a wide-area gas source tracing method, device, medium and product based on multi-level sensing asynchronous collaborative triggering, which can achieve efficient, accurate and low-cost pollution source location in complex wide-area environments.

[0009] To achieve the above objectives, this application provides the following solution:

[0010] In a first aspect, this application provides a wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering, the wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering includes:

[0011] A three-dimensional mesh generation method and an adaptive mesh mechanism are used to adaptively generate a three-dimensional mesh for the area to be measured, and the mesh generation result is obtained.

[0012] Based on the grid division results and sampling data, the environmental background information is detected to obtain the mean value of the environmental background noise;

[0013] The area to be measured is considered as the far field, and a random walk strategy and a weight avoidance mechanism are introduced to detect plumes.

[0014] The threshold for binary detection is determined based on the mean value of the background noise; and the sampling data is then subjected to binary calculation using the threshold for binary detection.

[0015] Based on the binary calculation results, the near-field region is determined according to the number of points of the measured data points;

[0016] Concentration tracking is performed in the near-field region, and an improved gradient calculation method and relative boundary distance are used to evaluate the near-field status in real time; the size of the near-field region is adjusted according to the evaluation results; and concentration detection is continuously judged to determine whether to exit the near-field region.

[0017] When switching from concentration detection in the near field region to the far field region, an improved Gaussian mixture model is used to preserve historical distribution characteristics, and then the binary detection loop is entered.

[0018] Determine whether to re-enter the near-field region based on real-time binary detection data;

[0019] Based on the concentration detection data of all near-field areas, the concentration peak is determined; and the concentration peak is used as the source location.

[0020] Optionally, the step of using a three-dimensional mesh generation method and an adaptive mesh mechanism to adaptively generate a three-dimensional mesh for the area to be measured, and obtaining the mesh generation result, specifically includes:

[0021] Determine the volume of the region to be measured based on its three-dimensional dimensions;

[0022] Using formula Determine the cube mesh size d;

[0023] Using the cubic mesh size, an adaptive 3D mesh is generated for the area to be measured, and the mesh generation result is obtained.

[0024] Where V is the volume of the region to be measured, V threshold V is the volume threshold. max For the large spatial domain threshold, V min For the small spatial domain threshold, d max For the maximum grid size, d min α is the minimum grid size, α is the size attenuation coefficient, and k is the linear adjustment coefficient.

[0025] Optionally, the step of determining the threshold for binary detection based on the mean of the ambient background noise, and performing binary calculations on the sampled data using the threshold for binary detection, specifically includes:

[0026] The threshold T for binary detection is determined using the formula T = μ + 2σ;

[0027] Using formula Determine the output of the binary calculation;

[0028] Where μ is the mean value of the ambient background noise. x i Let be the i-th sampled data, n be the number of sampled data, and σ be the standard deviation. s j Let j be the j-th data point sampled in real time.

[0029] Optionally, based on the binary calculation results, the near-field region is determined according to the number of measured data points, specifically including:

[0030] Based on the binary calculation results, determine the set of coordinates of the data points where the gas was detected;

[0031] The spatial distribution characteristics of gas points are determined based on the set of point coordinates; the spatial distribution characteristics of gas points include: spatial range and point set density;

[0032] Based on the spatial distribution characteristics of gas points, using the formula Determine the adaptive near-field size D; where R x R y and R z Let ρ be the spatial range, ρ be the point set density, and β be the density compensation factor.

[0033] Based on the adaptive near-field size, the directional normalized range weighting method is used to adjust the gas distribution directional characteristics.

[0034] The near-field region is determined based on the gas distribution direction characteristics.

[0035] Optionally, the concentration tracking based on the near-field region involves using an improved gradient calculation method and relative boundary distance to perform real-time near-field state assessment; and adjusting the near-field region size based on the assessment results; specifically including:

[0036] Using formula Determine the concentration slope i ;

[0037] Using formula Determine the border distance (border distance);

[0038] The size of the near-field region is adjusted based on the concentration slope, boundary distance, and the number of data points with 0 concentration detected in the current near-field region.

