An artificial intelligence inspection system and method based on drones

Abstract

The present invention discloses an unmanned aerial vehicle (UAV) artificial intelligence inspection system and method, which specifically relates to the field of UAV artificial intelligence technology. By acquiring magnetic field interference parameters and terrain complexity parameters in the pipeline inspection path in real time, and extracting magnetic field intensity fluctuation characteristics and obstacle density abnormal change characteristics, the navigation and positioning system accuracy weight assignment of each path is calculated in combination with machine learning methods such as polynomial regression to generate a comprehensive inspection accuracy index. By comparing the comprehensive accuracy index with the gradient standard threshold, the accuracy division of the inspection path is achieved: for incomplete accuracy inspections, the accuracy abnormality of the future navigation system is predicted, and the flight altitude is adjusted when necessary to avoid areas with dense obstacles. This effectively solves the problems of false detection and missed detection caused by magnetic field interference and complex terrain in the existing technology, significantly improves the efficiency, accuracy and reliability of UAV inspection tasks, and avoids major disasters caused by failure to detect leaks in a timely manner.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle (UAV) artificial intelligence technology, and in particular to an unmanned aerial vehicle (UAV) artificial intelligence inspection system and method. Background Art

[0002] Drone AI inspections are an advanced method that uses artificial intelligence (AI) technology combined with drones to automate the inspection of specific areas, equipment, or facilities. Equipped with high-resolution cameras, infrared sensors, or other detection equipment, drones are able to autonomously fly and collect data in complex environments. AI plays a key role in data analysis, processing real-time information through methods such as image recognition, pattern detection, and anomaly analysis to quickly identify faults or potential hazards. This technology is widely used in infrastructure monitoring (such as power grids, bridges, and railways), agriculture, energy, and security. For example, in the oil and gas industry, drone AI inspections can be used to detect pipeline leaks. AI uses thermal imaging to identify temperature anomalies and quickly locate the leak. These technologies significantly improve inspection efficiency and reduce labor costs and the need for dangerous work.

[0003] The existing technology has the following shortcomings:

[0004] Drone navigation systems can misalign their positioning due to magnetic interference and complex terrain, misidentifying healthy pipeline inspections as leaks while missing the actual leaking section. If this error goes undetected during inspection data analysis, the actual leak could worsen, eventually leading to a large-scale disaster. Furthermore, frequent navigation system errors may necessitate frequent interruptions and rescheduling of inspection missions, significantly reducing mission efficiency. Summary of the Invention

[0005] The purpose of the present invention is to provide a drone-based artificial intelligence inspection system and method to address the shortcomings of the background technology.

[0006] In order to achieve the above objectives, the present invention provides the following technical solutions: a drone-based artificial intelligence inspection method, comprising the following steps:

[0007] S1: Using drone-mounted sensors to obtain real-time magnetic field interference parameters and terrain complexity parameters along different pipeline inspection routes. The magnetic field interference parameters include ambient magnetic field intensity data, and the terrain complexity parameters include obstacle density data.

[0008] S2: Preprocessing the acquired environmental magnetic field strength data and obstacle density data, and extracting magnetic field strength fluctuation characteristics from the environmental magnetic field strength data and abnormal obstacle density change characteristics from the obstacle density data;

[0009] S3: Based on the extracted magnetic field intensity fluctuation characteristics and obstacle density abnormal change characteristics, the accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path is determined, and the weighted average summation calculation is performed to obtain the comprehensive inspection accuracy index;

[0010] S4: Compare the comprehensive inspection accuracy index in the pipeline inspection path with the gradient standard threshold, and classify the accuracy of the UAV pipeline inspection path according to the comparison result, and the classification results include accuracy inspection, incomplete accuracy inspection and inaccuracy inspection;

[0011] S5: For accurate inspections, the inspection continues along the established route without adjusting the flight strategy. For inaccurate inspections, the inspection mission of the drone is immediately interrupted, and the optimal inspection route is recalculated, with priority given to covering the abnormal pipeline section.

[0012] S6: For incomplete accuracy inspections, the accuracy anomaly of the drone’s navigation and positioning system in a subsequent fixed time period is predicted. If the accuracy anomaly is high, the drone’s inspection path is replanned, and the drone’s flight altitude is increased and areas with dense obstacles are avoided.

[0013] Preferably, in S2, the magnetic field intensity fluctuation characteristics in the extracted environmental magnetic field intensity data are analyzed to generate a magnetic field intensity continuous fluctuation index, and the method for obtaining the magnetic field intensity continuous fluctuation index is:

[0014] Record the output of the magnetometer sensor to obtain the time series data B(s) of the magnetic field intensity, where s is the time window. Divide the time series B(s) into short time windows w(n), each with a length of N samples. Apply Fourier transform to the signal in each time window and calculate the frequency component: ;in, is the complex spectrum value at time s and frequency f, B(n) is the signal sample in the window, w(n) is the window function value, is the frequency weight in complex form; take the modulus of the complex spectrum to get the amplitude spectrum , the expression is: ; For each time window s, calculate the spectral energy density , the expression is: ;in, and are the lowest and highest frequencies of interest, respectively, and E(s) represents the spectral energy density of the time window s; E(s) is normalized to: ;in, and are the minimum and maximum spectral energy in the whole sequence, is the normalized spectrum energy density; the normalized spectrum energy density is smoothed to generate the magnetic field intensity continuous fluctuation index HSK, which is expressed as: ; Where Z is the smoothing factor, HSK is the continuous fluctuation index of magnetic field intensity, for The continuous fluctuation index of magnetic field intensity at the moment.

[0015] Preferably, in S2, after analyzing the abnormal change characteristics of obstacle density in the extracted obstacle density data, an abnormal change index of obstacle density is generated. The method for obtaining the abnormal change index of obstacle density is:

[0016] Obstacle point cloud data is collected through lidar to form a point set ; Each point Contains its location coordinates , using Euclidean distance to calculate obstacle points and Distance between: ; For each point , find its k-nearest neighbor set , where k is the number of nearest neighbors specified by the user; k-nearest neighbor distance Yes The distance to its kth nearest neighbor: ;definition To its neighboring points The reachable distance , the expression is: ;

[0017] point The local reachable density is all points in its neighborhood The reciprocal mean of the reachable distance is expressed as: Where, yes The number of neighboring points of The local outlier factor is the ratio of the average local reachability density of the points in its neighborhood to its own local reachability density: ; Normalize the local outlier factor values of all points to generate the obstacle density abnormal change index, which is expressed as: ; Where GSH is the obstacle density abnormal change index, and are the minimum and maximum local outlier factor values in the dataset, respectively.

