Gas source tracing and positioning method, medium and system

Abstract

The invention relates to the technical field of gas source positioning, in particular to a gas source tracing positioning method, medium and system, and the method comprises the following steps: obtaining gas concentration observation information, local wind speed information and local wind direction information of a current position; establishing prior probability distribution and constructing an observation model; an Infotaxis search stage is executed; misleading criterion detection; if so, executing a Surge-Spiral search stage; when any one of the convergence conditions is met, determining that the gas source is successfully positioned and terminating the search; and when all convergence conditions are not met, returning according to the search stage before judgment. Compared with the prior art, the method solves the problem that an existing gas source tracing positioning method is carried out in a multi-machine collaborative search scene and lacks a method suitable for a single platform or a small-scale platform. According to the scheme, through a dual-stage and one-way constraint cooperation mechanism of an Infotaxis-Surge-Spiral hybrid strategy, rapid and robust traceability positioning is realized.

Description

Technical Field

[0001] This invention relates to the field of gas source location technology, specifically to a gas source tracing and location method, medium, and system. Background Technology

[0002] In emergency scenarios such as chemical reagent leaks, hazardous chemical accidents, and the spread of toxic and harmful gases, rapid and accurate location of the gas release source is a crucial prerequisite for effective control, rescue, and decontamination. Currently, utilizing unmanned systems equipped with gas sensors for atmospheric pollution monitoring and gas source tracing has become an important technological direction for improving emergency response capabilities. These systems must operate reliably in complex, low-concentration, and highly turbulent environments, placing extremely high demands on the real-time performance, robustness, and positioning accuracy of the algorithms.

[0003] Existing technologies mainly include: First, there are existing UAV-based atmospheric pollution monitoring and source tracing schemes based on single-drone hill-climbing algorithms and grid search. These schemes locate pollution sources by comparing point concentrations at different locations and using a local hill-climbing strategy to find the point of maximum concentration. However, overall, they still rely on reactive searches dependent on local gradients and are insufficiently adaptable to turbulent disturbances and intermittent plumes. Second, some studies have modeled the gas source search process as a partially observable Markov decision process, introducing Poisson observation models, Bayesian estimation, and particle filtering to perform probabilistic estimation of source term parameters. In multi-UAV collaborative scenarios, they combine niche particle swarm optimization and Infotaxis strategies to achieve collaborative search in multi-source scenarios, such as a three-dimensional spatial gas leak source localization method disclosed in CN117456137A and a gas source search method based on information entropy disclosed in CN110514567A. These methods theoretically have strong information utilization capabilities, but the algorithm structure is complex, the computational load is large, and they are mainly geared towards multi-source, multi-drone collaborative scenarios, which makes it difficult for single-drone applications in complex open environments and other scenarios with strict real-time constraints. Thirdly, at the level of a single algorithm, the Infotaxis algorithm, as a typical information-driven source tracing method, effectively solves the problem of traditional gradient tracking failure in low-concentration turbulent environments by maximizing the expected reduction in information entropy. However, this algorithm is quite sensitive to spatial discrete resolution and the overall shape of the information field: when the grid is too coarse, it is easy to detach from the plume, leading to premature convergence; when the grid is too fine, the computational complexity increases with the increase of the grid size multiplied by higher orders, making it difficult to meet the real-time requirements of engineering projects, and it also suffers from insufficient success rate in low-concentration, intermittent plume environments. In addition, traditional Surge-type algorithms borrow from the chemotaxis behavior of moths, relying on upwind tracing and lateral scanning for searching. Although they have a certain robustness to turbulent disturbances, lateral scanning is prone to the dispersion of the search range due to wind direction fluctuations, and there is a risk of getting stuck in a repetitive scanning dead loop. The overall search steps are relatively large, and the positioning accuracy and efficiency are relatively limited.

[0004] For example, the prior art CN106918367A discloses a method for a robot to actively search for and locate odor sources. It distinguishes the search method according to the wind speed, which may require traversal search or upwind search. However, this method takes a long time to find the initial location, and it is also very easy to be misled in the case of turbulent disturbance and intermittent smoke plumes, making it difficult to find the real odor source.

[0005] In summary, existing technologies lack an algorithm that can automatically trace sources based on environmental conditions and observation quality under single-platform or small-scale platform conditions. At the same time, existing methods lack quantitative identification of "information misleading" states and are prone to inefficient searches under the guidance of local erroneous information. In addition, existing methods also struggle to balance search success rate, path length, and computational cost. Summary of the Invention

[0006] The purpose of this invention is to provide a gas source tracing and location method, medium, and system to solve at least one of the aforementioned problems. This addresses the issue that existing gas source tracing and location methods are developed for multi-machine collaborative search scenarios, lacking suitable methods for single-platform or small-scale platforms. This solution achieves rapid and robust source tracing and location through a two-stage, unidirectional constraint collaborative mechanism using an Infotaxis-Surge-Spiral hybrid strategy.

