Unmanned aerial vehicle-mounted gas detection precision optimization method, system and device

Abstract

The invention belongs to the technical field of gas detection, and particularly discloses an unmanned aerial vehicle-mounted gas detection precision optimization method, system and device, and the method comprises the following steps: obtaining the flight state parameters of an unmanned aerial vehicle and the original concentration reading of an airborne gas sensor in real time; and inputting the flight state parameters into a preset simplified fluid dynamic model, and calculating to obtain an airflow disturbance characteristic quantity at the sensor mounting position. According to the method, the rotor airflow disturbance is accurately depicted by simplifying the fluid dynamic model, dynamic correction is performed in combination with the nonlinear compensation model, the accuracy of gas concentration measurement is remarkably improved in mechanism, the self-adaptive strategy fusing particle swarm optimization and Lyapunov stability is innovatively introduced, and the method has the advantages of being high in robustness and high in reliability. Global parameter optimization can be periodically carried out to minimize long-term errors, and parameters can be finely adjusted in real time based on a stability theory, so that high robustness, rapid convergence and continuous high precision of the system in a complex and changeable flight environment are ensured.

Description

Technical Field

[0001] This application belongs to the field of gas detection technology, and more specifically, relates to a method, system and device for optimizing the accuracy of gas detection on unmanned aerial vehicles. Background Technology

[0002] Confined spaces in power transmission facilities (such as substation interiors, underground cable tunnels, and enclosed compartments of power transmission equipment) have complex environmental structures, turbulent airflow, and pose a risk of accumulating various harmful or flammable gases, placing extremely high demands on operation, maintenance, and safety management. Traditional manual detection methods suffer from low efficiency, high safety risks, and limited coverage. Unmanned aerial vehicles (UAVs), with their flexibility, maneuverability, and ability to carry multiple sensors, are gradually becoming an important means of gas detection in confined spaces. However, when UAVs fly within confined spaces, the strong airflow disturbances generated by their rotors significantly interfere with the gas sensor's measurement process, leading to large deviations in concentration readings. Furthermore, changes in flight attitude and speed further exacerbate measurement uncertainty, severely impacting the reliability and accuracy of the detection data and hindering the effective application of UAVs in gas detection within confined spaces in power transmission facilities.

[0003] Existing UAV-borne gas detection technologies mostly rely on direct sensor readings or simple filtering based on fixed thresholds, lacking systematic modeling and dynamic compensation for the impact mechanism of airflow disturbances. Some studies have attempted to suppress interference by adding physical windbreaks or using mean filtering, but these methods are difficult to adapt to the rapidly changing flight states and complex flow field environments of UAVs in confined spaces. In addition, some compensation methods based on fixed parameter models are not robust enough to different flight conditions and cannot achieve long-term stable accuracy. Existing methods generally suffer from problems such as not fully considering the dynamic coupling relationship between flight state and airflow disturbances, lacking online adaptive adjustment mechanisms, and being unable to ensure both real-time performance and system stability. This results in large gas concentration measurement errors in high-dynamic scenarios such as confined spaces in power transmission plants.

[0004] Therefore, how to solve the problems of large gas concentration measurement errors caused by UAV flight attitude and rotor airflow disturbance in existing UAV-borne gas detection technology, as well as the lack of online adaptive capability and difficulty in guaranteeing system stability of traditional compensation methods, are current research challenges. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this application aims to provide a method, system, and device for optimizing the accuracy of unmanned aerial vehicle (UAV) gas detection. By combining fluid dynamics modeling, nonlinear compensation models, particle swarm optimization, and Lyapunov adaptive strategies, this method effectively solves the problems in existing technologies, such as large gas concentration measurement errors caused by UAV flight attitude and rotor airflow disturbances, and the lack of online adaptive capability and difficulty in ensuring system stability in traditional compensation methods.

