Dangerous article detection method based on unmanned aerial vehicle

Abstract

The invention relates to the technical field of unmanned aerial vehicle dangerous goods detection methods integrating environment perception, adaptive flight control, data intelligent correction and boundary decision, in particular to a dangerous goods detection method based on an unmanned aerial vehicle. Comprising the following steps: S1, acquiring an environment wind speed and a dangerous article concentration of a current position of an unmanned aerial vehicle, calculating a dangerous substance diffusion gradient, and further constructing a coupling disturbance factor in combination with the environment wind speed and the dangerous substance diffusion gradient; s2, in response to the coupling disturbance factor, calculating an optimal scanning speed, and determining an adaptive sampling period; s3, calculating a credibility index for data points in the original concentration data sequence according to the self-adaptive sampling period, performing weighted correction on the original concentration data by using the credibility index, and outputting a corrected concentration sequence; and S4, determining the boundary of the dangerous area by using the corrected concentration sequence. According to the method, the real-time synchronization of the detection behavior and the environment change is realized, and the detection efficiency and precision are effectively considered.

Description

Technical Field

[0001] This invention relates to the technical field of a method for detecting hazardous materials using unmanned aerial vehicles (UAVs), which integrates environmental perception, adaptive flight control, intelligent data correction, and boundary decision-making. Specifically, it is a method for detecting hazardous materials using unmanned aerial vehicles (UAVs). Background Technology

[0002] Traditional methods face significant challenges when detecting hazardous materials in complex and dynamic environments. These environments are characterized by constantly changing wind speeds and uneven dispersion of hazardous substances, which severely restricts the efficiency and accuracy of detection efforts.

[0003] Currently, traditional detection methods often struggle to balance detection speed and accuracy when dealing with such complex situations. The data they collect is easily affected by environmental factors, and the misjudgment rate is high when determining the boundaries of dangerous areas. These methods typically lack a closed-loop system that can integrate environmental perception, adaptive adjustment, data correction, and boundary decision-making, resulting in poor performance in dynamic real-world scenarios.

[0004] Therefore, how to overcome the interference of factors such as wind speed changes and uneven diffusion in complex dynamic environments, and achieve efficient, accurate and reliable determination of dangerous area boundaries, has become a technical problem that urgently needs to be solved in this field.

[0005] The information disclosed in the background section above is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention discloses a method for detecting hazardous materials based on unmanned aerial vehicles (UAVs). Specifically, the technical solution of this invention is as follows:

[0007] A method for detecting hazardous materials based on unmanned aerial vehicles (UAVs) includes:

[0008] S1. Obtain the ambient wind speed and hazardous material concentration at the current location of the drone, and calculate the hazardous material diffusion gradient based on the concentration and drone positioning information. Then, combine the ambient wind speed and hazardous material diffusion gradient to construct a coupling perturbation factor.

[0009] S2. In response to the coupling disturbance factor, calculate the optimal scanning speed and determine the adaptive sampling period based on the optimal scanning speed;

[0010] S3. Based on the adaptive sampling period, collect the original concentration data sequence, calculate the confidence index for the data points in the original concentration data sequence, and then use the confidence index to perform weighted correction on the original concentration data, and output the corrected concentration sequence.

[0011] S4. Using the corrected concentration sequence and based on the preset safe concentration threshold and stable gradient threshold, the spatial grid points are judged by dual criteria to determine the boundary of the danger zone.

[0012] Preferably, in S1, the calculation of the hazardous substance diffusion gradient includes:

[0013] Collect concentration measurements and coordinate vectors from two spatial locations using a drone;

[0014] The diffusion gradient of hazardous substances is calculated based on the concentration measurements and Euclidean distance between two spatial locations, combined with a preset reference length scale.

[0015] Preferably, in S1, the construction of the coupling perturbation factor includes:

[0016] The wind speed ratio is obtained by dividing the real-time ambient wind speed by the preset reference wind speed threshold.

[0017] The gradient ratio is obtained by dividing the real-time calculated hazardous substance diffusion gradient by a preset reference gradient threshold.

[0018] The coupling disturbance factor is obtained by multiplying the wind speed ratio and the gradient ratio, and then summing the sum with the basic terms.

[0019] Preferably, in S2, the calculation of the optimal scan speed includes:

[0020] An exponential decay model is adopted, taking the coupling disturbance factor as input, based on the preset maximum safe operating speed, and responding to the preset speed adjustment sensitivity coefficient to perform exponential decay calculation to obtain the optimal scanning speed.

[0021] Preferably, in S2, the determination of the adaptive sampling period includes:

[0022] Divide the preset sampling compensation coefficient by the optimal scanning speed to obtain the adjustment term;

[0023] The adaptive sampling period is obtained by adding the adjustment term to the basic minimum sampling period determined by the sensor hardware.