[0007] The objective of this invention is achieved through the following technical solution: The first aspect of this invention discloses a method for tracing and locating a gas source, comprising the following steps: S1: Obtain gas concentration observation information, local wind speed information, and local wind direction information at the current location; S2: Model and mesh the target area, establish a prior probability distribution for the candidate gas source locations at the center of each grid based on the gas diffusion model, and construct an observation model; S3: Execute the Infotaxis search phase: Based on the gas concentration observation information, perform Bayesian update to obtain the posterior probability distribution; calculate the pre-movement entropy and the expected reduction of information entropy for feasible movement actions, and select the action with the largest expected reduction of information entropy as the next movement direction and step size; S4: Misleading Criterion Detection: During the Infotaxis search, if at least one misleading criterion is met, the information misleading criterion is triggered, and the search phase is switched to the Surge-Spiral search phase; if not all misleading criteria are met, step S3 is repeated and the Infotaxis search phase is maintained. S5: Execute the Surge-Spiral search phase, including the Surge sub-phase and the Spiral sub-phase: determine the headwind direction based on local wind speed and direction information; S5-1: when the gas concentration observation information is greater than or equal to the set sensing threshold, enter the Surge sub-phase: move upward along the headwind direction and biased towards the direction of gas concentration increase or change trend indication and search for gas source; S5-2: when the gas concentration observation information is less than the set sensing threshold, enter the Spiral sub-phase: move in a circle with the position of the last effective Surge sub-phase as the center and search for plume signal; When any convergence condition is met, the gas source is determined to be successfully located and the search is terminated; when all convergence conditions are not met, the search is returned to step S3 or step S5 depending on the search stage before the determination; the convergence conditions include: (a) the pre-movement entropy is less than the fourth set threshold; (b) the gas concentration observation information is greater than the gas source confirmation threshold.

[0008] Preferably, in step S2, the gas diffusion model is a Gaussian diffusion model, a convection-diffusion equation, or a concentration field. The observation model is constructed using a Poisson distribution, and the arrival rate parameter of the Poisson distribution is obtained by mapping and transforming the theoretical concentration at each candidate gas source location. The theoretical concentration is obtained through a gas diffusion model.

[0009] Preferably, in step S3, the pre-movement entropy is the Shannon entropy corresponding to the posterior probability distribution; The expected reduction in information entropy is the difference between the post-decision entropy and the pre-decision entropy of a feasible movement action; the post-decision entropy is the cumulative value of the Shannon entropy corresponding to each of the possible gas concentration observation results of the feasible movement action, weighted and summed according to their occurrence probabilities.

[0010] Preferably, in step S4, the misleading criteria include: (a) within a continuous number of movement steps, the average rate of change of gas concentration observation information is negative and the absolute value of the average rate of change is greater than a first set threshold; (b) within a continuous number of movement steps, the increase in entropy before movement is greater than a second set threshold.

[0011] Preferably, in step S5, the step size of the Surge sub-stage is dynamically adjusted according to the distance from the gas source. The circular motion performed in the Spiral sub-stage involves determining candidate paths for the circular motion based on maneuverability constraints and environmental constraints, establishing a series of continuous observation points, sequentially traversing each observation point, and searching for plume signals.

[0012] Preferably, in step S5, the circular motion performed in the Spiral sub-stage skips the search of the arc region formed by the partial angle of the downwind direction.

[0013] Preferably, the circular motion performed in the Spiral sub-stage involves assigning probability weights to observation points along the circular motion path based on the posterior probability distribution output from the last Infotaxis search stage, and selecting regions with values ​​greater than a third set threshold for searching.

[0014] Preferably, in step S5, the circular motion performed in the Spiral sub-stage is as follows: if the circular motion performed at the current radius does not find a plume signal, the search radius is increased and the circular motion is performed again; when the circular motion finds a plume signal, the process switches to the Surge sub-stage for upward motion.

[0015] A second aspect of the present invention discloses a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the steps of the gas source tracing and locating method described in any of the preceding claims.

[0016] A third aspect of the present invention discloses a gas source tracing and positioning system, including a platform, a processor mounted on the platform, and a memory for executing instructions. When the processor executes the instructions, it implements the steps of any of the above-described gas source tracing and positioning methods.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1) Deep integration of information-driven and bio-inspired approaches: This invention does not simply replace the Infotaxis and Surge-Spiral scanning strategies with each other, but establishes a quantitative coupling relationship between the two through "information misleading criteria" and "probabilistic screening mechanism", so that the information entropy evaluation results directly participate in the path selection of the Surge-Spiral stage, thus forming a truly meaningful hybrid source tracing algorithm.

[0018] 2) Interpretable mode switching mechanism: By simultaneously examining the monotonicity of the concentration time series and the diffusion degree of the posterior probability distribution, the failure state of information-driven search is detected, avoiding the blind iteration of Infotaxis in the local abnormal information field, and improving the interpretability and adjustability of the algorithm behavior.

[0019] 3) Joint constraints of multi-source observation information: The multi-channel gas observation information output by the gas concentration sensing module (used to characterize the trend or direction of concentration change) is combined with the wind field information output by the wind speed and direction measurement module to provide joint constraints for the selection of the upward motion direction and the determination of the lateral scanning range, so that the search path is more in line with the actual plume transmission direction, and the risk of invalid scanning and repetitive cycles is reduced while maintaining the success rate.

[0020] 4) Optimization of the trade-off between computational overhead and search performance: When information is sufficient and the probability distribution is concentrated, Infotaxis is used for detailed search. When information is disturbed or the distribution is divergent, Surge-Spiral is used for large-step, coarse-grained search, thereby achieving an overall trade-off between search success rate, average number of steps and computational complexity. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the gas source tracing and positioning system of the present invention.