[0006] To achieve the above objectives, in a first aspect, this application provides a method for optimizing the accuracy of unmanned aerial vehicle (UAV) gas detection, comprising the following steps: S10, acquires the drone's flight status parameters and raw concentration readings from the onboard gas sensor in real time; S20, input the flight state parameters into a preset simplified fluid dynamics model to calculate the airflow disturbance characteristic quantity at the sensor installation location; S30, input the original concentration reading and airflow disturbance characteristics into the nonlinear compensation model to calculate the corrected gas concentration value; S40, based on the error between the corrected gas concentration value and the reference concentration value, the parameters of the nonlinear compensation model are dynamically adjusted using an adaptive strategy combining particle swarm optimization algorithm and Lyapunov stability theory.

[0007] As a further preferred embodiment, in step S20, the simplified fluid dynamics model is a multi-input multi-output function obtained by simplifying the Navier-Stokes equations and parametrically fitting them using computational fluid dynamics simulations or wind tunnel experimental data; its inputs include flight speed, rotor speed and attitude angle, and its outputs include airflow velocity and turbulence intensity at the sensor location.

[0008] As a further preferred embodiment, in step S30, the functional form of the nonlinear compensation model is:

[0009] in, For the corrected concentration, These are the original concentration readings. It is a vector containing the flight state parameters and airflow disturbance characteristics. and Let be the nonlinear function to be optimized.

[0010] As a further preferred embodiment, in step S40, the particle swarm optimization algorithm is executed periodically, and its optimization objective is to minimize the average measurement error over a period of time, wherein the measurement error is the difference between the corrected concentration and the reference concentration value.

[0011] As a further preferred embodiment, during the execution interval of the particle swarm optimization algorithm, an adaptive law based on Lyapunov stability theory is designed to fine-tune the parameters of the nonlinear compensation model in real time. The adaptive law ensures that the system error asymptotically converges to zero.

[0012] As a further preferred embodiment, the simplified fluid dynamics model and the nonlinear compensation model are pre-trained and parameter initialized using computational fluid dynamics simulation and wind tunnel experimental data during the initial offline phase.

[0013] Secondly, this application provides an unmanned aerial vehicle (UAV) gas detection accuracy optimization system for implementing the method described in any one of the above-mentioned methods, comprising: The flight status perception module is used to acquire flight status parameters; The gas sensor module is used to acquire raw concentration readings; The real-time fluid dynamics calculation module is used to run simplified fluid dynamics models and output airflow disturbance characteristics. The adaptive compensation calculation module is used to run the nonlinear compensation model and execute the adaptive strategy; The data storage and learning module is used to store model parameters and historical data.

[0014] As a further preferred embodiment, the gas sensor module further includes a reference sensor installed in a region affected by the rotor airflow below the influence threshold, for providing the reference concentration value.

[0015] Thirdly, this application provides an unmanned aerial vehicle (UAV) gas detection accuracy optimization device, including a device body, which is integrated into the flight control computer of the UAV or a separate mission computer. The device body includes a processor, a memory, and a computer program stored in the memory. The processor is communicatively connected to the memory, and the processor executes the computer program in the memory to implement the UAV gas detection accuracy optimization method as described in any one of the above.

[0016] As a further preferred embodiment, the main body of the device is communicatively connected to the UAV's flight control system, inertial measurement unit, and gas sensor via a data bus to acquire flight status parameters and raw concentration readings in real time.

[0017] The method, system, and apparatus for optimizing the accuracy of UAV-borne gas detection provided in this application have the following effects: (1) This application accurately characterizes the rotor airflow disturbance by simplifying the fluid dynamics model and performs dynamic correction by combining the nonlinear compensation model, which significantly improves the accuracy of gas concentration measurement from the mechanism perspective. This method innovatively introduces an adaptive strategy that integrates particle swarm optimization and Lyapunov stability, which can periodically perform global parameter optimization to minimize long-term error and fine-tune parameters in real time based on stability theory, ensuring the system's strong robustness, fast convergence and continuous high accuracy in complex and variable flight environments. (2) This application realizes a closed loop of detection, correction and learning through the collaborative work of modules such as flight perception, real-time calculation, adaptive compensation and data storage. It has good scalability and maintainability. The device can be directly integrated into the flight control or mission computer and seamlessly connected to each unit of the UAV through the data bus. It has high hardware integration and low communication latency, which makes it easy to deploy quickly on the embedded platform. It has significant engineering practical value. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method for optimizing the accuracy of UAV-borne gas detection provided in Embodiments 1 and 2 of this application; Figure 2 This is a flowchart of the method for optimizing the accuracy of UAV-borne gas detection provided in embodiments 3 and 4 of this application; Figure 3 This is a flowchart of the method for optimizing the accuracy of UAV-borne gas detection provided in Embodiment 5 of this application; Figure 4 This is a flowchart of the method for optimizing the accuracy of UAV-borne gas detection provided in Embodiment 6 of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0020] like Figure 1 As shown, this application provides a method for optimizing the accuracy of unmanned aerial vehicle (UAV) gas detection, including steps S10 to S40, which are detailed below: Step S10: Real-time acquisition of the UAV's flight status parameters and the raw concentration readings of the onboard gas sensors; through this step, key status information and raw sensor data during the UAV's flight process can be comprehensively acquired, providing the necessary data input for subsequent accurate compensation.