[0024] Preferably, in S3, the calculation of the credibility index includes:

[0025] Obtain the concentration readings of the current and previous moments, calculate the absolute value of the difference, and divide it by the preset sensor response limit threshold to obtain the concentration change rate.

[0026] A credibility index is generated by nonlinearly mapping the concentration change rate using the hyperbolic tangent function.

[0027] Preferably, in S3, the output of the corrected concentration sequence includes:

[0028] An exponential moving average filter is used, with the credibility index of the current data point as the dynamic weight.

[0029] The original concentration readings are weighted and averaged with the previously corrected concentration values, and the corrected concentration sequence is calculated recursively.

[0030] Preferably, in S4, the dual-criteria judgment includes:

[0031] When the average corrected concentration value of the spatial grid point is greater than or equal to the preset safe concentration threshold, and the spatial gradient of the corrected concentration at the spatial grid point is less than or equal to the preset stable gradient threshold, the spatial grid point is identified as a dangerous grid point.

[0032] If any condition is not met, the spatial grid point is determined to be a safe grid point.

[0033] Preferably, the determination of the boundary of the danger zone also includes:

[0034] The set of all grid points identified as hazardous is aggregated to form the boundary of the hazardous area.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] 1. This method constructs a coupling perturbation factor to quantify environmental wind speed and hazardous substance diffusion gradient into a unified index, and dynamically calculates the optimal scanning speed and adaptive sampling period based on this, realizing real-time synchronization between detection behavior and environmental changes, effectively balancing detection efficiency and accuracy.

[0037] 2. This method uses a confidence index to dynamically weight and correct the original concentration data, which can effectively suppress noise and distortion of sensor data under harsh environments, filter out abnormal jumps, and output a concentration sequence that better reflects the true concentration trend, thereby improving the reliability of the data.

[0038] 3. This method uses a dual criterion of safe concentration threshold and stable gradient threshold for boundary determination, which can effectively eliminate false boundary points with instantaneous high concentration caused by turbulence or noise, and ensure that the concentration distribution in the area determined as the boundary is relatively stable, which greatly reduces the misjudgment rate of traditional single threshold methods.

[0039] 4. This method constructs a complete technical closed loop from environmental quantification, behavior optimization, data correction to boundary reconstruction, transforming complex environmental disturbance problems into clear quantitative indicators and optimizing them step by step, which significantly improves the automation level and accuracy of boundary determination in the detection of hazardous materials in complex dynamic environments. Attached Figure Description

[0040] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0041] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0043] Example 1:

[0044] A method for detecting hazardous materials based on unmanned aerial vehicles (UAVs), comprising the following steps:

[0045] S1. Obtain the ambient wind speed and hazardous material concentration at the current location of the drone, and calculate the hazardous material diffusion gradient based on the concentration and drone positioning information. Then, combine the ambient wind speed and hazardous material diffusion gradient to construct a coupling perturbation factor.

[0046] S2. In response to the coupling disturbance factor, calculate the optimal scanning speed and determine the adaptive sampling period based on the optimal scanning speed;

[0047] S3. Based on the adaptive sampling period, collect the original concentration data sequence, calculate the confidence index for the data points in the original concentration data sequence, and then use the confidence index to perform weighted correction on the original concentration data, and output the corrected concentration sequence.

[0048] S4. Using the corrected concentration sequence and based on the preset safe concentration threshold and stable gradient threshold, the spatial grid points are judged by dual criteria to determine the boundary of the danger zone.

[0049] This embodiment provides a method for detecting hazardous materials based on unmanned aerial vehicles (UAVs), aiming to solve the problems of traditional detection methods in complex dynamic environments, such as wind speed changes and uneven diffusion of hazardous materials, which make it difficult to balance detection speed and accuracy, data is easily interfered with, and the boundary misjudgment rate is high. The method constructs a complete closed-loop system from environmental perception, flight control and sampling adaptive adjustment, data quality correction to boundary decision-making, so as to achieve efficient, accurate and reliable determination of hazardous area boundaries.

[0050] To ensure the quality of the detection data, it is recommended to install the sensor in an area away from the rotor downwash, such as above the fuselage or on the front boom, to reduce aerodynamic interference from the UAV itself. At the same time, the method has built-in safety and fault handling mechanisms: when sensor data is lost or abnormal, the system will switch to a preset safety mode, such as flying with fixed conservative parameters or hovering alarm; when the flight speed calculated by the algorithm is lower than the preset minimum speed threshold, it will force flight at the minimum speed to prevent mission stall, thereby ensuring the robustness of the entire process.