[0022] Figure 2 This is a schematic diagram of the overall process of the Infotaxis-Surge-Spiral hybrid source tracing and localization method of the present invention.

[0023] Figure 3 This is a schematic diagram of the probability distribution and pre-entropy evolution during the Infotaxis information-driven phase.

[0024] Figure 4 This diagram illustrates the triggering of information misleading criteria and mode switching.

[0025] Figure 5 This is a schematic diagram of the Spiral scan trajectory guided by probability screening during the Surge-Spiral phase.

[0026] The components include: 1-mounting platform; 2-gas concentration sensing module; 3-signal acquisition and processing components; 4-probe assembly; 5-wind speed and direction measurement module; and 6-extension probe. Detailed Implementation

[0027] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings. These embodiments are implemented based on the technical solution of the present invention, and detailed implementation methods and specific operation processes are given. However, the scope of protection of the present invention is not limited to the following embodiments.

[0028] The “misleading criterion” described in this invention refers to a series of quantitative judgment conditions used to determine whether the Infotaxis search phase fails due to a decline in information quality or environmental interference.

[0029] Currently, there is a lack of a hybrid source tracing algorithm that can deeply integrate information-driven strategies and bio-inspired tracking strategies on a single platform or small-scale platform, and can automatically switch modes based on environmental conditions and observation quality. At the same time, existing methods lack quantitative identification and mode switching mechanisms for "information misleading" states, making them prone to inefficient searches under the guidance of local erroneous information. In addition, existing methods also lack systematic utilization of the posterior probability distribution of gas source locations, failing to use probability information to constrain or filter subsequent scanning areas and path selection, and failing to form effective joint constraints between gas concentration observation and environmental information such as wind speed and direction, thus making it difficult to balance search success rate, path length, and computational cost.

[0030] To address the shortcomings of existing technologies, this invention proposes the following technical solution: a gas source tracing and localization method and system based on an Infotaxis-Surge-Spiral hybrid strategy, comprising the following steps: (1) Construct an environmental modeling framework that supports gas concentration observation and wind speed and direction observation, treat the gas source location as a random variable, and characterize the observation uncertainty under intermittent plume conditions based on a statistical observation model; (2) In the initial stage of the search, the Infotaxis algorithm is used to carry out information-driven search with the goal of maximizing the expected reduction in information entropy, so as to quickly compress the uncertain area of ​​gas source location; (3) Construct an “information misleading criterion” by using concentration trends and probability distribution patterns to judge in real time whether information-driven search has failed or fallen into an inefficient state, and switch to the Surge-Spiral stage once triggered; (4) In the Surge-Spiral stage, the wind speed and direction are used to estimate the comprehensive headwind direction. The two-stage cycle of Surge and Spiral is adopted: when there is a signal, the headwind direction is moved upstream; when there is no signal, the circular motion is scanned with the most recent effective plume position as the center, skipping the arc corresponding to the preset downwind angle interval to avoid dead loop; when not found, the radius of the circle is gradually expanded. (5) During the Spiral scan, the posterior probability distribution of the gas source location in the Infotaxis stage is used as a priori guide to screen candidate scan areas, skip low-probability candidate locations, and avoid invalid searches. Through the above-mentioned two-stage, unidirectional constraint collaborative mechanism, the gas source location is output when the probability distribution of the gas source location converges and the spatial distance meets the accuracy requirements, thus achieving fast and robust source tracing and positioning.

[0031] Example 1 A gas source tracing and localization method based on an Infotaxis-Surge-Spiral hybrid strategy includes the following steps: S1, the platform moves within the target area, uses the gas concentration sensing module on it to obtain the gas concentration observation information (at least including the gas concentration) at the current location, and uses the wind speed and direction measurement module to obtain the local wind speed and direction information. S2, based on the preset gas diffusion model, the target area is modeled as a grid state space of preset size; a uniformly distributed prior probability distribution is established for the candidate gas source positions at the center of each grid, and an observation model based on the gas concentration observation information at the current position is constructed. S3, Execute the Infotaxis search phase: Based on the gas concentration observation information at the current location, perform a Bayesian update on the probability distribution of the gas source location to obtain the posterior probability distribution; calculate the pre-movement entropy and the expected reduction of information entropy for each feasible movement action, and select the action with the largest reduction of expected information entropy as the next movement direction and step size to maximize the reduction of uncertainty in the gas source location. S4, Construct and monitor information misleading criteria: During the Infotaxis search process, continuously monitor the gas concentration change trend and the posterior probability distribution of the gas source location in the most recent few steps. When at least one of the following conditions is met, the information misleading criteria is triggered: (a) the plume leaves the grid path, that is, within the preset number of consecutive movement steps, the observed concentration shows a monotonically decreasing trend and the cumulative decrease exceeds the first set threshold; (b) the posterior probability distribution shows a significant diffusion or multi-peak splitting pattern in space, that is, within the preset number of consecutive movement steps, the increase in the pre-movement entropy is higher than the second set threshold. If the information misleading criterion is not triggered, the Infotaxis search phase is maintained, and steps S3-S4 are repeated; if the information misleading criterion is triggered, the process switches to the Surge-Spiral search phase. S5, Execute the Surge-Spiral search phase: (a) Determine the upwind direction based on wind speed and direction information; (b) Surge sub-phase: When the gas concentration is detected to be higher than the set sensing threshold, move upward along the upwind direction and biased towards the direction of rising gas concentration or the direction of change trend indication, and dynamically reduce the step size as the search progresses; (c) Spiral sub-phase: When the gas concentration is detected to be lower than the set sensing threshold, it is determined to leave the plume, and the plume is searched in a circular motion within the set initial radius around the position of the most recent effective Surge sub-phase. During the circular motion, the downwind part of the angle arc is skipped; if the current radius search does not find the plume, the circular radius is increased by the preset step size, and the circular motion search is repeated; In the Spiral sub-stage, based on the posterior probability distribution of the gas source location output from the current or most recent Infotaxis search stage obtained in step S3, the candidate regions for circular motion scanning are filtered, and regions with probability values ​​higher than the third set threshold in the posterior probability distribution are selected as the target regions for Spiral scanning, while regions with probability values ​​lower than the third set threshold are skipped directly.