[0018] Preferably, in S3, based on the extracted magnetic field intensity fluctuation characteristics and obstacle density abnormal change characteristics, the accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path is determined, and the weighted average summation calculation is performed to obtain the comprehensive inspection accuracy index, which is specifically:

[0019] The continuous fluctuation index of magnetic field intensity and the abnormal change index of obstacle density are converted into comprehensive feature vectors, which are used as inputs of the machine learning model. The machine learning model uses each set of comprehensive feature vectors to predict the accuracy weight assignment label of the UAV navigation and positioning system in each pipeline inspection path as the prediction target, and takes minimizing the sum of the prediction errors of the accuracy weight assignment labels of the UAV navigation and positioning systems in all pipeline inspection paths as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path is determined according to the model output results. Among them, the machine learning model is a polynomial regression model, and the comprehensive inspection accuracy index is obtained by weighted average summation calculation of the accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path.

[0020] Preferably, in S4, the comprehensive inspection accuracy index in the pipeline inspection path is compared with the gradient standard threshold, and the accuracy of the UAV pipeline inspection path is divided according to the comparison result, and the division results include accuracy inspection, incomplete accuracy inspection and inaccuracy inspection;

[0021] Comparing the obtained comprehensive inspection accuracy index of the pipeline inspection path with a gradient standard threshold, the gradient standard threshold including a first standard threshold and a second standard threshold, wherein the first standard threshold is less than the second standard threshold, and comparing the comprehensive inspection accuracy index of the pipeline inspection path with the first standard threshold and the second standard threshold respectively;

[0022] If the comprehensive inspection accuracy index in the pipeline inspection path is greater than the second standard threshold, it means that the accuracy of the UAV navigation and positioning system in the pipeline inspection path is high. At this time, a high-accuracy inspection signal is generated, and the accuracy of the UAV pipeline inspection path is classified as an accuracy inspection;

[0023] If the comprehensive inspection accuracy index in the pipeline inspection path is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the accuracy of the UAV navigation and positioning system in the pipeline inspection path is medium. At this time, a medium accuracy inspection signal is generated, and the accuracy of the UAV pipeline inspection path is classified as incomplete accuracy inspection;

[0024] If the comprehensive inspection accuracy index in the pipeline inspection path is less than the first standard threshold, it means that the accuracy of the UAV navigation and positioning system in the pipeline inspection path is low. At this time, a low-accuracy inspection signal is generated, and the accuracy of the UAV pipeline inspection path is classified as inaccurate inspection.

[0025] Preferably, in S6, for incomplete accuracy inspections, the accuracy abnormality of the drone navigation and positioning system in a subsequent fixed time period is predicted. If the accuracy abnormality is high, the drone inspection path is replanned, and the drone's flight altitude is increased and areas with dense obstacles are avoided. Specifically,

[0026] For incomplete accuracy inspection, that is, the comprehensive inspection accuracy index in the generated pipeline inspection path is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, in the future T time period, the accuracy abnormality of the navigation and positioning system is determined by the comprehensive prediction index express: ;in: is the predicted value of the magnetic field intensity fluctuation index in the future time t, GSH(t) is the predicted value of the obstacle density abnormal change index in the future time t, CI(t) is the comprehensive inspection accuracy index at the current time t, α, β, γ are weight coefficients, indicating the proportion of the influence of magnetic field fluctuation, obstacle density and current comprehensive accuracy on the predicted value, satisfying α+β+γ=1; the time series prediction model is used to make short-term predictions for HSK(t) and GSH(t): , ; Define the threshold Pthreshold of the abnormal degree of prediction accuracy: if , indicating that the accuracy is abnormally high and the path needs to be replanned; if , continue to inspect along the existing route.

[0027] Preferably, the path replanning goal is to avoid the obstacle-dense area with the shortest path and optimize the navigation accuracy. The objective function of the path planning problem is , the expression is: Where, is the length of path segment i, is the predicted abnormality level of path segment i, k is the penalty coefficient used to increase the path weight of high abnormality areas and avoid high-risk areas, and W is the total number of path segments;

[0028] The flight altitude adjustment rule is: define the adjusted flight altitude , the expression is: Where, is the current flight altitude, ΔH is the altitude increment coefficient; mark the high obstacle density area and set it as an impassable area during path planning. The regional density weight formula is: Where, is the regional weight, GSH(j) is the abnormal change index of obstacle density in the region, and when replanning the path, the regions whose regional weight is greater than the regional weight threshold preset according to historical data are excluded.

[0029] The present invention also provides an artificial intelligence inspection system based on drones, including a data acquisition module, a feature extraction module, a navigation and positioning accuracy assessment module, an accuracy classification module, a path planning and task control module, and a dynamic prediction and strategy optimization module;

[0030] Data acquisition module: uses drone-mounted sensors to acquire magnetic field interference parameters and terrain complexity parameters in different pipeline inspection paths in real time. The magnetic field interference parameters include ambient magnetic field intensity data, and the terrain complexity parameters include obstacle density data.

[0031] Feature extraction module: pre-processes the acquired environmental magnetic field strength data and obstacle density data, and extracts the magnetic field intensity fluctuation characteristics in the environmental magnetic field strength data and the obstacle density abnormal change characteristics in the obstacle density data;

[0032] Navigation and positioning accuracy assessment module: Based on the extracted magnetic field intensity fluctuation characteristics and abnormal obstacle density change characteristics, the module determines the accuracy weight of the drone navigation and positioning system in each pipeline inspection path, and calculates the weighted average sum of these values to obtain the comprehensive inspection accuracy index;

[0033] Accuracy classification module: compares the comprehensive inspection accuracy index in the pipeline inspection path with the gradient standard threshold, and classifies the accuracy of the UAV pipeline inspection path according to the comparison results. The classification results include accurate inspection, incomplete accuracy inspection and inaccuracy inspection;

[0034] Path planning and mission control module: For accurate inspections, the inspection continues along the established path without adjusting the flight strategy; for inaccurate inspections, the drone's inspection mission is immediately interrupted, the optimal inspection path is recalculated, and the abnormal pipeline section is covered first;

[0035] Dynamic prediction and strategy optimization module: For incomplete accuracy inspections, the accuracy anomaly of the drone's navigation and positioning system in a subsequent fixed time period is predicted. If the accuracy anomaly is high, the drone's inspection path is replanned, the drone's flight altitude is increased, and areas with dense obstacles are avoided.