[0039] Among them, C i C i+1 Let i be the concentration of data point i and data point i+1. Let i be the three-dimensional coordinates of the sampling point. Let D be the three-dimensional coordinates of sampling point i+1, ||·|| be the Euclidean distance, and D be the distance between the sampling points. k Let k be the length of each axis in the near-field region, k = x, y, z, where x, y, z are the coordinate axes, and p k c represents the current location of the drone. k This is the center of each axis length in the near-field region.

[0040] Optionally, when switching from near-field to far-field concentration detection, an improved Gaussian mixture model is used to preserve historical distribution characteristics, and a binary detection loop is entered. Specifically, this includes:

[0041] Obtain the average distribution density of the near-field region for the three most recent times;

[0042] The initial size of the near-field region for the next trigger is determined based on the average distribution density.

[0043] A spatial probability distribution model is determined using an improved Gaussian mixture model;

[0044] Binary entry detection is performed based on a spatial probability distribution model.

[0045] Optionally, determining the spatial probability distribution model using an improved Gaussian mixture model specifically includes:

[0046] Using formula Determine the confidence level P(x,y,z);

[0047] Where Z is the normalization coefficient, s is the spatial decay coefficient, and w iThe weighting is time-concentration composite, and it is updated with real-time concentration data each time it enters the near-field region. n is the total number of concentration measurement points participating in the weighted calculation, and p is the coordinate vector of the current position (x, y, z) in three-dimensional space. i Let be the spatial coordinate vector of the i-th measurement point.

[0048] Secondly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the wide-area gas tracing method based on multi-level sensor asynchronous collaborative triggering.

[0049] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the wide-area gas tracing method based on multi-level sensor asynchronous collaborative triggering.

[0050] Fourthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned wide-area gas tracing method based on multi-level sensor asynchronous collaborative triggering.

[0051] According to the specific embodiments provided in this application, this application has the following technical effects:

[0052] This application provides a wide-area gas source tracing method, device, medium, and product based on multi-level sensor asynchronous collaborative triggering. It employs a three-dimensional mesh generation method and an adaptive mesh mechanism to adaptively divide the area to be measured into a three-dimensional mesh, dynamically adjusting the mesh size in the absence of prior information, thus avoiding resource waste caused by fixed meshes. This application first detects plumes in the far-field region, and then determines the near-field region based on the number of measured data points according to the binary calculation results, i.e., formulating an event triggering strategy, significantly reducing the ineffective working time of high-precision sensors. Furthermore, this application's near-field-far-field dynamic conversion mechanism determines the near-field range in real time, reducing the area of ​​ineffective detection regions. Therefore, it can achieve efficient, accurate, and low-cost pollution source location in complex wide-area environments. Attached Figure Description

[0053] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a schematic diagram of a wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering in one embodiment of this application;

[0055] Figure 2 This is a schematic diagram illustrating the process of adjusting the size of the near-field region.

[0056] Figure 3 A schematic diagram of the process for determining exit from the near-field region;

[0057] Figure 4 A schematic diagram of the overall process for tracing and locating the source of gas. Detailed Implementation

[0058] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0059] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0060] In one exemplary embodiment, such as Figure 1 As shown, a wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering is provided, which includes the following steps S101 to S109. Wherein:

[0061] S101, the three-dimensional mesh generation method and adaptive mesh mechanism are used to adaptively generate a three-dimensional mesh for the area to be measured, and the mesh generation result is obtained;

[0062] Based on the three-dimensional dimensions (L) of the area to be measured x ,L y ,L z Determine the volume V = L of the region to be measured. x ×L y ×L z ;

[0063] Using formula Determine the cube mesh size d; where d∈[d min ,d max ];

[0064] Using the cubic mesh size, an adaptive 3D mesh is generated for the area to be measured, resulting in a d×d×d 3D cubic mesh.

[0065] Where V is the volume of the region to be measured, V threshold V is the volume threshold; when performing environmental monitoring tasks using small unmanned aerial vehicles (UAVs), it represents... threshold 300m3 ~500m 3 When tracing the source of pollution in a city / factory area, V threshold 1000m 3 ~3000m 3 V max The threshold value for the large spatial domain is 5000m. 3 Suitable for open areas or large industrial zones, V min This is a small spatial threshold, with a value of 200m. 3 Suitable for indoor or enclosed spaces, d max For the maximum grid size, d min α is the minimum grid size, α is the size attenuation coefficient, and k is the linear adjustment coefficient.