[0032] In the above method, the Infotaxis search phase can be dynamically switched to the Surge-Spiral search phase. Furthermore, when either of the two search phases meets one of the following convergence conditions, the gas source is determined to be successfully located and the search is terminated: (a) the pre-movement entropy calculated from the posterior probability distribution of the gas source location is lower than the fourth set threshold; (b) the gas concentration is detected to be continuously higher than the preset gas source confirmation threshold.

[0033] Furthermore, in step S2, the gas diffusion model is a Gaussian diffusion model, used to characterize the diffusion law of gas in the atmospheric environment; the grid state space is a reference size of 25×25 squares. In practical applications, the number of rows, columns, and single square side lengths of the grid can be adaptively adjusted according to the computing power of the platform, the area of ​​the target region, the gas diffusion characteristics, and the detection accuracy of the sensing module, so as to achieve a dynamic balance between search accuracy and computational efficiency; the observation model is constructed using a Poisson distribution. The arrival rate parameter of the Poisson distribution is obtained by mapping and transforming the theoretical concentration of each candidate gas source location. The theoretical concentration is calculated by the Gaussian diffusion model and used to characterize the observation uncertainty under intermittent plume conditions.

[0034] Furthermore, the calculation process for the expected reduction in information entropy in step S3 is as follows: the pre-movement entropy is the Shannon entropy corresponding to the posterior probability distribution of the current gas source location, used to quantify the uncertainty of the current gas source location; for each candidate movement point corresponding to a feasible movement action, all possible gas concentration observation results that may be obtained after moving to the candidate movement point are estimated one by one, and the occurrence probability corresponding to each gas concentration observation result is calculated; based on each gas concentration observation result and its corresponding occurrence probability, the post-decision entropy of each candidate movement point is calculated, and the post-decision entropy is specifically the cumulative value of the Shannon entropy corresponding to each possible gas concentration observation result after weighted summation according to its occurrence probability; the expected reduction in information entropy corresponding to each feasible movement action can be obtained by the difference between the pre-movement entropy and the post-decision entropy of each candidate movement point.

[0035] Furthermore, in step S4, the determination of condition (a) is as follows: based on a sliding time window of a preset length, calculate the average rate of change of the concentration observation value corresponding to the number of consecutive moving steps within the window. When the average rate of change is negative and its absolute value is greater than the first threshold, it is determined as "plume detaches from the grid path", satisfying condition (a); the determination of condition (b) is as follows: calculate the Shannon entropy of the posterior probability distribution and its change within a preset number of steps. When the change increases by more than the second threshold in several consecutive steps, it is determined as "posterior probability diffusion or multi-peak". When either condition "plume detaches from the grid path" or "posterior probability diffusion or multi-peak" is met, the mode switch to the Surge-Spiral stage is triggered.

[0036] Furthermore, step S7 involves screening candidate regions for the circular spiral scan, specifically including: (a) planning candidate paths for the circular motion scan based on the mobility of the platform and environmental constraints, and pre-setting a series of continuous spiral observation points along the path; (b) mapping the posterior probability distribution of the gas source location to each pre-set spiral observation point and obtaining the probability weight corresponding to each point; (c) traversing the pre-set spiral observation points in a pre-set scanning order, and skipping the location if the probability weight of the current observation point is lower than the third set threshold, and directly evaluating the next pre-set spiral observation point; (d) only driving the platform to move to the point to perform spiral scan observation when a pre-set spiral observation point with a probability weight higher than the third set threshold is encountered.

[0037] A gas source tracing and localization system for implementing the above-mentioned gas source tracing and localization method based on the Infotaxis-Surge-Spiral hybrid strategy includes: (a) a platform for moving within a target area; (b) a gas concentration sensing module installed on the platform for acquiring local gas concentration observations; (c) a wind speed and direction measurement module for acquiring atmospheric wind speed and direction information near the platform; and (d) a tracing decision and control module, which is communicatively connected to the gas concentration sensing module, the wind speed and direction measurement module, and the platform, and is capable of executing the above-mentioned method steps, generating search path decisions based on the Infotaxis-Surge-Spiral hybrid tracing algorithm, sending control commands to the platform to drive its movement, and outputting gas source location estimation results when the convergence criterion is met.