[0036] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0037] 1. The present invention uses real-time data collection, feature extraction, and comprehensive analysis based on drone-mounted sensors. It innovatively combines the magnetic field intensity fluctuation index with the obstacle density abnormal change index to propose a weighted assessment method for the accuracy of the navigation and positioning system, and accurately calculates the comprehensive inspection accuracy index. By comparing with the gradient standard threshold, the accuracy level of the inspection path can be efficiently divided, and corresponding inspection strategies can be adopted for different situations to achieve dynamic adjustment of inspection tasks. Especially in the scenario of incomplete accuracy inspection, the degree of abnormality of the future navigation system can be assessed in advance through the time series prediction model. Combined with path replanning and flight altitude optimization, it effectively avoids the waste of inspection resources and potential risks caused by false detection or missed detection.

[0038] 2. The present invention significantly improves the reliability, accuracy, and adaptability of UAV pipeline inspection tasks. In complex inspection environments, it can quickly identify and avoid high-risk areas, prioritize coverage of important pipeline sections, and ensure the comprehensiveness and real-time nature of inspections. Through path replanning and flight altitude adjustment methods, the risk of navigation errors caused by dense obstacles or magnetic field interference is minimized, while improving the efficiency of inspection tasks and the quality of data collection. The present invention effectively solves the problems of disaster risks and low efficiency caused by positioning errors in the existing technology, and provides strong technical support for UAV intelligent inspections in complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments described in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0040] Figure 1 Flow chart of the method of the present invention.

[0041] Figure 2 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0043] Example 1, please refer to Figure 1 and Figure 2As shown, the drone-based artificial intelligence inspection method described in this embodiment includes the following steps:

[0044] S1: Using drone-mounted sensors to obtain real-time magnetic field interference parameters and terrain complexity parameters along different pipeline inspection routes. The magnetic field interference parameters include ambient magnetic field intensity data, and the terrain complexity parameters include obstacle density data.

[0045] S2: Preprocessing the acquired environmental magnetic field strength data and obstacle density data, and extracting magnetic field strength fluctuation characteristics from the environmental magnetic field strength data and abnormal obstacle density change characteristics from the obstacle density data;

[0046] S3: Based on the extracted magnetic field intensity fluctuation characteristics and obstacle density abnormal change characteristics, the accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path is determined, and the weighted average summation calculation is performed to obtain the comprehensive inspection accuracy index;

[0047] S4: Compare the comprehensive inspection accuracy index in the pipeline inspection path with the gradient standard threshold, and classify the accuracy of the UAV pipeline inspection path according to the comparison result, and the classification results include accuracy inspection, incomplete accuracy inspection and inaccuracy inspection;

[0048] S5: For accurate inspections, the inspection continues along the established route without adjusting the flight strategy. For inaccurate inspections, the inspection mission of the drone is immediately interrupted, and the optimal inspection route is recalculated, with priority given to covering the abnormal pipeline section.

[0049] S6: For incomplete accuracy inspections, the accuracy anomaly of the drone’s navigation and positioning system in a subsequent fixed time period is predicted. If the accuracy anomaly is high, the drone’s inspection path is replanned, and the drone’s flight altitude is increased and areas with dense obstacles are avoided.

[0050] In S1, the magnetic field interference parameters and terrain complexity parameters in different pipeline inspection paths are obtained in real time through the drone-mounted sensors. The magnetic field interference parameters include the environmental magnetic field intensity data, and the terrain complexity parameters include the obstacle density data, specifically:

[0051] A highly sensitive three-axis magnetometer monitors the ambient magnetic field strength and its variations in real time. Integrated into the drone's navigation system, it compensates for the effects of magnetic field anomalies on navigation. The magnetometer measures the ambient magnetic field strength (in Gauss or Tesla) along the drone's flight path. Real-time magnetic field data is recorded and analyzed for significant magnetic interference sources (such as high-voltage power lines and ferromagnetic materials).

[0052] During drone flight, the system samples ambient magnetic field strength data every second and annotates its spatial coordinates (GPS location). Fluctuation characteristics of the magnetic field strength data, such as instantaneous intensity changes, interference frequency, and spatial distribution patterns, are extracted. Abnormal areas where magnetic field strength exceeds background noise are identified. High-magnetic-strength areas are associated with inspection routes, providing a basis for subsequent navigation route optimization.

[0053] High-resolution 3D LiDAR scans terrain features and constructs an obstacle distribution model. It detects vegetation, water bodies, and other physical obstacles in the terrain, supplementing features in areas LiDAR cannot detect. LiDAR uses the reflection time and intensity of pulsed lasers to obtain 3D obstacle distribution data. Multispectral cameras analyze spectral reflectance differences to distinguish terrain features such as buildings and vegetation.

[0054] LiDAR continuously scans along the drone's flight path, generating real-time point cloud data that records the height, density, and location of obstacles. A multispectral camera captures spectral images of the terrain surface and extracts object categories (such as trees and rocks). The number and density of obstacles within a unit path are calculated, and features of areas with high obstacle density are extracted. Areas with high obstacle density within the path are annotated and assigned a feasibility level (low, medium, or high).

[0055] A 3D terrain model is constructed by combining point cloud data and spectral images to evaluate terrain undulations, obstacle distribution, and inspection path passability.

[0056] Magnetic field interference parameters and terrain complexity parameters are integrated into the navigation system's data stream, forming a multi-dimensional parameter set for path assessment. The path analysis module evaluates the flight area segment by segment, assigning navigation accuracy weights to each segment. Filtering algorithms (such as Kalman filtering) remove sensor noise to ensure data accuracy. Machine learning models classify and analyze parameter features to predict the potential impact of interference areas on navigation.