[0066] S102, Based on the grid division results and sampling data, the environmental background information is detected to obtain the mean value of the environmental background noise;

[0067] As a specific embodiment, a set of low-precision gas sensors acquires sampling data x = x1, x2, ..., x n ], where n is the number of sampled data (e.g., n = 100);

[0068] Using formula Determine the mean value μ of the ambient background noise; x i This represents the i-th sampled data; the mean value represents the baseline level of environmental noise.

[0069] S103 treats the area to be measured as the far field, adopts a random walk strategy and introduces a weight avoidance mechanism to detect plumes; and performs a 5-second hover sampling on the reached grid area;

[0070] Specifically, a one-dimensional random walk describes the process by which an object moves according to random rules in a discrete space:

[0071]

[0072] Where, ξ i Let S be independent and identically distributed random variables, with S0 as the starting point and S... n This is the new position after the move.

[0073] Furthermore, the mathematical expression of the random walk strategy extended to three-dimensional space (x,y,z) is:

[0074]

[0075] Among them, independent and identically distributed random variables in three-dimensional space Each component is an independent random variable. As the starting point of three-dimensional space, The position after the movement in three-dimensional space.

[0076] Before each move, the access status of adjacent nodes is checked. If the target node is not accessed, migration is allowed; if it is accessed or belongs to a shielded area, it is marked as non-migration and the sampling point is recalculated. Each sampling point is sampled for 5 seconds during hovering, and the average value represents the final environmental data for that point.

[0077] S104: Determine the threshold for binary detection based on the mean value of the ambient background noise; and use the threshold for binary detection to perform binary calculation on the sampled data; if the value exceeds the threshold for binary detection, set it to 1, i.e. gas is detected; if the value is below the threshold for binary detection, set it to 0, i.e. gas is not detected, and output the binary data.

[0078] The standard deviation σ is calculated using the sample standard deviation formula with Bessel correction:

[0079]

[0080] The denominator uses n-1 (Bessel correction) to avoid underestimating the population standard deviation;

[0081] The threshold T for binary detection is determined using the formula T = μ + 2σ;

[0082] Using formula Determine the output of the binary calculation;

[0083] Among them, s j Let j be the j-th data point sampled in real time.

[0084] S105, Based on the binary calculation results, determine the near-field region based on the number of points of the measured data points;

[0085] S105 specifically includes:

[0086] S51, Based on the binary calculation results, determine the set of coordinates P = {p1, p2, ..., p...} of the data points where the gas was detected. m}(m≥3);

[0087] S52, determine the spatial distribution characteristics of gas points based on the set of point coordinates; the spatial distribution characteristics of gas points include: spatial range and point set density;

[0088] The formula for calculating the spatial range (maximum span in each axis) is as follows:

[0089] R x =max(x i )-min(x i );

[0090] R y =max(y i)-min(y i );

[0091] R z =max(z) i )-min(z i );

[0092] The formula for calculating the density of a point set is:

[0093]

[0094] Where m is the number of gas points, ρ is the density of the point set, and convex_hull_volume(P) represents the convex hull volume of the point set P, which is the volume of the smallest convex polyhedron that encloses all gas points, reflecting the "aggregation compactness of the points, that is, the convex hull volume reflects the degree of aggregation of the point set".

[0095] S53, based on the spatial distribution characteristics of gas points, using the formula The adaptive near-field size D is determined; where 0.8 is an empirical coefficient value, calibrated through extensive experiments to balance the "basic size dominated by spatial range" and the "density compensation correction amount," ensuring that the near-field size adapts to the initial range of most scenarios. R x R y and R z Let ρ be the spatial range, ρ be the point set density, and β be the density compensation factor, with a value of 5; using the formula... Determine the constraints, D min =2×d grid (e.g., 20m), L is the axial length of the initial region, d grid This is the base resolution for the grid.

[0096] S54, based on the adaptive near-field size, uses the directional normalized range weighting method to adjust the gas distribution direction characteristics;

[0097] The direction-normalized range weighted method is as follows:

[0098]

[0099] Wherein, γ is the directional strengthening coefficient, with a value of 0.3.