[0021] Step S20: Input the flight state parameters into the preset simplified fluid dynamics model and calculate the airflow disturbance characteristic quantity at the sensor installation location; This step uses a simplified hydrodynamic model to map flight state parameters into specific airflow disturbance characteristics at the sensor location. This model can accurately characterize the impact of the complex airflow generated by the rotor on sensor measurements, providing accurate physical feature inputs for subsequent compensation.

[0022] Step S30: Input the original concentration reading and airflow disturbance characteristics into the nonlinear compensation model to calculate the corrected gas concentration value; This step utilizes a nonlinear compensation model, combining the original readings with airflow disturbance characteristics, to dynamically correct the disturbed measurements. This method can compensate for the measurement bias introduced by airflow disturbances from a mechanistic perspective, significantly improving the accuracy of gas concentration measurements.

[0023] Step S40: Based on the error between the corrected gas concentration value and the reference concentration value, the parameters of the nonlinear compensation model are dynamically adjusted using an adaptive strategy that combines particle swarm optimization algorithm and Lyapunov stability theory. This step utilizes an adaptive strategy that integrates particle swarm optimization (PSO) and Lyapunov stability to dynamically optimize and fine-tune the parameters of the compensation model. The PSO algorithm periodically optimizes global parameters to minimize long-term errors, while the adaptive parameter law designed based on Lyapunov stability theory allows for real-time fine-tuning during optimization intervals, ensuring that the system error asymptotically converges to zero. This combined strategy ensures strong robustness, rapid convergence, and consistently high accuracy under complex and variable flight environments.

[0024] The method for optimizing the accuracy of UAV-borne gas detection provided in this application has the following effects: (1) This application accurately characterizes the rotor airflow disturbance by simplifying the fluid dynamics model and performs dynamic correction by combining the nonlinear compensation model, which significantly improves the accuracy of gas concentration measurement from the mechanism perspective. This method innovatively introduces an adaptive strategy that integrates particle swarm optimization and Lyapunov stability, which can periodically perform global parameter optimization to minimize long-term error and fine-tune parameters in real time based on stability theory, ensuring the system's strong robustness, fast convergence and continuous high accuracy in complex and variable flight environments. (2) This application realizes a closed loop of detection, correction and learning through the collaborative work of modules such as flight perception, real-time calculation, adaptive compensation and data storage. It has good scalability and maintainability. The device can be directly integrated into the flight control or mission computer and seamlessly connected to each unit of the UAV through the data bus. It has high hardware integration and low communication latency, which makes it easy to deploy quickly on the embedded platform. It has significant engineering practical value.

[0025] The following is a specific implementation example of this application: Example 1 Please see Figure 1 A method for optimizing the accuracy of unmanned aerial vehicle (UAV) gas detection includes the following steps: acquiring the flight state parameters of the UAV and the original concentration readings of the airborne gas sensor in real time; inputting the flight state parameters into a preset simplified fluid dynamics model to calculate the airflow disturbance characteristics at the sensor installation location; inputting the original concentration readings and the airflow disturbance characteristics into a nonlinear compensation model to calculate the corrected gas concentration value; and dynamically adjusting the parameters of the nonlinear compensation model based on the error between the corrected gas concentration value and the reference concentration value using an adaptive strategy combining particle swarm optimization algorithm and Lyapunov stability theory.