[0057] Based on the acquired parameters, the drone's inspection path is dynamically adjusted to prioritize avoiding areas of magnetic interference and dense obstacles. High-interference or high-complexity areas are marked to provide data support for subsequent mission planning. The acquired parameter data is stored in the cloud for inspection report generation and system model training and optimization.

[0058] S2: Preprocessing the acquired environmental magnetic field strength data and obstacle density data, and extracting magnetic field strength fluctuation characteristics from the environmental magnetic field strength data and abnormal obstacle density change characteristics from the obstacle density data.

[0059] Filter the raw data to remove random noise from the sampling process. Common methods include: Low-pass filtering: Removes high-frequency noise, such as rapidly changing, small disturbances. Combines historical and current state predictions to smooth the curves of magnetic field and obstacle data. Identifies and processes extreme values that may be caused by sensor errors, using median filtering or statistical methods (such as the triple standard deviation method) to eliminate outliers.

[0060] Calibrate sensor readings to ensure accurate measurements: Magnetic field data calibration: Compensate for zero drift or tilt errors in the magnetometer and calibrate the background magnetic field based on the drone's initial static position. Through multiple measurements and comparisons, correct for temporal deviation and spatial alignment of lidar and multispectral data. Normalize ambient magnetic field intensity data and obstacle density data. Normalization helps eliminate dimensional differences and facilitates subsequent algorithm analysis.

[0061] Extracting magnetic field intensity fluctuation characteristics includes: Fluctuation analysis: Calculating the first-order derivative: Analyzing the rate of change of magnetic field intensity over time or space, and capturing points of significant fluctuation: ΔB = B(t+1) − Bt; where Bt is the magnetic field intensity value at time t. A larger ΔB indicates a more severe fluctuation.

[0062] Fourier Transform (FFT): Converts magnetic field strength data from the time domain to the frequency domain, identifying the dominant frequency and amplitude characteristics of the fluctuations and highlighting possible interference source patterns.

[0063] Fluctuation amplitude statistics: Calculate the mean, variance, extreme value range and other indicators of the magnetic field strength to quantify the fluctuation intensity.

[0064] Extracting abnormal obstacle density change features includes:

[0065] Local density change analysis: Use sliding window technology to calculate the obstacle density in different spatial segments and identify abnormal changes in density: Dwindow = number of obstacles / window area; analyze the change rate of the density value of each sliding window.

[0066] Cluster analysis: Use the K-means or DBSCAN algorithm to cluster the obstacle distribution and locate dense obstacle areas. Abnormal changes are usually manifested as the density value of a certain area deviating significantly from the mean of the adjacent areas.

[0067] Terrain complexity index: A statistical index that comprehensively calculates the height, density, and distribution of obstacles, such as the standard deviation of the obstacle spacing distribution, to assess terrain complexity.

[0068] Set thresholds for magnetic field fluctuations and obstacle density changes, and mark areas exceeding the thresholds as "abnormal." Compare the currently extracted features with historical inspection data to identify possible abnormal change trends.

[0069] Magnetic field fluctuations and obstacle density characteristics are overlaid on the inspection route map, using color to represent the fluctuation amplitude and the intensity of abnormal density changes. LiDAR data is used to generate a 3D topographic map, visually displaying the distribution characteristics and dense areas of obstacles. Magnetic field fluctuation characteristics (such as dominant frequency and fluctuation amplitude) and obstacle density variation characteristics (such as abnormal density distribution) are output for subsequent analysis. Areas of high magnetic field fluctuations and dense obstacle concentrations are marked as high-risk areas, supporting inspection route optimization and task planning.

[0070] The magnetic field intensity continuous fluctuation index is generated by analyzing the magnetic field intensity fluctuation characteristics in the extracted environmental magnetic field intensity data. The method for obtaining the magnetic field intensity continuous fluctuation index is as follows:

[0071] Record the output of the magnetometer sensor to obtain a time series of magnetic field strength data, B(s), where s is the time window, typically in Tesla (T) or Gauss (G). Ensure that the sampling frequency, fs, is high enough to satisfy the sampling theorem (typically at least twice the dominant frequency of the signal). For example, if the dominant frequency of the magnetic field fluctuation is 10 Hz, then fs ≥ 20 Hz. Divide the time series B(s) into multiple short-term windows, w(n), each with a length of N samples. Window functions (such as Hanning or Hamming windows) can be used to reduce spectral leakage. The window function is: ; Apply the Fourier transform to the signal within each time window and calculate the frequency components: ;in, is the complex spectrum value at time s and frequency f, B(n) is the signal sample in the window, w(n) is the window function value, is the frequency weight in complex form. Take the modulus of the complex spectrum to get the amplitude spectrum , the expression is: ; For each time window s, calculate the spectral energy density , the expression is: ;in, and are the lowest and highest frequencies of interest, respectively (for example, for low-frequency fluctuations, you can set fmin=0.1 Hz and fmax=10 Hz). E(s) represents the spectral energy density in time window s.

[0072] In order to make the fluctuation index comparable with the magnetic field intensity changes in different time windows or paths, E(s) is normalized: ;in, and are the minimum and maximum spectral energy in the whole sequence, is the normalized spectral energy density. The normalized spectral energy density is smoothed to eliminate short-term fluctuation interference, and a continuous magnetic field intensity fluctuation index HSK is generated. The expression is: ; where Z is the smoothing factor, and its value range is 0 < Z < 1 (usually 0.8 to 0.95). HSK is the continuous magnetic field intensity fluctuation index, is the continuous magnetic field intensity fluctuation index at time

[0073] When the continuous magnetic field intensity fluctuation index is large, it indicates that the magnetic field environment on the inspection path fluctuates violently, and there may be strong magnetic field interference sources (such as high-voltage transmission lines, substations or large industrial equipment). This can have a significant impact on the navigation and positioning systems of the UAV. For example, the magnetometer data may be interfered, resulting in navigation direction deviation or flight trajectory drift. In addition, severe magnetic field interference may affect the collaborative work of GPS and IMU (Inertial Measurement Unit), leading to a decrease in positioning accuracy or even temporary failure. In this case, the navigation and positioning accuracy of the inspection path is significantly reduced, and backup navigation strategies (such as RTK-GPS or visual positioning) must be adopted to ensure the reliability of the inspection task.