[0051] A method for detecting hazardous materials based on unmanned aerial vehicles (UAVs) aims to accurately and reliably delineate the boundaries of hazardous areas through a series of dynamic and adaptive steps. This method primarily addresses the adaptive detection problem along predetermined detection paths, such as gridded scanning or boundary-flying. The specific path planning algorithm can employ existing technologies in this field. In this embodiment, the method's specific steps include:

[0052] S1. Obtain the ambient wind speed and hazardous material concentration at the current location of the UAV. Based on the concentration and UAV positioning information, calculate the hazardous material diffusion gradient. Then, combine the ambient wind speed and the hazardous material diffusion gradient to construct a coupled perturbation factor. Ambient wind speed refers to the real-time wind speed at the location of the UAV. Its function is to quantify the direct impact of atmospheric flow on the diffusion of hazardous materials. It is obtained in real time through an airborne wind speed sensor. Hazardous material concentration refers to the content of a specific hazardous material in the air. Its function is to provide a basis for judging the degree of danger. It is obtained in real time through an airborne hazardous material detector. Hazardous material diffusion gradient is a dimensionless relative rate of change that characterizes the steepness and irregularity of the spatial distribution of concentration. Its purpose is to quantify the non-uniformity of hazardous material diffusion. Coupled perturbation factor is a single indicator that uniformly measures the comprehensive negative impact of changes in ambient wind speed and concentration gradient on the detection task. Its purpose is to simplify the complex multivariate environmental problem into a response problem to a single perturbation indicator, providing a decision-making basis for subsequent dynamic optimization.

[0053] S2. In response to the coupling disturbance factor, the optimal scanning speed is calculated, and the adaptive sampling period is determined in combination with the optimal scanning speed. The optimal scanning speed refers to the optimal scanning speed that the UAV should adopt under the current environmental disturbance to balance detection efficiency and data quality. Its purpose is to dynamically adjust the detection rhythm according to the severity of the environment. The more complex the environment, the slower the speed and the more detailed the scan. The adaptive sampling period refers to the time interval for the sensor to collect data. Its purpose is to ensure that the UAV can maintain a constant spatial sampling density at different flight speeds and prevent the loss of key information due to passing too fast during high-speed flight.

[0054] S3. Based on the adaptive sampling period, the original concentration data sequence is collected, and a confidence index is calculated for each data point in the original concentration data sequence. The confidence index is then used to perform weighted correction on the original concentration data, and the corrected concentration sequence is output. The confidence index is a dimensionless index between 0 and 1, used to evaluate the reliability of each collected concentration data point. Its purpose is to quantify the unreliability of data mutations caused by severe environmental disturbances or the sensor reaching its response limit. The corrected concentration sequence is a data sequence after confidence-weighted smoothing, which aims to filter out noise and abnormal jumps in the original data and output a smoother, more reliable data stream that better reflects the true concentration trend.

[0055] S4. Using the corrected concentration sequence and based on preset safe concentration thresholds and stable gradient thresholds, spatial grid points are judged using a dual criterion to determine the boundary of the hazardous area. The safe concentration threshold is a concentration limit preset based on relevant safety regulations or toxicological data, and is the basic legal or industry standard for judging whether a certain area is dangerous. The stable gradient threshold is an empirical parameter used to distinguish between stable diffusion boundaries and anomalous data jumps. Its purpose is to add a judgment dimension to ensure that the area judged as the boundary not only meets the concentration standard, but also has a relatively stable concentration distribution, thereby eliminating false boundaries of instantaneous high concentration caused by factors such as turbulence. Through this dual criterion judgment, that is, considering both the absolute value of concentration and the spatial rate of change of concentration, spatial grid points are classified, and finally all grid points judged as dangerous are summarized to form the boundary of the hazardous area.

[0056] This invention, by constructing a coupled perturbation factor, quantifies wind speed and concentration gradient—two core environmental disturbance sources—into a unified index for the first time. Based on this, it achieves serial dynamic optimization of flight speed and sampling period, ensuring real-time synchronization between detection behavior and environmental changes. Furthermore, by introducing a credibility index to dynamically weight and correct the raw data, it effectively suppresses noise and distortion in sensor data under harsh environments. Finally, by employing both concentration and gradient criteria for boundary determination, it significantly reduces the misjudgment rate caused by data jumps in traditional single-threshold methods. This method forms a complete technical closed loop of environmental quantification, behavior optimization, data correction, and boundary reconstruction, significantly improving the automation level, data reliability, and accuracy of boundary determination in the detection of hazardous materials in complex dynamic environments.

