Multi-type unmanned aerial vehicle maintenance method and system

Abstract

The invention discloses a multi-type unmanned aerial vehicle maintenance method and system, and the method comprises the following steps: S1, obtaining the surface fault data of an unmanned aerial vehicle through recognizing the appearance image of the unmanned aerial vehicle; s2, establishing communication connection with the unmanned aerial vehicle to obtain internal parameters of the unmanned aerial vehicle; s3, sending the surface fault data and the internal parameters of the unmanned aerial vehicle to a preset database, and generating an unmanned aerial vehicle maintenance scheme after fault analysis and data comparison; and S4, repairing or maintaining the unmanned aerial vehicle according to the unmanned aerial vehicle maintenance scheme. The AI large model is adopted for deep machine learning, information such as technical data and maintenance cases of various unmanned aerial vehicles is comprehensively collected, a rich unmanned aerial vehicle knowledge base is constructed, surface fault data and internal parameters of the unmanned aerial vehicles are automatically obtained, the knowledge base and a 3D model base are combined for intelligent analysis, and a maintenance scheme is generated. The method realizes standardized fault diagnosis process and automatic maintenance operation, and has the advantages of improving the maintenance efficiency, improving the fault diagnosis accuracy and reducing the manual dependence degree.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicles (UAVs), and in particular to a maintenance method and system for multiple types of UAVs. Background Art

[0002] Drones are high-precision devices with a high failure rate of components. They require regular maintenance, which requires technicians to perform on a maintenance platform. This is time-consuming and labor-intensive to locate the fault point, reducing maintenance efficiency.

[0003] Drone maintenance primarily relies on manual inspection and repair by maintenance personnel on maintenance platforms. This traditional approach presents numerous problems: First, maintenance personnel spend a significant amount of time troubleshooting, particularly for surface faults like fuselage cracks and propeller deformation, as well as abnormalities in internal parameters such as the flight control system and powertrain. Second, the accuracy of fault diagnosis depends largely on the maintenance personnel's experience, lacking a standardized diagnostic process. Third, the maintenance process requires extensive reference to technical documentation and maintenance manuals, resulting in low efficiency.

[0004] To address the aforementioned issues, patent application number 202210381187.9 discloses a drone maintenance method, apparatus, device, and storage medium. The method includes: obtaining the model, brand, and version information of the drone to be repaired; determining the drone's maintenance instructions from a pre-set drone maintenance database based on the model, brand, and version information; and displaying the maintenance instructions so that a maintenance worker can use them to repair the drone.

[0005] The above technical solution downloads the maintenance manual of the drone. The maintenance personnel can first repair the drone according to the maintenance manual. They can first focus on checking the key components of the drone, because the key components have a relatively high failure rate. This can improve the efficiency of drone maintenance, improve the targeted maintenance, and save the workload of maintenance personnel.

[0006] However, this solution only provides static maintenance guidance and lacks intelligent fault diagnosis. Maintenance personnel still need to independently determine the type and severity of the fault, making it difficult to accurately identify complex issues such as fuselage wear and abnormal motor speed. Furthermore, the solution lacks support for visualizing the drone's three-dimensional structure, preventing maintenance personnel from intuitively understanding the specific location and damage of faulty components. More importantly, existing technologies are unable to integrate surface fault detection with internal parameter analysis, making it difficult to develop a systematic maintenance plan, resulting in inconsistent maintenance quality.

[0007] Therefore, it is necessary for us to improve the above-mentioned prior art to overcome the above-mentioned defects. Summary of the Invention

[0008] The purpose of the present invention is to provide a multi-type UAV maintenance method, system, electronic device and computer-readable storage medium, which has the advantages of improving maintenance efficiency, increasing fault diagnosis accuracy and reducing manual dependence.

[0009] The above technical objectives of the present invention are achieved through the following technical solutions:

[0010] A multi-type drone maintenance method includes the following steps:

[0011] S1. Obtain surface fault data of the drone by identifying its appearance image;

[0012] S2. Establish a communication connection with the drone to obtain the drone's internal parameters;

[0013] S3. Send the drone's surface fault data and internal parameters to a pre-set database for fault analysis and data comparison, generating a drone maintenance plan.

[0014] S4. Perform repair or maintenance operations on the drone according to the drone maintenance plan.

[0015] Furthermore, the present application also proposes that surface fault data is obtained by identifying the appearance image of the drone through an image recognition algorithm, and the surface fault data includes fuselage cracks, fuselage wear and propeller deformation data.

[0016] Furthermore, the present application also proposes that the internal parameters of the drone are obtained through a wireless communication module that can establish a communication connection with the drone, and the internal parameters include flight control system parameters and power system parameters.

[0017] Furthermore, the present application also proposes that the flight control system parameters include attitude data, flight speed and flight altitude; the power system parameters include motor speed, battery voltage and current.

[0018] Furthermore, the present application also proposes that the database includes a knowledge base and a 3D model library of drones;

[0019] The knowledge base is constructed as follows:

[0020] First, AI big model technology is used to collect technical documents, maintenance manuals, and fault case data for different brands and models of drones. The collected data is then cleaned, classified, and labeled, and input into a deep learning model for training to build a drone knowledge base.

[0021] The construction method of the 3D model library is as follows:

[0022] Using 3D modeling software, based on the drone's design drawings or actual measurement data, an accurate 3D model is created for each type of drone to form a 3D model library; the 3D model has full-scale display, zoom in, zoom out and rotation functions, making it easy to view details during fault analysis.