[0026] By acquiring flight status parameters and gas sensor readings in real time, and combining simplified fluid dynamics and nonlinear compensation models for dynamic correction, the accuracy of gas concentration measurement is significantly improved. Its advantages lie in the use of an adaptive strategy to dynamically adjust model parameters, enabling it to adapt to complex and changing flight environments, enhancing the system's real-time performance and robustness. Furthermore, the method has a simple structure, making it easy to implement and deploy.

[0027] Example 2 Please see Figure 1 A method for optimizing the accuracy of airborne gas detection on an unmanned aerial vehicle (UAV) includes the following steps: acquiring the UAV's flight state parameters and the raw concentration readings of the airborne gas sensor in real time; inputting the flight state parameters into a pre-defined simplified fluid dynamics model, which is a multi-input multi-output function obtained by simplifying the Navier-Stokes equations and parametrically fitting computational fluid dynamics simulations or wind tunnel experimental data; its inputs include flight speed, rotor speed, and attitude angle, and its outputs include airflow velocity and turbulence intensity at the sensor location; in the initial offline phase, the simplified fluid dynamics model and the nonlinear compensation model are pre-trained and parameter-initialized using computational fluid dynamics simulations and wind tunnel experimental data to calculate the airflow disturbance characteristics at the sensor installation location; inputting the raw concentration readings and airflow disturbance characteristics into the nonlinear compensation model, whose functional form is:

[0028] in, For the corrected concentration, These are the original concentration readings. It is a vector containing flight state parameters and airflow disturbance characteristics. and For the nonlinear function to be optimized, the corrected gas concentration value is calculated. Based on the error between the corrected gas concentration value and the reference concentration value, an adaptive strategy combining particle swarm optimization (PSO) and Lyapunov stability theory is used to dynamically adjust the parameters of the nonlinear compensation model. The PSO algorithm is executed periodically, and its optimization objective is to minimize the average measurement error over a period of time. The measurement error is the difference between the corrected concentration value and the reference concentration value. During the execution interval of the PSO algorithm, an adaptive law based on Lyapunov stability theory is designed to fine-tune the parameters of the nonlinear compensation model in real time. The adaptive law ensures that the system error asymptotically converges to zero.

[0029] Building upon Example 1, the construction methods for the simplified fluid dynamics model and the nonlinear compensation model are further clarified. The model is pre-trained based on CFD simulation or wind tunnel experimental data, ensuring the initial accuracy and physical rationality of the model. Its advantages lie in combining particle swarm optimization algorithm and Lyapunov stability theory, periodically optimizing parameters and fine-tuning them in real time. This not only effectively reduces long-term measurement errors but also ensures stable convergence of the system, improving the reliability and adaptability of the method.

[0030] Example 3 Please see Figure 2 A method for optimizing the accuracy of gas detection on an unmanned aerial vehicle (UAV) includes the following steps: acquiring the flight state parameters of the UAV and the raw concentration readings of the onboard gas sensor in real time; inputting the flight state parameters into a preset simplified fluid dynamics model, the function of which is:

[0031] in, : The airflow velocity components (m / s) in the X and Y directions of the body coordinate system at the sensor location. : Turbulence intensity, characterizing the degree of airflow turbulence (dimensionless). Flight speed (m / s); : Average rotor speed (rad / s); Pitch angle and roll angle (rad); : The coefficient matrix obtained through training with CFD data; : An eigenvector containing flight parameters and their polynomial combinations.

[0032] The simplified fluid dynamics model is a multi-input multi-output function obtained by simplifying the Navier-Stokes equations and parametrically fitting computational fluid dynamics simulations or wind tunnel experimental data. Its inputs include flight speed, rotor speed, and attitude angle, while the outputs include airflow velocity and turbulence intensity at the sensor location. In the initial offline phase, the simplified fluid dynamics model and the nonlinear compensation model are pre-trained and parameter-initialized using computational fluid dynamics simulations and wind tunnel experimental data to calculate the airflow disturbance characteristics at the sensor installation location. The original concentration readings and airflow disturbance characteristics are then input into the nonlinear compensation model, whose functional form is:

[0033] in, For the corrected concentration, These are the original concentration readings. It is a vector containing flight state parameters and airflow disturbance characteristics. and For the nonlinear function to be optimized, the corrected gas concentration value is calculated. Based on the error between the corrected gas concentration value and the reference concentration value, an adaptive strategy combining particle swarm optimization (PSO) and Lyapunov stability theory is used to dynamically adjust the parameters of the nonlinear compensation model. The PSO algorithm is executed periodically, and its optimization objective is to minimize the average measurement error over a period of time. The measurement error is the difference between the corrected concentration value and the reference concentration value. During the execution interval of the PSO algorithm, an adaptive law based on Lyapunov stability theory is designed to fine-tune the parameters of the nonlinear compensation model in real time. The adaptive law ensures that the system error asymptotically converges to zero.

[0034] Define error: Design a Lyapunov function with the following form:

[0035] in, For parameter deviation vectors, A positive adaptive gain; to ensure (System stable), derive parameters The adaptive law is:

[0036] in: Compensation parameters Real-time adjustment rate; : The sensitivity (gradient) of the corrected concentration to the compensation parameter; this formula means that the system will automatically and stably adjust its own parameters to eliminate the error based on the magnitude and direction of the error.

[0037] A simplified fluid dynamics model and a nonlinear compensation model are defined using specific mathematical formulas. Lyapunov stability theory is introduced to derive the parameter adaptive law, ensuring that the system error asymptotically converges to zero. Its advantages lie in its rigorous theory, clear model expression, ease of simulation and practical application, and real-time response to airflow disturbances, significantly improving the accuracy and stability of gas detection. It is suitable for high-dynamic flight scenarios.

[0038] Example 4 Please see Figure 2 A method for optimizing the accuracy of gas detection on unmanned aerial vehicles (UAVs) includes the following steps: Real-time acquisition of UAV flight status parameters and raw concentration readings from onboard gas sensors. ; The flight state parameters are input into a pre-defined simplified fluid dynamics model to calculate the airflow disturbance characteristics at the sensor installation location. Original concentration readings By inputting the airflow disturbance characteristic quantity into a nonlinear compensation model, the corrected gas concentration value is calculated. The nonlinear compensation model satisfies the following functional relationship:

[0039] in, It is a feature vector containing flight state parameters and airflow disturbance characteristics. and For about vectors Nonlinear compensation function, eigenvector Specifically represented as ; Based on the corrected gas concentration value Compared with reference concentration value error An adaptive strategy combining particle swarm optimization and Lyapunov stability theory is used to dynamically adjust the nonlinear compensation function. and The parameters.

[0040] The simplified fluid dynamics model has the following functional form:

[0041] in, : The airflow velocity components (m / s) in the X and Y directions of the body coordinate system at the sensor location. : Turbulence intensity, characterizing the degree of airflow turbulence (dimensionless). Flight speed (m / s); : Average rotor speed (rad / s); Pitch angle and roll angle (rad); : The coefficient matrix obtained through training with CFD data; : An eigenvector containing flight parameters and their polynomial combinations.

[0042] The particle swarm optimization algorithm is executed periodically, and its optimization objective is to minimize the fitness function. ,in For the first The error from each sampling is used to search for the optimal combination of parameters that minimizes J by iteratively updating the position and velocity of the particles.

[0043] The adaptive strategy based on Lyapunov stability theory derives its parameter adjustment law based on the following steps: Define the Lyapunov function:

[0044] in Positive adaptive gain The vector representing the deviation between the parameters of the compensation function and their ideal values; to satisfy the Lyapunov stability condition. Derive the parameters of the compensation function. The adaptive adjustment law is:

[0045] in, For the corrected concentration versus parameter Sensitivity.

[0046] Parameter adjustment law Discretized into Used to measure parameters in discrete-time systems Perform real-time fine-tuning.

[0047] The composition of eigenvectors and the discretization implementation of the adaptive strategy are further refined, making parameter adjustment more suitable for digital control systems. Its advantages lie in its clear optimization objective (minimizing the sum of squared errors), the efficient search for the global optimum using the particle swarm optimization algorithm, and the Lyapunov adaptive law for fine-tuning parameters in discrete-time systems, thus improving the method's practicality and real-time performance, making it suitable for embedded platform deployment.