[0057] Example 2:

[0058] In S1, the calculation of the hazardous substance diffusion gradient includes:

[0059] Collect concentration measurements and coordinate vectors from two spatial locations using a drone;

[0060] The diffusion gradient of hazardous substances is calculated based on the concentration measurements and Euclidean distance between two spatial locations, combined with a preset reference length scale.

[0061] Based on Example 1, this embodiment specifies the calculation method for the diffusion gradient of hazardous substances in S1. Its purpose is to provide a standardized, dimensionless gradient calculation method that can objectively reflect the drastic degree of relative concentration change, without being affected by the magnitude of the absolute concentration value.

[0062] The calculation includes the following steps:

[0063] The drone collects concentration measurements and coordinate vectors at two spatial locations; specifically, the drone continuously acquires two spatial locations within a short period of time, with their coordinate vectors being... , and their corresponding concentration measurements , ;

[0064] Based on the concentration measurements and Euclidean distance between two spatial points, and in conjunction with a preset reference length scale, the diffusion gradient of hazardous substances is calculated; in this embodiment, the calculation is implemented through a specific mathematical model.

[0065] To ensure diffusion gradient To reflect the dimensionless characteristics of the property and the degree of its relative change, a dimensionless diffusion gradient is introduced. The calculation formula is as follows:

[0066]

[0067] in, The dimensionless diffusion gradient is the output of this step.

[0068] The concentration values ​​at the two sampling points [in ppm] were obtained using an airborne hazardous materials detector.

[0069] The Euclidean distance between two sampling points [in meters] is calculated using UAV positioning information;

[0070] The reference length scale [in meters] is a preset parameter that is used to evaluate the size of the characteristic space of the concentration gradient. It is set according to the task requirements, such as twice the size of the UAV body or the minimum spatial resolution of 5 meters required by the task.

[0071] Minimal concentration positive number [unit: ppm], is an engineering constant set to prevent the denominator of the concentration from being zero, for example, 1e-6;

[0072] Minimal positive distance [unit: m], is an engineering constant set to prevent the denominator of distance from being zero, for example 1e-6;

[0073] Through the above-described specific calculation method, this invention can obtain a standardized, dimensionless hazardous substance diffusion gradient. This gradient value not only has a clear physical meaning, characterizing the relative rate of change of concentration at a specific spatial scale, but also has good scale invariance. This makes subsequent decisions based on this gradient, such as the construction of coupling perturbation factors, more robust and reliable, effectively improving the adaptability of the entire detection system to hazardous sources with different concentration levels and different diffusion modes. It should be noted that this gradient calculation method assumes that the concentration field is quasi-static within a short period between two samplings. In scenarios with extremely high accuracy requirements, more complex spatiotemporal models can be introduced for compensation.

[0074] Example 3:

[0075] In S1, the construction of the coupling perturbation factor includes:

[0076] The wind speed ratio is obtained by dividing the real-time ambient wind speed by the preset reference wind speed threshold.

[0077] The gradient ratio is obtained by dividing the real-time calculated hazardous substance diffusion gradient by a preset reference gradient threshold.

[0078] Multiply the wind speed ratio by the gradient ratio and sum the result with the basic term to obtain the coupling disturbance factor;

[0079] Based on Example 1, this embodiment specifies the construction method of the coupled disturbance factor in S1. Its purpose is to integrate the two multi-dimensional disturbance information of real-time changing environmental wind speed and hazardous substance diffusion gradient into a single, standardized quantitative index, thereby simplifying the subsequent decision-making model.

[0080] The construction process includes:

[0081] The wind speed ratio is obtained by dividing the real-time ambient wind speed by the preset reference wind speed threshold.

[0082] The gradient ratio is obtained by dividing the real-time calculated hazardous substance diffusion gradient by a preset reference gradient threshold.

[0083] Multiply the wind speed ratio by the gradient ratio and sum the result with the basic term to obtain the coupling disturbance factor;

[0084] In this embodiment, a coupling perturbation factor is introduced to uniformly measure the comprehensive negative impact of changes in ambient wind speed and concentration gradient on the detection mission. The calculation formula is as follows:

[0085]

[0086] in, The dimensionless coupling perturbation factor is the output of this step.

[0087] Real-time ambient wind speed [unit: m / s], collected in real time by an airborne wind speed sensor;

[0088] The reference wind speed threshold [in m / s] is a key adjustable parameter that represents the critical wind speed at which the drone platform can maintain stable flight. It is determined based on the official wind resistance specifications of the drone hardware platform, such as 8 m / s, or through precise calibration via actual wind tunnel flight tests.

[0089] The dimensionless hazardous substance diffusion gradient is calculated in real time. [Dimensionless] is obtained by the steps in Example 2.