[0100] S55 determines the near-field region based on the gas distribution direction characteristics.

[0101] The near-field region is defined as a three-dimensional region with O as its geometric center:

[0102]

[0103] S106 performs concentration tracking based on the near-field region (turns off the low-precision sensor and triggers the high-precision sensor to collect concentration information), uses an improved gradient calculation method and relative boundary distance to evaluate the near-field state in real time, adjusts the size of the near-field region based on the evaluation results, and continuously determines whether to exit the concentration detection in the near-field region.

[0104] Because a random walk strategy is used for sampling, the distance between two sampling points is not fixed and is unknown. The concentration gradient calculated based on the slope of the time difference may have a large error for distant sampling points. Therefore, it is necessary to eliminate the influence of distance and use a spatially distance-normalized gradient to replace the traditional time difference slope. The improved concentration gradient formula is as follows:

[0105]

[0106] Among them, C i C i+1 Let i be the concentration of data point i and data point i+1. Let i be the three-dimensional coordinates of the sampling point. Let i be the three-dimensional coordinates of the sampling point i+1, and ||·|| be the Euclidean distance.

[0107] Using formula Determine the boundary distance border dist (between the drone and the near-field boundary); border dist is [0,1], representing the relative distance of the current position from a certain boundary;

[0108] Based on the above near-field conditions, the two indicators (concentration slope) are calculated. i Based on the distance to the boundary (border_dist) and the number of 0-concentration points detected in the current near field (set as N0), the following near field region adjustment logic is executed. The near field adjustment process is as follows: Figure 2 As shown:

[0109] (1) EXPAND (Extend Near Field Region): When the concentration is decreasing and the drone is approaching the boundary, it may be that there is still gas outside the boundary of the near field region. In this case, it is recommended to expand the near field region.

[0110] If slope<-0.3and border_dist<0.2,then:state←EXPAND;

[0111] (2) SHRINK (Shrink Near Field Region): If the current point is far from the boundary (located in the central region), but multiple zero-concentration points are detected, it indicates that the near field region may have expanded beyond the set range and should be shrunk.

[0112] If N0≥3andborder_dist>0.5,then:state←SHRINK;

[0113] (3) EXIT (Exit Near Field) If no gas is detected at most points in the near field area (e.g., 5 consecutive detections are 0), it means there is no plume in the near field area, and you should exit the near field mode immediately:

[0114] If N0 ≥ 5, then: state ← EXIT;

[0115] (4) HOLD (Keep unchanged): Maintain the current near-field region state when the above three special conditions are not met.

[0116] Else:state←HOLD;

[0117] Based on the near-field region obtained after the previous dynamic adjustment, a comprehensive near-field exit judgment is performed to determine whether to truly exit the near-field mode and switch to far-field mode. Specifically, the near-field exit judgment process is as follows: Figure 3 As shown, the decision condition (should_exit_nearfield) has the following three levels:

[0118] (1) Forced Exit Conditions (Highest Priority)

[0119] Forced exit will occur if any of the following conditions are met:

[0120] 1. The concentration value is 0 for 5 consecutive times;

[0121] 2. Concentration decay rate >15% for three consecutive tests (negative growth);

[0122] Judgment criteria: This indicates that the gas source has disappeared or the drone has completely moved out of the plume range;

[0123] (2) Comprehensive risk assessment (main decision-making mechanism)

[0124] Overall risk score:

[0125] Overall risk score = 0.6 × boundary risk + 0.4 × gradient stability;

[0126] Exit condition: Exit score > 0.7;

[0127] Specifically, the evaluation metrics mainly include boundary risk and gradient stability:

[0128] Calculate the boundary risk value (0-1):

[0129] BoundaryRisk = min(risk) x risk y riskz );

[0130] Specifically, the formula for calculating boundary risk for each coordinate axis (X, Y, Z) in three-dimensional space is as follows:

[0131] p k This is the current location of the drone. These are the near-field boundary coordinates; O k The coordinates of the geometric center of the near-field region are... The ratio of the distance to the left boundary. This represents the ratio of the distance to the right boundary.