[0038] Furthermore, the platform is a drone, and the gas concentration sensing module and wind speed and direction measurement module are installed on the drone. During flight, the drone performs maneuvering flight according to the search path generated by the Infotaxis-Surge-Spiral hybrid strategy to complete the source tracing and location of the gas source.

[0039] Furthermore, the gas concentration sensing module is a gas monitoring module mounted on the mounting platform, including a probe assembly and a signal acquisition and processing assembly; the probe assembly includes at least one gas sensing unit, used to acquire gas observation information for different target gases such as organic gases and inorganic gases; the signal acquisition and processing assembly is used to acquire the output of the gas sensing unit and obtain gas concentration observation information to assist in the selection of the upward motion direction and the determination of the lateral scanning range in step S6.

[0040] Furthermore, the source tracing decision and control module includes: (a) a sensor data preprocessing unit, used to filter, normalize, and aggregate the time window of the raw signal output by the gas concentration sensing module; (b) an environment modeling and probability update unit, used to perform Bayesian updates based on the Gaussian diffusion model and the observation model constructed using the Poisson distribution, calculate the posterior probability distribution of the gas source location and its corresponding Shannon entropy, estimate the prior probability distribution of each candidate moving point (the point after moving according to each feasible moving action) and calculate the post-decision entropy, providing data support for the determination of information misleading criteria and the calculation of the expected reduction in information entropy; (c) a mode determination and switching unit, used to perform information misleading criteria based on the gas concentration change trend and the posterior probability distribution shape, and control the unidirectional switching from the Infotaxis search stage to the Surge-Spiral search stage; (d) a path planning and control unit, used to combine the probability screening results, wind speed and direction information, and the maneuver constraints of the platform to generate the next movement command and send it to the platform.

[0041] Furthermore, this gas source tracing and locating system can be applied in tracing gas leaks in sudden environmental events or locating gas leaks in urban underground pipelines; the system is deployed for environmental monitoring, safety supervision, and pipeline leak location. The phased collaborative mechanism of the Infotaxis-Surge-Spiral hybrid strategy is suitable for complex environments with low concentrations, strong turbulence, and intermittent plume occurrences in application scenarios.

[0042] Example 2 A gas source tracing and localization method and system based on an Infotaxis-Surge-Spiral hybrid strategy is disclosed. The system includes a platform, a gas concentration sensing module, a wind speed and direction measurement module, and a tracing decision and control module. The platform can be a drone, unmanned vehicle, or other autonomous vehicle capable of movement, used to perform search tasks in two-dimensional or three-dimensional space. The gas concentration sensing module, located at the front or bottom of the platform, includes multiple gas sensing chips and can acquire local multi-point concentration information through multi-channel acquisition. The wind speed and direction measurement module can employ an ultrasonic anemometer, a mechanical anemometer, or a comprehensive estimation unit based on inertial navigation and airflow models to output local instantaneous wind speed and direction. The tracing decision and control module, typically implemented by an embedded computing unit or a host computer, executes the proposed Infotaxis-Surge-Spiral hybrid tracing and localization algorithm and issues motion control commands to the platform.

[0043] In this embodiment, the platform can be a carrier with autonomous movement capabilities (such as a drone), for example a multi-rotor drone, such as... Figure 1 As shown. Specifically, the platform 1 is equipped with a gas concentration sensing module 2, a signal acquisition and processing component 3, a probe component 4, a wind speed and direction measurement module 5, and an extension probe 6. Part of the gas concentration sensing module 2 and the wind speed and direction measurement module 5 are respectively located at the end of the extension probe 6 at the front of the platform 1, so that the gas and wind field observation points are far away from the rotor downwash and the area disturbed by the aircraft. The gas concentration sensing module 2 further includes the probe component 4 and the signal acquisition and processing component 3. The probe component 4 is located at the end of the extension probe 6 and is used to generate multi-channel gas observation signals. The signal acquisition and processing component 3 can be located inside or near the platform 1 and is connected to the probe component 4 via cables or a bus. It is used to acquire and process multi-channel signals to form gas observation information for source tracing calculations and indication information reflecting changes in gas concentration. The wind speed and direction measurement module 5 can be arranged adjacent to the probe component 4 or at a preset distance. It is used to output local instantaneous wind speed and direction information, which is then fused or weighted by the source tracing decision and control module to obtain comprehensive wind direction information. The source tracing decision and control module communicates with the flight control or motion control unit of the onboard platform 1, and is used to generate search path decisions and issue control commands based on gas observation information and wind field information.

[0044] The system workflow is as follows: After the platform enters the target area, it periodically collects gas concentration and wind field information. The source tracing decision and control module runs the method of this scheme to generate a search path until the gas source is located.

[0045] The overall process of the Infotaxis-Surge-Spiral hybrid source tracing and localization method involved in this solution is as follows: Figure 2As shown, the specific steps are as follows: First, initialize the probability distribution of the gas source location and algorithm parameters (including the initial radius of the Spiral stage, various thresholds, etc.). The platform, carrying sensing modules (including a gas concentration sensing module and a wind speed and direction measurement module), enters the target area to begin the search. The Infotaxis search phase is executed first, performing Bayesian updates and Shannon entropy calculations based on gas concentration observation information to select the optimal movement action. Simultaneously monitor information misleading criteria; if not triggered, the Infotaxis search continues; if triggered, switch to the Surge-Spiral stage, using a combined Surge upwind search and Spiral circular scan. During the Spiral stage, a probability filtering mechanism optimizes the scanning area until the convergence condition is met, at which point the gas source location is output, and the search terminates.