[0048] Example 5 Please see Figure 3 An unmanned aerial vehicle (UAV) gas detection accuracy optimization system includes: a flight state perception module for acquiring flight state parameters; a gas sensor module for acquiring raw concentration readings, the gas sensor module also including a reference sensor installed at a location less affected by rotor airflow for providing reference concentration values; a real-time fluid dynamics calculation module for running a simplified fluid dynamics model and outputting airflow disturbance characteristics; an adaptive compensation calculation module for running a nonlinear compensation model and executing an adaptive strategy; and a data storage and learning module for storing model parameters and historical data.

[0049] The modular design facilitates system expansion and maintenance, the reference sensor provides a reliable baseline value, the data storage supports historical data learning and model iteration optimization, and the overall system works collaboratively to significantly improve the accuracy of gas detection and environmental adaptability.

[0050] Example 6 Please see Figure 4A device for optimizing the accuracy of gas detection on a UAV includes a main body, which is integrated into the flight control computer of the UAV or a separate mission computer. The main body includes a processor, a memory, and a computer program stored in the memory. The processor is communicatively connected to the memory. The processor executes the computer program in the memory to implement the aforementioned accuracy optimization method. The main body is communicatively connected to the flight control system, inertial measurement unit, and gas sensor of the UAV via a data bus to acquire flight status parameters and raw concentration readings in real time.

[0051] With high hardware integration, it seamlessly connects with the flight control system, IMU and gas sensors via a data bus, enabling low-latency real-time processing. This facilitates rapid deployment and application on actual UAV platforms and has high engineering practical value.

[0052] Example 7 Precise hovering and local concentration mapping for routine inspection of cable tunnels System and technology adaptation: Flight platform: Small hexacopter UAV with LiDAR SLAM (Simultaneous Localization and Mapping) capabilities to adapt to tunnel environments without GPS; Sensors: Equipped with a laser methane detector and an electrochemical hydrogen sulfide sensor; Core challenges: The narrow space inside the tunnel and the complex vortex formed by the rotor downwash airflow hitting the walls and cables cause the sensor readings to fluctuate violently, making it difficult to determine the true concentration.

[0053] Implementation method: Targeted airflow modeling: Simplifying the input of the fluid dynamics model to include not only flight speed and attitude angle More importantly, it introduces the shortest distance to the sidewalls / top walls. The model is trained by simulating the flow field at different locations within the tunnel using CFD, and its output... (turbulence intensity) and Strong correlation.

[0054] Near-wall compensation algorithm: The nonlinear compensation model is modified as follows:

[0055] Among them, the function The dilution effect of near-wall extreme turbulence on readings was specifically modeled. When When smaller, It will be significantly greater than 1 to compensate for the low reading caused by the airflow being dispersed.

[0056] "Precise Hover" Adaptive Mode: The drone performs automatic hovering measurements at suspected leak points (such as near valves or joints). The optimization objective of the PSO algorithm is to obtain a stable concentration reading (minimizing variance) within a 10-second hover, rather than the absolute value. The system automatically adjusts parameters to quickly converge the output curve to a stable plateau, which represents the reliable local concentration. A Lyapunov stabilizer ensures smooth, non-oscillating readings during this process.

[0057] In summary, this system integrates airborne lidar and a compensation calculation unit. It enables drones to not only fly safely within tunnels but also achieve "accurate measurements upon landing," effectively distinguishing between concentration fluctuations caused by flight disturbances and actual hazardous gas accumulation, providing maintenance personnel with precise risk location maps.

[0058] Example 8 Rapid Leak Source Tracing System for Emergency Detection in Utility Tunnels System Composition: Flight Platform: A quadcopter UAV with a high-strength, fully enclosed anti-collision cage design and explosion-proof certification. It integrates a multi-hop self-organizing network communication module to ensure a stable data link with the entrance base station within the long and deep utility tunnel; Sensing Unit: Equipped with a wide-range VOCs (volatile organic compounds) sensor with extremely fast response, and optionally a miniature ultrasonic anemometer to sense the actual background airflow within the utility tunnel; Computing Unit: The onboard computer has sufficient computing power to process gas data, airflow information, and its own position and attitude in real time, and to run complex source tracing algorithms.