[0132] The final boundary risk is the minimum value of the boundary risk values ​​of the three axes; the overall risk is determined by comprehensively assessing the proximity of the UAV to the boundary in all dimensions and adopting the minimum value principle, with the most dangerous direction; decision application: (1) 0.4 safe zone (it is recommended to maintain or expand the near field); (2) 0.2-0.4 warning zone (it is recommended to maintain the current state); (3) <0.2 danger zone (it is recommended to shrink or withdraw from the near field); the higher the value, the closer to the boundary;

[0133] Calculate gradient stability (0-1) as follows:

[0134]

[0135] Where, N 梯度 N represents the number of gradients. 方向变化 For the number of changes, For average amplitude, For the gradient magnitude variance,

[0136] (3) Boundary Hesitation Judgment (Auxiliary Conditions)

[0137] 1. Check the last 5 location records;

[0138] 2. The number of times the boundary risk value > 0.6 (within 30% of the boundary area) is ≥ 3;

[0139] The boundary loitering analysis indicates that the drone repeatedly and ineffectively probed the boundary area.

[0140] S107, when switching from concentration detection in the near field region to the far field region, an improved Gaussian mixture model is used to preserve historical distribution characteristics and enter the binary detection loop;

[0141] S107 specifically includes:

[0142] S71, Obtain the average distribution density of the near-field region in the most recent three tests.

[0143] S72, determine the initial size D of the next near-field region trigger based on the average distribution density. init ;

[0144] S73, using an improved Gaussian mixture model to determine the spatial probability distribution model;

[0145] Using formula Determine the confidence level P(x,y,z);

[0146] Where Z is the normalization coefficient (which needs to be recalculated after each update). s is the spatial attenuation coefficient, w i The weighting is time-concentration composite, and it is updated with real-time concentration data each time it enters the near-field region. n is the total number of concentration measurement points participating in the weighted calculation, and p is the coordinate vector of the current position (x, y, z) in three-dimensional space. i Let be the spatial coordinate vector of the i-th measurement point. s = s0 × (1 + α0 · distribution dispersion). And each time it enters the near-field region, it updates the data using real-time concentration data.

[0147] S74 performs binary entry detection based on a spatial probability distribution model.

[0148] S108 determines whether to re-enter the near-field region based on real-time binary detection data;

[0149] Specifically, the verification conditions include:

[0150] The specific calculation process is as follows:

[0151] 1. For each detected gas point (x) i ,y i ,z i ), calculate its confidence level P(x) in the historical probability model. i ,y i ,z i );

[0152] 2. If there exists at least one point satisfying P(x) i ,y i ,z i If the value is greater than 0.7, then the spatial verification is passed;

[0153] The real-time update rule for the probabilistic model P(x,y,z) is as follows: each time the near field is entered, the weights are updated with the newly acquired concentration data.

[0154]

[0155] Based on the above conditions, a final decision will be made as to whether to re-enter near-field mode.

[0156] S109, determine the concentration peak based on the concentration detection data of all near-field areas; and take the concentration peak as the source location.

[0157] like Figure 4 As shown, based on the concentration data information of the near-field region obtained in S106 to S108 above, the concentration peak is locked until the sensor array detects a clear and significant concentration gradient in the near-field region. The UAV tracks this gradient and locates a relatively stable concentration peak point or a very small high-concentration core area. When the position of the highest concentration no longer drifts significantly, this point / area is considered to be the place where the gas leakage or release is most intense, i.e., the source location.

[0158] This application has the following effects:

[0159] (1) Improved efficiency (reduced search time and resource consumption)

[0160] By employing a multi-level asynchronous collaborative triggering mechanism (the high-precision sensor is activated only after the low-precision sensor detects the plume), the ineffective working time of the high-precision sensor is significantly reduced. Furthermore, it is shown that the high-precision sensor is activated only for 10%–20% of the total working time, significantly reducing system energy consumption (measured power consumption reduction of 35%) and extending the UAV's endurance. Through adaptive mesh generation and random walk strategies, in environments without prior information, the resolution is dynamically adjusted using a volume-driven mesh size formula, avoiding resource waste caused by fixed meshes. Compared to traditional uniform meshes, search efficiency is improved by 40% (measured data). Through a near-field-far-field dynamic conversion mechanism, the near-field range is determined in real time based on boundary risk values ​​(BoundaryRisk = min(risk_x, risk_y, risk_z)) and gradient stability, reducing the area of ​​ineffective detection regions. This mechanism shows that it reduces the near-field detection area by 30% while increasing the positioning speed by 50%.