[0074] When the continuous magnetic field intensity fluctuation index is small, it indicates that the magnetic field environment on the inspection path is relatively stable and there are no significant external interference sources. This usually means that the working conditions of the UAV navigation and positioning systems are relatively ideal, the magnetometer can normally sense the earth's magnetic field, and the deviation of the navigation direction and flight trajectory is small. In addition, the collaborative performance of navigation components such as GPS and IMU can also be fully exerted, ensuring high-precision path planning and positioning. In this case, the accuracy of the inspection task is relatively high, the data acquisition results are more reliable, and there is no need to adjust the flight path additionally or rely on redundant navigation means.

[0075] After analyzing the abnormal change characteristics of the obstacle density in the extracted obstacle density data, an abnormal change index of obstacle density is generated. The method for obtaining the abnormal change index of obstacle density is:

[0076] Collect obstacle point cloud data through lidar to form a point set ; each point contains its position coordinates , and use the Euclidean distance to calculate the distance between obstacle points and : ; the distance can reflect the spatial density characteristics between points.

[0077] For each point , find its k-nearest neighbor point set , where k is the number of nearest neighbors specified by the user (usually taking 5 to 15, adjusted according to the data scale).

[0078] k-nearest neighbor distance Yes The distance to its kth nearest neighbor: ;definition To its neighboring points The reachable distance , the expression is: ;

[0079] point The local reachable density is all points in its neighborhood The reciprocal mean of the reachable distance is expressed as: Where, yes The number of neighboring points of The local outlier factor is the ratio of the average local reachability density of the points in its neighborhood to its own local reachability density: ;when >1, indicates a point The density of a point is lower than the average density of its neighborhood and may be an outlier.

[0080] Normalize the local outlier factor values of all points to generate the obstacle density abnormal change index, which is expressed as: ; Where GSH is the obstacle density abnormal change index, and are the minimum and maximum local outlier factor values in the dataset, respectively.

[0081] A high obstacle density anomaly variation index indicates significant local anomalies in the distribution of obstacles along the inspection path. For example, certain areas may contain dense clusters of obstacles, prominent obstacle groups, or sudden, complex terrain. In this case, the accuracy of the drone's navigation and positioning systems may be challenged: path planning becomes more difficult, and obstacle avoidance algorithms must adjust their flight paths more frequently, potentially causing navigation delays or even misjudgments. Furthermore, high-density obstacle areas may block GPS signals or affect visual positioning, further reducing positioning accuracy, leading to decreased inspection efficiency and mission reliability.

[0082] When the Obstacle Density Abnormal Variation Index is low, it indicates that obstacles along the inspection path are relatively evenly distributed or have a low density, indicating low environmental complexity. In this scenario, the drone's navigation and positioning systems operate smoothly: path planning can generally proceed according to preset settings without requiring frequent adjustments; the obstacle avoidance algorithm is less burdened, reducing the risk of navigation delays. Furthermore, uniform terrain conditions facilitate the stable operation of the visual navigation system, resulting in high positioning accuracy, thus ensuring efficient inspections and complete data collection.

[0083] S3: Based on the extracted magnetic field intensity fluctuation characteristics and abnormal obstacle density change characteristics, the accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path is determined, and the weighted average summation calculation is performed to obtain the comprehensive inspection accuracy index.

[0084] The continuous fluctuation index of magnetic field intensity and the abnormal change index of obstacle density are converted into comprehensive feature vectors, which are used as inputs of the machine learning model. The machine learning model uses each set of comprehensive feature vectors to predict the accuracy weight assignment label of the UAV navigation and positioning system in each pipeline inspection path as the prediction target, and takes minimizing the sum of the prediction errors of the accuracy weight assignment labels of the UAV navigation and positioning systems in all pipeline inspection paths as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path is determined according to the model output results. Among them, the machine learning model is a polynomial regression model, and the comprehensive inspection accuracy index is obtained by weighted average summation calculation of the accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path.

[0085] The accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path is obtained by obtaining the corresponding function expression from the comprehensive feature vector training data of the trained machine learning model: Where, is the output function of the model, HSK is the continuous fluctuation index of magnetic field intensity, GSH is the abnormal change index of obstacle density, Assign accuracy weights to the drone navigation and positioning systems in each pipeline inspection path.

[0086] S4: Compare the comprehensive inspection accuracy index in the pipeline inspection path with the gradient standard threshold, and divide the accuracy of the UAV pipeline inspection path according to the comparison result. The division results include accuracy inspection, incomplete accuracy inspection and inaccuracy inspection.

[0087] Comparing the obtained comprehensive inspection accuracy index of the pipeline inspection path with a gradient standard threshold, the gradient standard threshold including a first standard threshold and a second standard threshold, wherein the first standard threshold is less than the second standard threshold, and comparing the comprehensive inspection accuracy index of the pipeline inspection path with the first standard threshold and the second standard threshold respectively;

[0088] If the comprehensive inspection accuracy index in the pipeline inspection path is greater than the second standard threshold, it means that the accuracy of the UAV navigation and positioning system in the pipeline inspection path is high. At this time, a high-accuracy inspection signal is generated, and the accuracy of the UAV pipeline inspection path is classified as an accuracy inspection;

[0089] If the comprehensive inspection accuracy index in the pipeline inspection path is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the accuracy of the UAV navigation and positioning system in the pipeline inspection path is medium. At this time, a medium accuracy inspection signal is generated, and the accuracy of the UAV pipeline inspection path is classified as incomplete accuracy inspection;

[0090] If the comprehensive inspection accuracy index in the pipeline inspection path is less than the first standard threshold, it means that the accuracy of the UAV navigation and positioning system in the pipeline inspection path is low. At this time, a low-accuracy inspection signal is generated, and the accuracy of the UAV pipeline inspection path is classified as inaccurate inspection.

[0091] S5: For accurate inspections, the inspection continues along the established route without adjusting the flight strategy. For inaccurate inspections, the inspection mission of the drone is immediately interrupted, the optimal inspection route is recalculated, and the abnormal pipeline section is covered first.