[0090] The reference gradient threshold, [dimensionless], is a key adjustable parameter defined as the critical gradient level at which sensor readings begin to show unstable phenomena such as frequent saturation or failure. It is determined based on statistical data analysis of historical hazardous material leakage events or through fluid dynamics simulation.

[0091] This embodiment successfully transforms the complex bivariate problem—wind speed, concentration gradient, and environmental disturbance—into a problem involving a single index by constructing this coupled disturbance factor. The response issue; The magnitude of the value intuitively quantifies the overall severity of the current environment for the detection mission, providing a clear and quantitative basis for dynamic optimization of subsequent flight and sampling parameters, greatly simplifying the design complexity of the control system and improving the robustness of the decision.

[0092] Example 4:

[0093] In S2, the calculation of the optimal scan speed includes:

[0094] An exponential decay model is adopted, taking the coupling disturbance factor as input, based on the preset maximum safe operating speed, and responding to the preset speed adjustment sensitivity coefficient to perform exponential decay calculation to obtain the optimal scanning speed;

[0095] Based on Example 1, this embodiment specifies the calculation method for the optimal scanning speed in S2. Its purpose is to establish a direct mapping relationship from environmental disturbances to flight behavior, so that the UAV can automatically and quickly adjust its detection speed according to the severity of the environment.

[0096] The calculation process is as follows: using the exponential decay model, taking the coupling disturbance factor as input, based on the preset maximum safe operating speed, and responding to the preset speed adjustment sensitivity coefficient, the exponential decay calculation is performed to obtain the optimal scanning speed;

[0097] In this embodiment, the speed decision is achieved through a custom speed control law based on an exponential decay model; an optimal scan speed is introduced. The calculation formula is as follows:

[0098]

[0099] in, Optimal scan speed [in m / s] is the output of this step.

[0100] Maximum safe operating speed (in m / s) refers to the speed at which the drone operates in an ideal, undisturbed environment. The maximum flight speed under the given conditions is preset based on the performance of the UAV and the safety requirements of the mission;

[0101] The velocity adjustment sensitivity coefficient, [dimensionless], is a custom core adjustable parameter that controls the degree of velocity decay as a function of perturbations; it is determined through offline calibration experiments. Specifically, a calibration dataset containing multiple stable perturbation environments is constructed, where each environment has a known, constant coupling perturbation factor. In each environment, different flight speeds were tested. Find a speed that optimizes a preset detection performance evaluation function, such as a cost function composed of false alarm rate and false alarm rate. This yields a set of data points. Finally, through regression analysis methods such as nonlinear least squares, the formula was analyzed. By performing a fitting process, the optimal solution can be obtained. value;

[0102] : Coupling perturbation factor, [dimensionless], is calculated by the steps in Example 3;

[0103] By employing this exponentially decaying solution method, this invention establishes a highly efficient negative feedback regulation mechanism; when environmental disturbances occur... As the speed increases, the scanning speed The speed automatically and non-linearly decreases, increasing the drone's dwell time and perception time in harsh areas; conversely, when the environment is favorable, the speed increases to accelerate the detection process. This adaptive speed adjustment capability enables the drone to maximize global detection efficiency while ensuring detection accuracy, solving the drawbacks of the traditional fixed-speed scanning mode.

[0104] Example 5:

[0105] In S2, the determination of the adaptive sampling period includes:

[0106] Divide the preset sampling compensation coefficient by the optimal scanning speed to obtain the adjustment term;

[0107] The adaptive sampling period is obtained by adding the adjustment term to the basic minimum sampling period determined by the sensor hardware.

[0108] Based on Example 1, this embodiment specifies the method for determining the adaptive sampling period in S2; its purpose is to cooperate with the decision of the optimal scanning speed in the previous stage to ensure that no matter how fast or slow the UAV flies, the sampling point density on its spatial path remains at a constant and ideal level.

[0109] The determination process includes:

[0110] Divide the preset sampling compensation coefficient by the optimal scanning speed to obtain the adjustment term;

[0111] The adaptive sampling period is obtained by adding the adjustment term to the basic minimum sampling period determined by the sensor hardware.

[0112] In this embodiment, the period is tuned based on the idea of ​​ensuring a constant spatial sampling density, and is achieved through a linear adjustment model; an adaptive sampling period is introduced. The calculation formula is as follows:

[0113]

[0114] The denominator in this formula The dimensions are meters per second, molecules The dimension of is meters per second, therefore the dimension of the fractional term is seconds per second, and... and The units (seconds / times) should remain consistent.

[0115] in, The adaptive sampling period [in seconds per sampling] is the calculation output of this step.

[0116] The minimum sampling period (in seconds) is the minimum sampling period that ensures the sampling period will not fall below the hardware limit. Its origin is determined by the technical specifications of the sensor hardware and is an inherent property of the sensor.