[0059] Implementation Methods: Dynamic Airflow Field Fusion: The system not only relies on flight state-based predictive models but also integrates actual environmental airflow data measured by an onboard miniature anemometer. This fusion perception strategy enables the system to more accurately distinguish between the "wind" generated by the UAV's own movement and the actual background ventilation within the pipe gallery, providing reliable directional guidance for headwind tracking; Motion Interference Decoupling and Compensation: One of the core functions of the compensation algorithm is to decouple the dynamic impact of UAV flight motion on gas readings from the true gradient of environmental concentration. By correcting the readings in real time, the system can more purely reflect the concentration changes caused by "spatial location changes," thus highlighting the true concentration distribution gradient; Intelligent Source Tracing Guidance: The adaptive algorithm is deeply coupled with the flight control system. By continuously analyzing the compensated concentration data and airflow direction, the algorithm dynamically evaluates the effectiveness of the current flight path and provides the flight control system with flight direction suggestions that are "most likely pointing to the leak source." The system can guide the UAV to autonomously execute headwind search or concentration gradient climbing strategies, significantly shortening the leak source location time.

[0060] In summary, within the low-visibility, complex, and dangerous environment of the utility tunnel, it not only resisted airflow interference to conduct accurate measurements, but also proactively and intelligently went against the flow to pinpoint the source of the leak, providing crucial intelligence for subsequent emergency response and greatly ensuring personnel safety and controlling the situation.

[0061] Example 9 Inspection and gas monitoring of power transmission tunnels System Composition: Platform: A medium-sized hexacopter UAV with IP67 dust and water resistance to adapt to the humid environment and potential condensation within the culvert. The airframe employs a low electromagnetic interference design to minimize potential impact on power transmission cables; Sensing Unit: Integrates multiple gas detection modules (catalytic combustion CH4 sensor, infrared CO2 sensor, electrochemical O2 sensor), lidar (for precise positioning and terrain mapping in low-light environments), a high-precision IMU (inertial measurement unit), and a wide-angle vision camera; Computing Unit: Equipped with an onboard computer with edge computing capabilities, capable of simultaneously processing gas data, 3D point cloud information, and visual data, and running complex environmental perception and compensation algorithms in real time; Communication Unit: Employs multi-mode communication links, utilizing pre-deployed communication repeaters or cables for transmission deep within the culvert to ensure the continuity of data and control signals.

[0062] Implementation Method: Terrain-Assisted Airflow Modeling and Compensation: The system deeply integrates real-time 3D point cloud data generated by LiDAR with flight attitude and velocity information. The built-in fluid dynamics model not only considers the UAV's own motion but, more importantly, incorporates its relative positional relationships with culvert walls, cable trays, and puddles. When the UAV approaches curved ceilings or narrow passages, the model can predict the "Venturi effect" or vortex generation zone caused by spatial contraction and provide predictive compensation for sensor readings. For example, when traversing narrow areas, it automatically enhances the correction for the dilution effect of readings caused by airflow acceleration, ensuring that the data accurately reflects the actual gas enrichment at that location; Autonomous Identification of Hazardous Areas Through Multimodal Data Fusion: The system binds precisely compensated gas concentration data with 3D spatial location. Algorithms automatically identify "ventilation dead zones" (such as the ends of culvert branches and recessed areas) and "high-risk accumulation zones" (such as spatial units where methane concentrations consistently exceed safety thresholds). Simultaneously, visual information is integrated to record images of key equipment such as cable joints and insulators and perform correlation analysis with gas anomalies. Once a location is found to simultaneously exhibit both visual anomalies (such as water stains or corrosion) and abnormal gas readings, the system will immediately mark that location as the highest priority and issue a detailed alarm. Adaptive inspection paths and long-term trend analysis: the drone does not simply repeat fixed routes. Based on a "gas environment map" constructed from historical inspection data inside the culvert, the system can dynamically adjust the focus areas and sampling density for each flight. For areas where potential hazards have been previously detected or where gas concentrations are showing an upward trend, the system will automatically increase hovering detection time and flight coverage density. All compensated and calibrated data is recorded long-term, forming a "culvert gas safety and health record" for analyzing gas sources and diffusion patterns.