[0092] For accuracy inspections, this indicates that the navigation and positioning systems perform reliably and accurately on the current path. Handling measures include: Continue inspections along the established path: No need to adjust the flight strategy, and perform tasks normally along the pre-planned inspection path. Action steps: Maintain the current flight altitude, speed, and navigation strategy of the drone. Collect data (such as thermal imaging, high-definition images, or gas detection data) as planned. Avoid unnecessary path adjustments to improve mission efficiency. Regularly monitor the stability of the navigation system during the inspection process to ensure that accuracy remains at a high level. If real-time data changes significantly (such as abnormally increased magnetic field fluctuations or a sudden increase in obstacle density), dynamically evaluate the mission status.

[0093] Inaccurate inspections indicate poor performance and low accuracy of the navigation and positioning systems on the current route. Actions include: Immediately abort the mission: Action steps: The drone automatically returns home or flies to a safe standby point to avoid further low-quality data collection. Mark the inspection status of the current route as "failed" and record information about possible interference sources (such as areas with strong magnetic fields or dense obstacles). This prevents continued flight from causing false or missed inspections and wasting mission resources. Prioritize inspections of abnormal pipeline sections to ensure that inspections in high-risk areas are completed as quickly as possible. Utilize data on fluctuations in magnetic field intensity and abnormal changes in obstacle density to identify areas of severe interference and exclude them from route planning. Use alternative navigation modes (such as RTK-GPS or vision-based navigation) to improve navigation accuracy. Generate an optimal inspection route using a path planning algorithm (such as A* or Dijkstra's algorithm) that avoids interference areas. The drone re-executes the inspection along the new route, focusing on previously interfered areas. Adjust flight altitude or speed to adapt to the new environmental complexity.

[0094] After completing the mission, the two inspection results are compared to ensure that all critical pipeline sections are effectively covered. Path failure information is fed back to the navigation model to optimize subsequent mission planning.

[0095] S6: For incomplete accuracy inspections, the accuracy anomaly of the drone’s navigation and positioning system in a subsequent fixed time period is predicted. If the accuracy anomaly is high, the drone’s inspection path is replanned, and the drone’s flight altitude is increased and areas with dense obstacles are avoided.

[0096] For incomplete accuracy inspection, that is, the comprehensive inspection accuracy index in the generated pipeline inspection path is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, in the future T time period, the accuracy abnormality of the navigation and positioning system is determined by the comprehensive prediction index express: ;in: is the predicted value of the magnetic field intensity fluctuation index at future time t (based on time series prediction, such as the ARIMA model). GSH(t) is the predicted value of the obstacle density abnormal change index at future time t (based on the obstacle distribution model). CI(t) is the comprehensive inspection accuracy index at current time t. α, β, and γ are weight coefficients, representing the proportion of the impact of magnetic field fluctuation, obstacle density, and current comprehensive accuracy on the predicted value, satisfying α + β + γ = 1.

[0097] Use time series prediction models (such as LSTM) to make short-term predictions for HSK(t) and GSH(t): , ; Define the threshold Pthreshold of the abnormal degree of prediction accuracy: if , indicating that the accuracy is abnormally high and the path needs to be replanned; if , continue to inspect along the existing route.

[0098] The goal of path replanning is to avoid obstacle-dense areas with the shortest path and optimize navigation accuracy. The objective function of the path planning problem is: Where, is the length of path segment i, is the predicted anomaly level of path segment i, k is a penalty factor used to increase the path weight in areas with high anomalies and avoid high-risk areas, and W is the total number of path segments. Path planning can use the A* algorithm or the Dijkstra algorithm to dynamically adjust path weights to achieve optimal path calculation.

[0099] The flight altitude adjustment rule is: define the adjusted flight altitude , the expression is: Where, is the current flight altitude, and ΔH is the altitude increment coefficient (which can be set according to the distribution characteristics of obstacle density).

[0100] Mark areas with high obstacle density and set them as inaccessible areas during path planning. The regional density weight formula is: Where, is the regional weight, and GSH(j) is the abnormal change index of obstacle density in the region. During path replanning, regions with regional weights greater than the regional weight threshold preset based on historical data are excluded.

[0101] In this embodiment, drone-mounted sensors acquire real-time magnetic field interference parameters (including ambient magnetic field intensity data) and terrain complexity parameters (including obstacle density data) along the pipeline inspection path. This data is preprocessed to extract characteristics of magnetic field intensity fluctuations and abnormal changes in obstacle density. Based on these extracted characteristics, accuracy weights for the drone's navigation and positioning systems are calculated, and a comprehensive inspection accuracy index is derived through weighted average summation. After comparing the comprehensive inspection accuracy index with a gradient standard threshold, the inspection path accuracy is classified into three categories: accurate inspection, incomplete accuracy inspection, and inaccuracy inspection. For accurate inspections, no flight strategy adjustment is required. For inaccuracy inspections, the mission is immediately terminated and the route is replanned, prioritizing coverage of abnormal pipeline sections. For incomplete accuracy inspections, the accuracy of the navigation and positioning systems is predicted over time. If the degree of anomaly is high, the route is replanned, increasing the flight altitude and avoiding areas with dense obstacles to ensure the reliability and accuracy of the inspection mission.

[0102] Example 2: The drone-based artificial intelligence inspection system described in this example includes a data acquisition module, a feature extraction module, a navigation and positioning accuracy assessment module, an accuracy classification module, a path planning and mission control module, and a dynamic prediction and strategy optimization module.

[0103] Data acquisition module: uses drone-mounted sensors to acquire magnetic field interference parameters and terrain complexity parameters in different pipeline inspection paths in real time. The magnetic field interference parameters include ambient magnetic field intensity data, and the terrain complexity parameters include obstacle density data.

[0104] Feature extraction module: pre-processes the acquired environmental magnetic field strength data and obstacle density data, and extracts the magnetic field intensity fluctuation characteristics in the environmental magnetic field strength data and the obstacle density abnormal change characteristics in the obstacle density data;

[0105] Navigation and positioning accuracy assessment module: Based on the extracted magnetic field intensity fluctuation characteristics and abnormal obstacle density change characteristics, the module determines the accuracy weight of the drone navigation and positioning system in each pipeline inspection path, and calculates the weighted average sum of these values to obtain the comprehensive inspection accuracy index;

[0106] Accuracy classification module: compares the comprehensive inspection accuracy index in the pipeline inspection path with the gradient standard threshold, and classifies the accuracy of the UAV pipeline inspection path according to the comparison results. The classification results include accurate inspection, incomplete accuracy inspection and inaccuracy inspection;

[0107] Path planning and mission control module: For accurate inspections, the inspection continues along the established path without adjusting the flight strategy; for inaccurate inspections, the drone's inspection mission is immediately interrupted, the optimal inspection path is recalculated, and the abnormal pipeline section is covered first;

[0108] Dynamic prediction and strategy optimization module: For incomplete accuracy inspections, the accuracy anomaly of the drone's navigation and positioning system in a subsequent fixed time period is predicted. If the accuracy anomaly is high, the drone's inspection path is replanned, the drone's flight altitude is increased, and areas with dense obstacles are avoided.