[0161] (2) Improved positioning accuracy (enhanced anti-interference and stability)

[0162] By introducing a spatial probability distribution model (an improved Gaussian mixture model), a time-concentration composite weight is introduced. The mechanism for fusing historical data effectively suppresses misjudgments caused by transient interference. Experiments show that in scenarios with sudden changes in wind speed, the positioning error is reduced by 60% compared to the traditional gradient method (average error <1.5m). Through adaptive near-field adjustment, the near-field size is dynamically corrected based on the gas distribution direction characteristics. Furthermore, the near-field region is matched with the plume morphology. In actual measurements, the accuracy in the Z-axis direction is improved by 25% (especially suitable for complex terrain). Through a triple near-field exit verification mechanism—forced exit (continuous zero-value detection), boundary risk assessment (multi-dimensional risk calculation), and gradient stability analysis—early exit / delay is avoided. The false exit rate is reduced to below 5% (compared to >20% for traditional methods).

[0163] (3) Improved environmental adaptability (robustness in complex scenarios)

[0164] By employing a dynamic background noise threshold, an automatic threshold calculation based on the mean plus two standard deviations (T = μ + 2σ) is used to adapt to different environmental backgrounds (such as high background noise in industrial areas). During testing, the plume recognition accuracy remained >92% within a background noise range of 0–100 ppm. Historical distribution density was determined through near-field initialization driven by historical data. Then calculate the initial near-field size ( This accelerates the repeated search process. Compared to the first round of search, the second location time is reduced by 70%.

[0165] (4) Optimize resource utilization (cost-benefit balance)

[0166] By employing a high- and low-precision sensor linkage strategy, the high-precision sensor is activated only in the near-field region (duty cycle <20%), extending the lifespan of expensive sensors (estimated lifespan increased by 3 times) and reducing the cost per mission by 45%. Deduplication avoidance mechanisms and spatial clustering verification enable the shielding of visited areas during random walks (ξ). i Historical locations are excluded during generation, and a new spatial clustering condition (max) is added. i P(x i ,y i ,z i (>0.7), reducing redundant sampling points. The number of measured sampling points was reduced by 35%.

[0167] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal. The computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The I / O interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering.

[0168] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0169] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0170] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0171] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0172] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0173] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0174] In this application, all actions to acquire signals, information, or data are carried out in compliance with the relevant data protection laws and policies of the country where the location is situated, and with the authorization granted by the owner of the relevant device.

[0175] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0176] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering, characterized in that, The wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering includes: A three-dimensional mesh generation method and an adaptive mesh mechanism are used to adaptively generate a three-dimensional mesh for the area to be measured, and the mesh generation result is obtained. Based on the grid division results and sampling data, the environmental background information is detected to obtain the mean value of the environmental background noise; The area to be measured is considered as the far field, and a random walk strategy and a weight avoidance mechanism are introduced to detect plumes. The threshold for binary detection is determined based on the mean value of the background noise; and the sampling data is then subjected to binary calculation using the threshold for binary detection. Based on the binary calculation results, the near-field region is determined according to the number of points of the measured data points; Concentration tracking is performed in the near-field region, and an improved gradient calculation method and relative boundary distance are used to evaluate the near-field status in real time; the size of the near-field region is adjusted according to the evaluation results; and concentration detection is continuously judged to determine whether to exit the near-field region. When switching from concentration detection in the near field region to the far field region, an improved Gaussian mixture model is used to preserve historical distribution characteristics, and then the binary detection loop is entered. Determine whether to re-enter the near-field region based on real-time binary detection data; Based on the concentration detection data of all near-field areas, the concentration peak is determined; and the concentration peak is used as the source location.