[0046] In this embodiment, the environmental modeling process is achieved through a gas concentration sensing module in conjunction with a Gaussian diffusion model: the target area is discretized into a two-dimensional grid of preset size, and each center position within the grid is defined as a potential gas source candidate point; based on the gas diffusion and propagation law, an observation model is established and a Poisson distribution is used to characterize the observation uncertainty, wherein the arrival rate parameter of the Poisson distribution is obtained by mapping the theoretical concentration of each candidate gas source location, and the theoretical concentration is calculated by the Gaussian diffusion model; the gas concentration sensing module collects the actual gas concentration observation value (or its equivalent count form) at the current location, and uses the observation value as a Poisson random variable, and by comparing the actual observation value with the model prediction value corresponding to the theoretical concentration, the probability update of each potential gas source candidate point is completed, forming the posterior probability distribution of the gas source location.

[0047] The above modeling can utilize simplified convection-diffusion equations or the concentration field in a statistical sense provided by a particle system simulation environment. This invention is not limited to a specific diffusion model form.

[0048] In one optional embodiment, the probe assembly includes multiple gas sensing units, which can be semiconductor, electrochemical, or other types of gas-sensitive devices. Different sensing materials or ranges can be selected based on the target application scenario to improve the response capability to different target gases or different concentration ranges. As an example, the probe assembly can integrate multiple sensing units to generate response signals for target gases such as H2S, ethanol, acetone, VOCs, formaldehyde, and CH4. The above is merely an example and does not constitute a limitation. This method and system are not targeted at specific gas types and are applicable to various gas leak tracing scenarios that meet diffusion and propagation characteristics. Multiple gas sensing units can be integrated into the probe housing using a planar array or similar method, and gas observation information can be obtained through multi-channel acquisition. The signal acquisition and processing component can generate indication information reflecting the trend or direction of gas concentration change based on time window difference, threshold discrimination, trend fitting, or a combination thereof, to assist in the selection of the upward movement direction and the determination of the lateral scanning range in step S6.

[0049] In this embodiment, the core steps of the Infotaxis-Surge-Spiral hybrid source tracing and localization algorithm in the Infotaxis stage include: using a uniform distribution or a coarse prior as the initial probability distribution of the gas source location; after each observation, calculating the likelihood function of each candidate location under the current observation according to the Poisson observation model, and updating the posterior probability distribution using the Bayesian formula, wherein the observation information is obtained by the multi-channel output of the gas concentration sensing module after preprocessing by the signal acquisition and processing component, and is used to characterize the gas concentration observation state at the current moment; calculating the Shannon entropy (i.e., the pre-movement entropy) of the current posterior probability distribution; for each candidate action (e.g., moving a certain step in eight nearby directions), estimating the possible observation results and their probabilities under this action, and thus obtaining the expected information entropy (i.e., the post-decision entropy) after the action; comparing the pre-movement entropy with the post-decision entropy of each action, and selecting the action that can produce the largest expected entropy reduction as the next decision.

[0050] Through the above process, the Infotaxis stage can quickly narrow the search range when information is abundant and the plume structure is relatively stable. For example... Figure 3 As shown, the evolution of probability distribution and information entropy in the Infotaxis information-driven stage is illustrated below: Initially (time step 0), the uncertainty of the gas source location is the highest, and the entropy value before movement is close to 6; as the search progresses (time steps 5-15), the platform continuously updates the posterior probability distribution based on observation information, gradually focusing on high-probability areas, and the entropy value before movement shows a decreasing trend. By time step 20, the entropy value drops below 1, indicating that the uncertainty of the gas source location has been significantly reduced, and the target gas source is close to being locked. This example intuitively demonstrates the core advantage of the Infotaxis algorithm in rapidly compressing uncertain regions of gas source location by maximizing the expected reduction in information entropy.

[0051] To avoid invalid iterations of Infotaxis in areas with abnormal local information, this invention sets up an information misleading criterion in the embodiments: linear fitting or difference calculation is performed on the concentration observations of the most recent few steps (e.g., 2 to 5 steps). When the average rate of change is negative and the absolute value exceeds a preset threshold, it indicates that the plume has deviated from the grid path. The concentration observations are obtained by summarizing the multi-channel gas observation information output by the gas concentration sensing module within a preset time window; linear fitting or difference calculation is performed on the moving entropy of the most recent few steps (e.g., 2 to 5 steps). If its increase exceeds a preset threshold, it indicates that the posterior probability is in a diffuse or multi-peak state.