[0109] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0110] It should be understood that the term "and / or" as used herein simply describes a relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the associated objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0111] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0112] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A drone-based artificial intelligence inspection method, characterized by: The following steps are involved: S1: Using drone-mounted sensors to obtain real-time magnetic field interference parameters and terrain complexity parameters along different pipeline inspection routes. The magnetic field interference parameters include ambient magnetic field intensity data, and the terrain complexity parameters include obstacle density data. S2: Preprocess the acquired environmental magnetic field strength data and obstacle density data, and extract the magnetic field strength fluctuation characteristics in the environmental magnetic field strength data and the obstacle density abnormal change characteristics in the obstacle density data, specifically including: The magnetic field intensity continuous fluctuation index is generated by analyzing the magnetic field intensity fluctuation characteristics in the extracted environmental magnetic field intensity data. The method for obtaining the magnetic field intensity continuous fluctuation index is as follows: the output of the magnetometer sensor is recorded to obtain the time series data B(s) of the magnetic field intensity; for each time window s, the spectrum energy density is calculated and normalized to obtain the spectrum energy density; the normalized spectrum energy density is smoothed to generate the magnetic field intensity continuous fluctuation index; The obstacle density abnormal change index is generated after analyzing the abnormal change characteristics of the obstacle density in the extracted obstacle density data. The method for obtaining the obstacle density abnormal change index is as follows: the obstacle point cloud data is collected by lidar, and the obstacle point is calculated using the Euclidean distance. and The distance between each point , find its k-nearest neighbor set, where k is the number of nearest neighbors specified by the user, and calculate To its neighboring points The reachable distance; calculate the point based on the reachable distance The local reachability density of all points in the neighborhood The reciprocal mean of the reachable distance; calculation point The local outlier factor is the average local reachability density of the neighborhood points and the point The local outlier factor values of all points are normalized to generate the obstacle density abnormal change index; S3: Based on the extracted magnetic field intensity fluctuation characteristics and obstacle density abnormal change characteristics, the accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path is determined, and the weighted average summation calculation is performed to obtain the comprehensive inspection accuracy index; S4: Compare the comprehensive inspection accuracy index in the pipeline inspection path with the gradient standard threshold, and classify the accuracy of the UAV pipeline inspection path according to the comparison result, and the classification results include accuracy inspection, incomplete accuracy inspection and inaccuracy inspection; S5: For accurate inspections, the inspection continues along the established route without adjusting the flight strategy. For inaccurate inspections, the inspection mission of the drone is immediately interrupted, and the optimal inspection route is recalculated, with priority given to covering the abnormal pipeline section. S6: For incomplete accuracy inspections, the accuracy anomaly of the drone’s navigation and positioning system in a subsequent fixed time period is predicted. If the accuracy anomaly is high, the drone’s inspection path is replanned, and the drone’s flight altitude is increased and areas with dense obstacles are avoided.

2. The drone-based artificial intelligence inspection method according to claim 1, characterized in that: In S2, the magnetic field intensity fluctuation characteristics in the extracted environmental magnetic field intensity data are analyzed to generate a magnetic field intensity continuous fluctuation index. The method for obtaining the magnetic field intensity continuous fluctuation index is: Record the output of the magnetometer sensor to obtain the time series data B(s) of the magnetic field intensity, where s is the time window. Divide the time series B(s) into short time windows w(n), and the length of each window is N samples; Apply the Fourier transform to the signal within each time window and calculate the frequency components: ;in, is the complex spectrum value at time s and frequency f, B(n) is the signal sample in the window, w(n) is the window function value, is the frequency weight in complex form; take the modulus of the complex spectrum to get the amplitude spectrum , the expression is: ; For each time window s, calculate the spectral energy density , the expression is: ;in, and are the lowest and highest frequencies of interest, respectively, and E(s) represents the spectral energy density of the time window s; E(s) is normalized to: ;in, and are the minimum and maximum spectral energy in the whole sequence, is the normalized spectrum energy density; the normalized spectrum energy density is smoothed to generate the magnetic field intensity continuous fluctuation index HSK, which is expressed as: ; Where Z is the smoothing factor, HSK is the continuous fluctuation index of magnetic field intensity, for The continuous fluctuation index of magnetic field intensity at the moment.

3. The drone-based artificial intelligence inspection method according to claim 2, characterized in that: In S2, the obstacle density abnormal change characteristics in the extracted obstacle density data are analyzed to generate an obstacle density abnormal change index. The obstacle density abnormal change index is obtained as follows: Obstacle point cloud data is collected through lidar to form a point set ; Each point Contains its location coordinates , using Euclidean distance to calculate obstacle points and Distance between: ; For each point , find its k-nearest neighbor set , where k is the number of nearest neighbors specified by the user; k-nearest neighbor distance Yes The distance to its kth nearest neighbor: ;definition To its neighboring points The reachable distance , the expression is: ; point The local reachable density is all points in its neighborhood The reciprocal mean of the reachable distance is expressed as: Where, yes The number of neighboring points of The local outlier factor is the ratio of the average local reachability density of the points in its neighborhood to its own local reachability density: ; Normalize the local outlier factor values of all points to generate the obstacle density abnormal change index, which is expressed as: ; Where GSH is the obstacle density abnormal change index, and are the minimum and maximum local outlier factor values in the dataset, respectively.