[0117] The sampling compensation coefficient [unit: meters / scan] is a key adjustable parameter. It represents the desired number of meters the UAV flies to complete at least one effective sample, indicating the specific spatial resolution requirements of the mission. If high-density scanning is required, then... Take the smaller value; if it is a coarse scan, then Take the larger value;

[0118] The optimal scanning speed [in m / s] is calculated using the steps in Example 4.

[0119] This embodiment dynamically links the sampling period with the optimal scanning speed, ensuring that the UAV maintains a constant spatial sampling resolution throughout the entire detection mission, regardless of changes in flight speed. This effectively avoids information loss due to sparse sampling points in high-speed flight areas or resource waste due to overly dense sampling points in low-speed flight areas, achieving optimized allocation of detection resources and providing a spatially uniform and constant-quality data foundation for subsequent data processing and boundary analysis.

[0120] To improve system robustness, this step also imposes boundary constraints on the output values; when the optimal scan speed is calculated... Below the preset minimum safe speed At that time, the actual execution speed will be forcibly set to [the specified value]. Accordingly, the calculated sampling period It will also be limited to a preset maximum value. To avoid sampling interruptions due to excessively low flight speed and ensure the continuity of the exploration mission.

[0121] Example 6:

[0122] In S3, the calculation of the credibility index includes:

[0123] Obtain the concentration readings of the current and previous moments, calculate the absolute value of the difference, and divide it by the preset sensor response limit threshold to obtain the concentration change rate.

[0124] A credibility index is generated by nonlinearly mapping the concentration change rate using the hyperbolic tangent function.

[0125] Based on Example 1, this embodiment specifies the calculation method of the credibility index in S3; its purpose is to assign a quantitative and standardized reliability score to each collected raw concentration data point in order to identify and mark abnormal data that may be caused by severe environmental disturbances or sensor saturation.

[0126] The calculation process includes:

[0127] Obtain the concentration readings of the current and previous moments, calculate the absolute value of the difference, and divide it by the preset sensor response limit threshold to obtain the concentration change rate.

[0128] A credibility index is generated by nonlinearly mapping the concentration change rate using the hyperbolic tangent function.

[0129] In this embodiment, the core of the calculation is to utilize the nonlinear mapping properties of the hyperbolic tangent function; a confidence index is introduced. The calculation formula is as follows:

[0130]

[0131] in, The dimensionless confidence index of the i-th data point is the output of this step, and its range is (0,1].

[0132] The current and previous concentration readings, [in ppm], [sourced from the sensor according to...]. [Obtained through periodic data collection];

[0133] [Sensor response limit threshold], [unit: ppm], is a key parameter that represents the maximum concentration change that the sensor can stably respond to within a sampling period. It is mainly determined based on the technical specifications of the sensor hardware or by conducting a step impact experiment on the sensor with a standard concentration of gas.

[0134] : Confidence decay coefficient, [dimensionless], a user-defined parameter that controls the rate at which confidence decreases with the rate of change of concentration; calibrated through sensor characteristic experiments; for example, applying a known concentration step signal to the sensor under laboratory conditions and acquiring its complete transient response data sequence; analyzing the sequence, marking data points in a stable state as high-confidence samples, and marking transient data points in the rising / falling edge or overshoot phase as low-confidence samples; The purpose of determining the value is to optimize so that The formula can distinguish between these two types of samples to the greatest extent, thereby achieving effective discrimination between the true dynamic response and potential data distortion;

[0135] By calculating this credibility index, the present invention provides a dynamic and quantitative weight for subsequent data correction; it can automatically identify and punish abnormal data points that change too drastically, exceed physical laws or hardware capabilities, and assign them lower credibility; this makes the data correction process no longer a blind smoothing filter, but an intelligent weighting based on the quality of the data itself, which greatly improves the fidelity and reliability of the corrected data.

[0136] Example 7:

[0137] In S3, the output of the corrected concentration sequence includes:

[0138] An exponential moving average filter is used, with the credibility index of the current data point as the dynamic weight.

[0139] The corrected concentration sequence is obtained by recursively averaging the original concentration readings with the corrected concentration values ​​from the previous time step.

[0140] Based on Example 1, this embodiment specifies the output method of the corrected concentration sequence in S3. Its purpose is to use the confidence index calculated in the previous stage to smooth the original concentration data, filter out noise and abnormal jumps, and finally output a reliable data sequence that better reflects the true concentration trend.

[0141] The output process is as follows: using an exponential moving average filter, the confidence index of the current data point is used as a dynamic weight; the original concentration reading is weighted and averaged with the previously corrected concentration value, and the corrected concentration sequence is calculated recursively.