[0063] In summary, this study overcame numerous challenges, including the complex internal structure of culverts, the unique electromagnetic environment, and significant airflow disturbances. It achieved precise, spatial, and intelligent monitoring of hazardous gases, providing a solid data foundation and technical support for ensuring the safe and stable operation and predictive maintenance of high-voltage transmission lines.

[0064] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for optimizing the accuracy of gas detection on unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: S10, acquires the drone's flight status parameters and raw concentration readings from the onboard gas sensor in real time; S20, input the flight state parameters into a preset simplified fluid dynamics model to calculate the airflow disturbance characteristic quantity at the sensor installation location; S30, input the original concentration reading and airflow disturbance characteristics into the nonlinear compensation model to calculate the corrected gas concentration value; S40, based on the error between the corrected gas concentration value and the reference concentration value, the parameters of the nonlinear compensation model are dynamically adjusted using an adaptive strategy combining particle swarm optimization algorithm and Lyapunov stability theory.

2. The method for optimizing the accuracy of UAV-borne gas detection as described in claim 1, characterized in that, In step S20, the simplified fluid dynamics model is a multi-input multi-output function obtained by simplifying the Navier-Stokes equations and parametric fitting using computational fluid dynamics simulations or wind tunnel experimental data; its inputs include flight speed, rotor speed and attitude angle, and its outputs include airflow velocity and turbulence intensity at the sensor location.

3. The method for optimizing the accuracy of UAV-borne gas detection as described in claim 1, characterized in that, In step S30, the functional form of the nonlinear compensation model is: in, For the corrected concentration, These are the original concentration readings. It is a vector containing the flight state parameters and airflow disturbance characteristics. and Let be the nonlinear function to be optimized.

4. The method for optimizing the accuracy of UAV-borne gas detection as described in claim 1, characterized in that, In step S40, the particle swarm optimization algorithm is executed periodically, and its optimization objective is to minimize the average measurement error over a period of time, wherein the measurement error is the difference between the corrected concentration and the reference concentration value.

5. The method for optimizing the accuracy of UAV-borne gas detection as described in claim 4, characterized in that, During the execution interval of the particle swarm optimization algorithm, an adaptive law based on Lyapunov stability theory is designed to fine-tune the parameters of the nonlinear compensation model in real time. The adaptive law ensures that the system error asymptotically converges to zero.

6. The method for optimizing the accuracy of UAV-borne gas detection as described in claim 1, characterized in that, In the initial offline phase, the simplified fluid dynamics model and the nonlinear compensation model are pre-trained and parameter initialized using computational fluid dynamics simulations and wind tunnel experimental data.

7. A UAV-borne gas detection accuracy optimization system for implementing the method of any one of claims 1 to 6, characterized in that, include: The flight status perception module is used to acquire flight status parameters; The gas sensor module is used to acquire raw concentration readings; The real-time fluid dynamics calculation module is used to run simplified fluid dynamics models and output airflow disturbance characteristics. The adaptive compensation calculation module is used to run the nonlinear compensation model and execute the adaptive strategy; The data storage and learning module is used to store model parameters and historical data.

8. The UAV-borne gas detection accuracy optimization system as described in claim 7, characterized in that, The gas sensor module also includes a reference sensor installed in a region affected by the rotor airflow below the influence threshold, for providing the reference concentration value.

9. A device for optimizing the accuracy of gas detection on a UAV, characterized in that, The device includes a main body, which is integrated into the flight control computer of the UAV or a separate mission computer. The main body includes a processor, a memory, and a computer program stored in the memory. The processor is communicatively connected to the memory, and the processor executes the computer program in the memory to implement the UAV-borne gas detection accuracy optimization method as described in any one of claims 1 to 6.

10. The UAV-borne gas detection accuracy optimization device as described in claim 9, characterized in that, The main body of the device is connected to the UAV's flight control system, inertial measurement unit, and gas sensor via a data bus to acquire flight status parameters and raw concentration readings in real time.

Images