[0052] When any of the above conditions are met, the current local optimization strategy based on information entropy is considered unreliable, triggering a unidirectional mode switch from the Infotaxis stage to the Surge-Spiral stage. The threshold value can be calibrated according to the simulation scenario and actual application requirements. For example... Figure 4 As shown, the specific example process of information misleading criterion dual-condition triggering and mode switching is as follows: The two broken lines in the figure correspond to the feature changes of the two criteria, respectively. Figure 4 As shown by the broken line of a, in time steps 15-17, the gas concentration shows a monotonically decreasing trend and the cumulative decrease exceeds the first set threshold, indicating that the platform has deviated from the main plume band, corresponding to the "plume leaving the grid path" criterion trigger scenario. Figure 4 As shown by the broken line in b, during time steps 14-16, the pre-movement entropy value shows a continuous upward trend and the increase exceeds the second set threshold, indicating that the posterior probability distribution exhibits a significant diffusion or multi-peak splitting pattern, and the uncertainty of the gas source location increases in the opposite direction, corresponding to the "posterior probability diffusion or multi-peak" criterion trigger scenario. When either condition in the two scenarios is met, it is determined that the information-driven strategy is misled, and the system immediately switches from the Infotaxis stage to the Surge-Spiral stage, re-capturing the plume signal through the coordinated action of upwind tracing and circular scanning, avoiding falling into inefficient search.

[0053] In this embodiment, the Infotaxis-Surge-Spiral hybrid source tracing and localization algorithm operates as follows in the Surge-Spiral phase: The instantaneous and historical wind directions output by the wind speed and direction measurement module are used to obtain the comprehensive wind direction through time-weighted averaging. If the current concentration observation is higher than a set threshold, the system enters Surge mode: the platform moves along the comprehensive counter-wind direction and towards the side of local concentration increase, with the step size gradually decreasing as it approaches the gas source. The direction of the local concentration increase is determined by the gas observation changes within adjacent time steps or the same time window, providing directional indication rather than a precise numerical gradient. If the current concentration observation is lower than the threshold, it is determined that the platform has left the main plume zone, and the system enters Spiral mode: the location of the most recently detected effective plume is recorded as the center, an initial circumference radius is set (e.g., 20~50m), and the system moves clockwise... Alternatively, it can perform a counter-clockwise circular motion, skipping the downwind arc area of ​​-75° to 75° during the circular motion to avoid dead loops caused by airflow circulation. The downwind arc area is determined relative to the overall wind direction, and its angle range can be preset or adjusted according to environmental parameters. During the circular motion scanning process, the latest posterior probability distribution from the Infotaxis stage is called to assign probability weights to candidate observation points on the circular path. Points with weights below the set threshold are directly ignored, and only high-probability points are scanned. Ignoring points includes not entering the corresponding position to perform a scan or not performing effective observation sampling at that position. If a plume signal is not detected again after completing a full scan within the current radius, the circular radius is increased by a preset step size (e.g., 10~20m), and the above circular motion and probability screening process is repeated. Once a significant odor signal is detected again, the system switches back to Surge mode to continue tracing upwards. Figure 5 As shown, a specific example of the scanning trajectory guided by probability screening in the Spiral stage is as follows: Taking the most recent valid plume location as the center, the initial scanning radius is set to 20m. By calling the latest posterior probability distribution in the Infotaxis stage, the probability weight of candidate observation points on the circular path is evaluated, and high-probability observation areas ("legal search positions" in the figure) are selected. Areas with weights lower than the third set threshold ("skipped search positions" in the figure) are directly skipped. When no plume is found within this radius, the radius is adjusted to 30m by a preset step size (e.g., 10m), and the above probability screening and circular scanning process is repeated until the plume signal is recaptured, effectively reducing invalid scanning paths and improving search efficiency.

[0054] The above process provides a "priority search zone" at the information level by probability distribution and a "rough transmission path" at the physical level by wind field information, thereby achieving the synergy between information-driven and wind direction tracking.

[0055] In this embodiment, the Infotaxis-Surge-Spiral hybrid source tracing and localization algorithm has the following comprehensive process: Initialize the probability distribution and algorithm parameters (including the initial circumference radius, radius increment, and downwind skip angle in the Spiral phase), and enter the Infotaxis phase; at each time step, perform: collect gas concentration observation information and wind field information, Bayesian update, information entropy, and calculate misleading criteria. If no misleading criterion is triggered, continue with Infotaxis action selection; if the misleading criterion is triggered, switch to the Surge-Spiral phase, loop between Surge and Spiral, and call the probability filtering mechanism during the Spiral process; regardless of the phase, as long as the pre-movement entropy is less than the termination threshold, or the detected gas concentration is greater than the preset threshold, it is determined that the gas source has been found, the gas source location estimation result is output, and the algorithm terminates.

[0056] It should be noted that in various application scenarios of this patent, the algorithm parameters can be adjusted accordingly to meet different needs, and this invention does not limit this.

[0057] In summary, this invention proposes a gas source tracing and localization method and system based on an Infotaxis-Surge-Spiral hybrid strategy. The system includes a platform, a gas concentration sensing module, a wind speed and direction measurement module, and a tracing decision and control module. The gas concentration sensing module acquires multi-point concentration information of intermittent plumes within the target area, the wind speed and direction measurement module senses the atmospheric flow characteristics in real time, and the tracing decision and control module executes the Infotaxis-Surge-Spiral hybrid tracing and localization algorithm based on the observation data. This method uses the Infotaxis algorithm as the initial search strategy, constructs the gas source location probability distribution through Bayesian updates, and calculates the information entropy change to maximize the expected information gain in the gas source search. During the search, the concentration trend and probability distribution pattern are monitored. When specific conditions are met, it is determined that the information-driven strategy is misled, and the system switches to the Surge-Spiral stage, performing an upward tracing motion along the upwind direction. When the odor signal is interrupted, a circular scanning motion is performed. Simultaneously, the gas source probability distribution in the Infotaxis stage is used to filter the scanning area, focusing on searching high-probability zones. This invention, through phased collaboration and unidirectional constraints, maintains a high success rate compared to a single algorithm in low-concentration turbulence scenarios, reduces the number of search steps and computational complexity, and improves the real-time performance and robustness of source tracing and localization.