4. The drone-based artificial intelligence inspection method according to claim 3 is characterized by: In S3, based on the extracted magnetic field intensity fluctuation characteristics and obstacle density abnormal change characteristics, the accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path is determined, and the weighted average summation calculation is performed to obtain the comprehensive inspection accuracy index, which is specifically: The continuous fluctuation index of magnetic field intensity and the abnormal change index of obstacle density are converted into comprehensive feature vectors, which are used as inputs of the machine learning model. The machine learning model uses each set of comprehensive feature vectors to predict the accuracy weight assignment label of the UAV navigation and positioning system in each pipeline inspection path as the prediction target, and takes minimizing the sum of the prediction errors of the accuracy weight assignment labels of the UAV navigation and positioning systems in all pipeline inspection paths as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the model training is stopped. The accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path is determined according to the model output results. Among them, the machine learning model is a polynomial regression model, and the comprehensive inspection accuracy index is obtained by weighted average summation calculation of the accuracy weight assignment of the UAV navigation and positioning system in each pipeline inspection path.

5. The method for inspection based on drone artificial intelligence according to claim 4, characterized in that: In S4, the comprehensive inspection accuracy index in the pipeline inspection path is compared with the gradient standard threshold, and the accuracy of the UAV pipeline inspection path is divided according to the comparison result. The division results include accuracy inspection, incomplete accuracy inspection and inaccuracy inspection; Comparing the obtained comprehensive inspection accuracy index of the pipeline inspection path with a gradient standard threshold, the gradient standard threshold including a first standard threshold and a second standard threshold, wherein the first standard threshold is less than the second standard threshold, and comparing the comprehensive inspection accuracy index of the pipeline inspection path with the first standard threshold and the second standard threshold respectively; If the comprehensive inspection accuracy index in the pipeline inspection path is greater than the second standard threshold, it means that the accuracy of the UAV navigation and positioning system in the pipeline inspection path is high. At this time, a high-accuracy inspection signal is generated, and the accuracy of the UAV pipeline inspection path is classified as an accuracy inspection; If the comprehensive inspection accuracy index in the pipeline inspection path is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it means that the accuracy of the UAV navigation and positioning system in the pipeline inspection path is medium. At this time, a medium accuracy inspection signal is generated, and the accuracy of the UAV pipeline inspection path is classified as incomplete accuracy inspection; If the comprehensive inspection accuracy index in the pipeline inspection path is less than the first standard threshold, it means that the accuracy of the UAV navigation and positioning system in the pipeline inspection path is low. At this time, a low-accuracy inspection signal is generated, and the accuracy of the UAV pipeline inspection path is classified as inaccurate inspection.

6. The drone-based artificial intelligence inspection method according to claim 5, characterized in that: In S6, for incomplete accuracy inspections, the accuracy anomaly of the drone's navigation and positioning system in a subsequent fixed time period is predicted. If the accuracy anomaly is high, the drone's inspection path is replanned, and the drone's flight altitude is increased and areas with dense obstacles are avoided. Specifically: For incomplete accuracy inspection, that is, the comprehensive inspection accuracy index in the generated pipeline inspection path is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, in the future T time period, the accuracy abnormality of the navigation and positioning system is determined by the comprehensive prediction index express: ;in: is the predicted value of the magnetic field intensity fluctuation index in the future time t, GSH(t) is the predicted value of the obstacle density abnormal change index in the future time t, CI(t) is the comprehensive inspection accuracy index at the current time t, α, β, γ are weight coefficients, indicating the proportion of the influence of magnetic field fluctuation, obstacle density and current comprehensive accuracy on the predicted value, satisfying α+β+γ=1; the time series prediction model is used to make short-term predictions for HSK(t) and GSH(t): , ; Define the threshold Pthreshold of the abnormal degree of prediction accuracy: if , indicating that the accuracy is abnormally high and the path needs to be replanned; if , continue to inspect along the existing route.

7. The drone-based artificial intelligence inspection method according to claim 6, characterized in that: The goal of path replanning is to avoid the obstacle-dense area with the shortest path and optimize the navigation accuracy. The objective function of the path planning problem is , the expression is: Where, is the length of path segment i, is the predicted abnormality level of path segment i, k is the penalty coefficient used to increase the path weight of high abnormality areas and avoid high-risk areas, and W is the total number of path segments; The flight altitude adjustment rule is: define the adjusted flight altitude , the expression is: Where, is the current flight altitude, ΔH is the altitude increment coefficient; mark the high obstacle density area and set it as an impassable area during path planning. The regional density weight formula is: Where, is the regional weight, GSH(j) is the abnormal change index of obstacle density in the region, and when replanning the path, the regions whose regional weight is greater than the regional weight threshold preset according to historical data are excluded.

8. A drone-based artificial intelligence inspection system, used to implement the drone-based artificial intelligence inspection method according to any one of claims 1 to 7, characterized in that: It includes data acquisition module, feature extraction module, navigation and positioning accuracy assessment module, accuracy classification module, path planning and task control module, and dynamic prediction and strategy optimization module; Data acquisition module: uses drone-mounted sensors to acquire magnetic field interference parameters and terrain complexity parameters in different pipeline inspection paths in real time. The magnetic field interference parameters include ambient magnetic field intensity data, and the terrain complexity parameters include obstacle density data. Feature extraction module: pre-processes the acquired environmental magnetic field strength data and obstacle density data, and extracts the magnetic field intensity fluctuation characteristics in the environmental magnetic field strength data and the obstacle density abnormal change characteristics in the obstacle density data; Navigation and positioning accuracy assessment module: Based on the extracted magnetic field intensity fluctuation characteristics and abnormal obstacle density change characteristics, the module determines the accuracy weight of the drone navigation and positioning system in each pipeline inspection path, and calculates the weighted average sum of these values to obtain the comprehensive inspection accuracy index; Accuracy classification module: compares the comprehensive inspection accuracy index in the pipeline inspection path with the gradient standard threshold, and classifies the accuracy of the UAV pipeline inspection path according to the comparison results. The classification results include accurate inspection, incomplete accuracy inspection and inaccuracy inspection; Path planning and mission control module: For accurate inspections, the inspection continues along the established path without adjusting the flight strategy; for inaccurate inspections, the drone's inspection mission is immediately interrupted, the optimal inspection path is recalculated, and the abnormal pipeline section is covered first; Dynamic prediction and strategy optimization module: For incomplete accuracy inspections, the accuracy anomaly of the drone's navigation and positioning system in a subsequent fixed time period is predicted. If the accuracy anomaly is high, the drone's inspection path is replanned, the drone's flight altitude is increased, and areas with dense obstacles are avoided.

Images