[0142] In this embodiment, the data correction process employs an improved exponential moving average filter with dynamically adjusted filter weights; a corrected concentration value is introduced. The calculation formula is as follows:

[0143]

[0144] in, The corrected concentration value [in ppm] is the recursive calculation output of this step;

[0145] Raw concentration readings [in ppm] are obtained from the sensor.

[0146] The credibility index of the current data point, [dimensionless], is calculated by the steps in Example 6;

[0147] The previously corrected concentration value [in ppm]; to initiate this recursive calculation, initial conditions need to be set. Specifically, the first correction value is directly taken from the first original concentration value, i.e. From the second data point ( Begin by recursively calculating using this formula;

[0148] This embodiment upgrades the traditional exponential moving average filter into an intelligent, adaptive filter by introducing a confidence index as a dynamic weight. It can smooth noise while preserving the details of the true signal to the greatest extent possible. Compared to fixed-weight filters, it can better handle sudden, non-Gaussian noise impacts, ultimately producing a corrected concentration sequence. It is both smooth and continuous, and does not overly obscure the true concentration change trend, providing a high-quality data foundation for the final boundary decision.

[0149] Example 8:

[0150] In S4, the dual-criteria judgment includes:

[0151] When the average corrected concentration value of the spatial grid point is greater than or equal to the preset safe concentration threshold, and the spatial gradient of the corrected concentration at the spatial grid point is less than or equal to the preset stable gradient threshold, the spatial grid point is identified as a dangerous grid point.

[0152] If any condition is not met, the spatial grid point is determined to be a safe grid point;

[0153] Based on Example 1, this embodiment further specifies the dual-criteria judgment method in S4. Its purpose is to overcome the defect of the traditional single concentration threshold method in complex environments, which is prone to misjudging data noise or unstable turbulence as dangerous area boundaries. By introducing gradient constraints, the accuracy and robustness of boundary judgment are improved.

[0154] The judgment process is as follows:

[0155] When the average corrected concentration value of the spatial grid point is greater than or equal to the preset safe concentration threshold, and the spatial gradient of the corrected concentration at the spatial grid point is less than or equal to the preset stable gradient threshold, the spatial grid point is identified as a dangerous grid point.

[0156] If any condition is not met, the spatial grid point is determined to be a safe grid point;

[0157] In this embodiment, the dual-criteria judgment is implemented through a custom boundary decision function based on heuristic classification rules; the boundary state determination result of spatial grid point j is introduced. Logical judgment expression:

[0158]

[0159] in, The boundary state determination result of spatial grid point j is a Boolean or binary value, where 1 represents a dangerous area and 0 represents a safe area.

[0160] The spatial average of all corrected concentration values ​​within the grid point j region [in ppm] is calculated by considering multiple values ​​falling within that grid point. The values ​​are obtained by averaging.

[0161] The safe concentration threshold (in ppm) is a legal or industry standard value that is preset based on the toxicological data of the hazardous material to be tested and relevant emergency response safety procedures.

[0162] The magnitude of the spatial gradient of the corrected concentration at grid point j [in ppm / m] is obtained by calculating the rate of concentration change between grid point j and its neighboring grid points using numerical methods, such as the finite difference method.

[0163] The stable gradient threshold [unit: ppm / m] is a key empirical parameter used to distinguish stable diffusion boundaries from anomalous data jumps; its source can be obtained and set by analyzing the typical concentration gradient range of stable boundary regions in a large number of real or simulated leakage events.

[0164] By introducing the spatial gradient of concentration as a second criterion, this invention can effectively filter out pseudo-boundary points that meet the concentration threshold but not the gradient stability condition, caused by instantaneous turbulence or sensor noise. This dual constraint significantly improves the reliability of boundary determination, making the final determined boundary of the danger zone smoother, more reasonable, and more in line with the laws of physical diffusion, greatly reducing the misjudgment rate and providing a more accurate decision-making basis for subsequent emergency response.

[0165] Example 9:

[0166] The determination of the boundaries of the danger zone also includes:

[0167] The set of all grid points identified as hazardous is compiled to form the boundary of the hazardous area;

[0168] This embodiment further elaborates on the steps for determining the boundary of the hazardous area based on the method in Embodiment 8; its purpose is to integrate the judgment results of the discrete spatial grid points in the previous step into a continuous and complete geometric boundary of the hazardous area.

[0169] The determination process includes: summarizing the set of all grid points identified as hazardous to form the boundary of the hazardous area;

[0170] Specifically, after completing the detection of all spatial grid points within the probe airspace... After applying the dual criteria, a series of results will be obtained. Grid points with a value of 1 are called hazardous grid points. The operation in this step is to collect all these hazardous grid points in a spatial coordinate system. The outer envelope or contour line of this set of hazardous grid points is defined as the boundary of the hazardous area output by the present invention. In practice, this set of hazardous grid points can be processed by algorithms such as convex hull algorithm, contour line generation algorithm or edge detection algorithm in image processing to generate a smooth, closed boundary line for visualization or geographic information system analysis.