2. The wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering according to claim 1, characterized in that, The method employs a three-dimensional mesh generation method and an adaptive mesh mechanism to adaptively generate a three-dimensional mesh for the area to be measured, resulting in mesh generation results, specifically including: Determine the volume of the region to be measured based on its three-dimensional dimensions; Using formula Determine the cube mesh size d; Using the cubic mesh size, an adaptive 3D mesh is generated for the area to be measured, and the mesh generation result is obtained. Where V is the volume of the region to be measured, V threshold V is the volume threshold. max For the large spatial domain threshold, V min For the small spatial domain threshold, d max For the maximum grid size, d min α is the minimum grid size, α is the size attenuation coefficient, and k is the linear adjustment coefficient.

3. The wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering according to claim 1, characterized in that, The step of determining the threshold for binary detection based on the mean of the ambient background noise, and then using the threshold for binary detection to perform binary calculations on the sampled data, specifically includes: The threshold T for binary detection is determined using the formula T = μ + 2σ; Using formula Determine the output of the binary calculation; Where μ is the mean value of the ambient background noise. x i Let be the i-th sampled data, n be the number of sampled data, and σ be the standard deviation. s j Let j be the j-th data point sampled in real time.

4. The wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering according to claim 1, characterized in that, Based on the binary calculation results, the near-field region is determined according to the number of measured data points, specifically including: Based on the binary calculation results, determine the set of coordinates of the data points where the gas was detected; The spatial distribution characteristics of gas points are determined based on the set of point coordinates; the spatial distribution characteristics of gas points include: spatial range and point set density; Based on the spatial distribution characteristics of gas points, using the formula Determine the adaptive near-field size D; where R x R y and R z Let ρ be the spatial range, ρ be the point set density, and β be the density compensation factor. Based on the adaptive near-field size, the directional normalized range weighting method is used to adjust the gas distribution directional characteristics. The near-field region is determined based on the gas distribution direction characteristics.

5. The wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering according to claim 1, characterized in that, The concentration tracking is performed based on the near-field region, and an improved gradient calculation method and relative boundary distance are used to evaluate the near-field state in real time. And adjust the near-field area size based on the evaluation results; specifically including: Using formula Determine the concentration slope i ; Using formula Determine the border distance (border distance); The size of the near-field region is adjusted based on the concentration slope, boundary distance, and the number of data points with 0 concentration detected in the current near-field region. Among them, C i C i+1 Let i be the concentration of data point i and data point i+1. Let i be the three-dimensional coordinates of the sampling point. Let D be the three-dimensional coordinates of sampling point i+1, ||·|| be the Euclidean distance, and D be the distance between the sampling points. k Let k be the length of each axis in the near-field region, k = x, y, z, where x, y, z are the coordinate axes, and p k c represents the current location of the drone. k This is the center of each axis length in the near-field region.

6. The wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering according to claim 1, characterized in that, When switching from near-field to far-field concentration detection, an improved Gaussian mixture model is used to preserve historical distribution characteristics, and a binary detection loop is entered. Specifically, this includes: Obtain the average distribution density of the near-field region for the three most recent times; The initial size of the near-field region for the next trigger is determined based on the average distribution density. A spatial probability distribution model is determined using an improved Gaussian mixture model; Binary entry detection is performed based on a spatial probability distribution model.

7. The wide-area gas source tracing method based on multi-level sensor asynchronous collaborative triggering according to claim 6, characterized in that, The determination of the spatial probability distribution model using the improved Gaussian mixture model specifically includes: Using formula Determine the confidence level P(x,y,z); Where Z is the normalization coefficient, s is the spatial decay coefficient, and w i The weighting is time-concentration composite, and it is updated with real-time concentration data each time it enters the near-field region. n is the total number of concentration measurement points participating in the weighted calculation, and p is the coordinate vector of the current position (x, y, z) in three-dimensional space. i Let be the spatial coordinate vector of the i-th measurement point.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the wide-area gas tracing method based on multi-level sensing asynchronous collaborative triggering as described in any one of claims 1-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the wide-area gas tracing method based on multi-level sensor asynchronous collaborative triggering as described in any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the wide-area gas tracing method based on multi-level sensor asynchronous collaborative triggering as described in any one of claims 1-7.

Images