[0058] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0059] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0060] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0061] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0062] The above description of the embodiments is provided to enable those skilled in the art to understand and use the invention. It will be apparent to those skilled in the art that various modifications can be made to these embodiments, and the general principles described herein can be applied to other embodiments without inventive effort. Therefore, the present invention is not limited to the above embodiments, and any improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the invention should be within the protection scope of the present invention.

Claims

1. A method for tracing and locating a gas source, characterized in that, Includes the following steps: S1: Obtain gas concentration observation information, local wind speed information, and local wind direction information at the current location; S2: Model and mesh the target area, establish a prior probability distribution for the candidate gas source locations at the center of each grid based on the gas diffusion model, and construct an observation model; S3: Execute the Infotaxis search phase: Based on the gas concentration observation information, perform Bayesian update to obtain the posterior probability distribution; calculate the pre-movement entropy and the expected reduction of information entropy for feasible movement actions, and select the action with the largest expected reduction of information entropy as the next movement direction and step size; S4: Misleading Criterion Detection: During the Infotaxis search, if at least one misleading criterion is met, the information misleading criterion is triggered, and the search phase is switched to the Surge-Spiral search phase; if not all misleading criteria are met, step S3 is repeated and the Infotaxis search phase is maintained. S5: Execute the Surge-Spiral search phase, including the Surge sub-phase and the Spiral sub-phase: determine the headwind direction based on local wind speed and direction information; S5-1: when the gas concentration observation information is greater than or equal to the set sensing threshold, enter the Surge sub-phase: move upward along the headwind direction and biased towards the direction of gas concentration increase or change trend indication and search for gas source; S5-2: when the gas concentration observation information is less than the set sensing threshold, enter the Spiral sub-phase: move in a circle with the position of the last effective Surge sub-phase as the center and search for plume signal; When any convergence condition is met, the gas source is determined to be successfully located and the search is terminated; when all convergence conditions are not met, the search is returned to step S3 or step S5 depending on the search stage before the determination; the convergence conditions include: (a) the pre-movement entropy is less than the fourth set threshold; (b) the gas concentration observation information is greater than the gas source confirmation threshold.

2. The gas source tracing and positioning method according to claim 1, characterized in that, In step S2, the gas diffusion model is a Gaussian diffusion model, a convection-diffusion equation, or a concentration field. The observation model is constructed using a Poisson distribution, and the arrival rate parameter of the Poisson distribution is obtained by mapping and transforming the theoretical concentration at each candidate gas source location. The theoretical concentration is obtained through a gas diffusion model.

3. The gas source tracing and positioning method according to claim 1, characterized in that, In step S3, the pre-movement entropy is the Shannon entropy corresponding to the posterior probability distribution; The expected reduction in information entropy is the difference between the post-decision entropy and the pre-decision entropy of a feasible movement action; the post-decision entropy is the cumulative value of the Shannon entropy corresponding to each of the possible gas concentration observation results of the feasible movement action, weighted and summed according to their occurrence probabilities.

4. The gas source tracing and positioning method according to claim 1, characterized in that, In step S4, the misleading criteria include: (a) within a continuous number of movement steps, the average rate of change of gas concentration observation information is negative and the absolute value of the average rate of change is greater than a first set threshold; (b) within a continuous number of movement steps, the increase in entropy before movement is greater than a second set threshold.

5. The gas source tracing and positioning method according to claim 1, characterized in that, In step S5, the step size of the Surge sub-stage is dynamically adjusted based on the distance from the gas source. The circular motion performed in the Spiral sub-stage involves determining candidate paths for the circular motion based on maneuverability constraints and environmental constraints, establishing a series of continuous observation points, sequentially traversing each observation point, and searching for plume signals.

6. The gas source tracing and positioning method according to claim 1, characterized in that, In step S5, the circular motion performed in the Spiral sub-stage skips the search of the arc region formed by the partial angle of the downwind direction.

7. The gas source tracing and positioning method according to claim 1, characterized in that, In step S5, the circular motion performed in the Spiral sub-stage involves assigning probability weights to the observation points on the circular motion path based on the posterior probability distribution output from the last Infotaxis search stage, and selecting regions with values ​​greater than a third set threshold for searching.

8. The gas source tracing and positioning method according to claim 1, characterized in that, In step S5, the circular motion performed in the Spiral sub-stage is as follows: if the circular motion performed at the current radius does not find a plume signal, the search radius is increased and the circular motion is performed again; when the circular motion finds a plume signal, the process switches to the Surge sub-stage for upward motion.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, which can be executed by one or more processors to implement the steps of the gas source tracing and locating method according to any one of claims 1 to 8.

10. A gas source tracing and positioning system, characterized in that, It includes a platform, a processor mounted on the platform, and a memory for executing instructions. When the processor executes the instructions, it implements the steps of the gas source tracing and locating method according to any one of claims 1 to 8.

Images