[0171] This embodiment summarizes all hazardous grid points that meet the dual criteria, transforming discrete judgment results into a macroscopic and continuous hazardous area shape. It is a key closed-loop link in achieving the final determination of the hazardous area. It provides a clear and intuitive geometric boundary that can be directly used for drawing emergency evacuation maps, establishing isolation zones, and formulating subsequent disposal decisions, making the results of the entire detection method highly practical and operable.

[0172] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention; any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0173] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for detecting hazardous materials based on unmanned aerial vehicles (UAVs), characterized in that, The specific steps include: S1. Obtain the ambient wind speed and hazardous material concentration at the current location of the drone, and calculate the hazardous material diffusion gradient based on the concentration and drone positioning information. Then, combine the ambient wind speed and hazardous material diffusion gradient to construct a coupling perturbation factor. S2. In response to the coupling disturbance factor, calculate the optimal scanning speed and determine the adaptive sampling period based on the optimal scanning speed; S3. Based on the adaptive sampling period, collect the original concentration data sequence, calculate the confidence index for the data points in the original concentration data sequence, and then use the confidence index to perform weighted correction on the original concentration data, and output the corrected concentration sequence. S4. Using the corrected concentration sequence and based on the preset safe concentration threshold and stable gradient threshold, the spatial grid points are judged by dual criteria to determine the boundary of the danger zone.

2. The method for detecting hazardous materials based on unmanned aerial vehicles according to claim 1, characterized in that: In S1, the calculation of the hazardous substance diffusion gradient includes: Collect concentration measurements and coordinate vectors from two spatial locations using a drone; The diffusion gradient of hazardous substances is calculated based on the concentration measurements and Euclidean distance between two spatial locations, combined with a preset reference length scale.

3. The method for detecting hazardous materials based on unmanned aerial vehicles according to claim 1, characterized in that: In S1, the construction of the coupling perturbation factor includes: The wind speed ratio is obtained by dividing the real-time ambient wind speed by the preset reference wind speed threshold. The gradient ratio is obtained by dividing the real-time calculated hazardous substance diffusion gradient by a preset reference gradient threshold. The coupling disturbance factor is obtained by multiplying the wind speed ratio and the gradient ratio, and then summing the sum with the basic terms.

4. The method for detecting hazardous materials based on unmanned aerial vehicles according to claim 1, characterized in that: In S2, the calculation of the optimal scan speed includes: An exponential decay model is adopted, taking the coupling disturbance factor as input, based on the preset maximum safe operating speed, and responding to the preset speed adjustment sensitivity coefficient to perform exponential decay calculation to obtain the optimal scanning speed.

5. The method for detecting hazardous materials based on unmanned aerial vehicles according to claim 1, characterized in that: In S2, the determination of the adaptive sampling period includes: Divide the preset sampling compensation coefficient by the optimal scanning speed to obtain the adjustment term; The adaptive sampling period is obtained by adding the adjustment term to the basic minimum sampling period determined by the sensor hardware.

6. The method for detecting hazardous materials based on unmanned aerial vehicles according to claim 1, characterized in that: In S3, the calculation of the credibility index includes: Obtain the concentration readings of the current and previous moments, calculate the absolute value of the difference, and divide it by the preset sensor response limit threshold to obtain the concentration change rate. A credibility index is generated by nonlinearly mapping the concentration change rate using the hyperbolic tangent function.

7. The method for detecting hazardous materials based on unmanned aerial vehicles according to claim 1, characterized in that: In S3, the output of the corrected concentration sequence includes: An exponential moving average filter is used, with the credibility index of the current data point as the dynamic weight. The original concentration readings are weighted and averaged with the previously corrected concentration values, and the corrected concentration sequence is calculated recursively.

8. A method for detecting hazardous materials based on unmanned aerial vehicles according to claim 1, characterized in that: In S4, the dual-criteria judgment includes: When the average corrected concentration value of the spatial grid point is greater than or equal to the preset safe concentration threshold, and the spatial gradient of the corrected concentration at the spatial grid point is less than or equal to the preset stable gradient threshold, the spatial grid point is identified as a dangerous grid point. If any condition is not met, the spatial grid point is determined to be a safe grid point.

9. A method for detecting hazardous materials based on unmanned aerial vehicles according to claim 8, characterized in that: The determination of the boundaries of the danger zone also includes: The set of all grid points identified as hazardous is aggregated to form the boundary of the hazardous area.

Images