[0023] Furthermore, this application also proposes that after fault analysis and data comparison, the maintenance decision module generates a drone maintenance plan based on the results of fault analysis and data comparison, combined with the structural characteristics of the 3D model library and the technical documents, maintenance manuals and fault case data in the knowledge base; the content of the drone maintenance plan includes maintenance steps, maintenance tools and parts replacement recommendations.

[0024] Furthermore, the present application also proposes that the drone maintenance plan is executed by an execution module, which includes a multi-axis robotic arm system. The end of the multi-axis robotic arm system is used to be equipped with maintenance tools. The multi-axis robotic arm system plans the motion path and operation steps according to the drone maintenance plan to perform repair or maintenance operations on the drone.

[0025] A multi-type UAV maintenance system, including

[0026] The visual recognition module is used to obtain the appearance images of the UAV from different angles and analyze the appearance images through image recognition algorithms to obtain the surface fault data of the UAV;

[0027] Wireless communication module, used to obtain the internal parameters of the drone by establishing communication with the drone;

[0028] The data processing module is used to perform fault analysis and data comparison on the surface fault data and internal parameters of the UAV with the standard data stored in the database;

[0029] The maintenance decision module is used to generate a maintenance plan for the drone based on the results of fault analysis and data comparison, combined with the drone's knowledge base and 3D model library;

[0030] The execution module is used to plan the motion path and operation steps according to the drone maintenance plan and perform repair or maintenance operations on the drone.

[0031] An electronic device includes: at least one processor and at least one memory; the memory is used to store one or more program instructions; the processor is used to run the one or more program instructions to execute the above method.

[0032] The present application also proposes a computer-readable storage medium, which contains one or more program instructions, and the one or more program instructions are used to execute the above method.

[0033] In summary, the present invention has the following beneficial effects:

[0034] Using AI large models for deep machine learning, we comprehensively collect technical data, maintenance cases and other information of various types of drones, build a rich drone knowledge base, and intelligently generate the most appropriate repair or maintenance plan based on the characteristics and requirements of different types of drones with the help of the drone knowledge base.

[0035] From the above, it can be seen that the present application provides a multi-type drone maintenance method, system, electronic device and computer-readable storage medium, which uses AI large models for deep machine learning, comprehensively collects technical data, maintenance cases and other information of various types of drones, and builds a rich drone knowledge base. By automatically obtaining drone surface fault data and internal parameters, combining the knowledge base with the 3D model library for intelligent analysis to generate maintenance plans, it realizes standardized fault diagnosis processes and automated maintenance operations, which has the advantages of improving maintenance efficiency, improving fault diagnosis accuracy and reducing manual dependence. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of Example 1 described in the present invention.

[0037] Figure 2 It is a schematic diagram of the fixing portion in Example 1 described in the present invention. DETAILED DESCRIPTION

[0038] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below with reference to diagrams and specific embodiments.

[0039] like Figure 1 and Figure 2 As shown, the present invention proposes a multi-type UAV maintenance method, which includes obtaining surface fault data by identifying the appearance image of the UAV; obtaining internal parameters by establishing a communication connection with the UAV; sending the surface fault data and internal parameters to a database for fault analysis and data comparison to generate a maintenance plan; and performing repair or maintenance operations according to the maintenance plan.

[0040] Among them, identifying the appearance image of the drone refers to using a multi-angle image acquisition device to obtain complete visual information of the drone's outer shell. Specifically, this can be achieved by using a ring-arranged high-definition camera array to ensure that there is no blind spot coverage on the fuselage surface. Establishing a communication connection refers to communicating with the drone's data interface through a wireless transmission protocol. Specifically, this can be achieved by using Bluetooth, Wi-Fi, or a dedicated radio frequency module to meet the communication needs of different types of equipment. Fault analysis in the database refers to pattern matching the collected data with pre-stored standard parameters. Specifically, a convolutional neural network algorithm can be used to process image data, and a time series analysis algorithm can be used to process sensor data. Generating a maintenance plan refers to outputting a set of operating instructions based on the type and severity of the fault. Specifically, a decision tree algorithm can be used in combination with a maintenance knowledge base to generate a step-by-step operating guide.

[0041] Specifically, when a drone undergoing maintenance enters the work area, a circular camera array automatically captures images of its exterior. The image processing unit identifies deformation features in the propeller, while the communication module simultaneously reads data indicating abnormal motor speed fluctuations. These two types of data are synchronously transmitted to the central processing unit and compared with the standard parameters of similar drones stored in a database. The analysis system discovered that propeller deformation has led to a decrease in aerodynamic efficiency, which in turn has increased motor load and caused speed fluctuations. Based on this diagnostic result, the maintenance plan generation module calls the installation parameters for this propeller model from the 3D model library and instructs the robotic arm to perform the replacement operation. The entire process requires no human intervention, forming a closed-loop control from data acquisition to maintenance completion.

[0042] Compared with existing technologies, traditional methods require manual inspection of the exterior and manual connection to the debug interface to read data, which carries the risk of missed detections and operational delays. This solution utilizes automated image recognition and wireless communication technology to simultaneously collect surface damage and internal parameters, eliminating detection blind spots caused by human factors. Existing maintenance solutions rely on the experience and judgment of maintenance personnel, while this solution, through database matching and algorithmic analysis of historical cases, can identify the correlation between complex faults, for example, simultaneously discovering the causal relationship between structural damage and electrical parameter anomalies.

[0043] Through the above-mentioned technical solutions, this application achieves automated diagnosis and precise maintenance of drone faults. Surface image recognition technology effectively detects subtle structural damage, avoiding oversights that can occur during manual visual inspections. Wireless reading of internal parameters overcomes physical interface limitations and improves data collection efficiency. The intelligent analysis capabilities of the database replace empirical judgments, ensuring the scientific nature and repeatability of maintenance plans. This method significantly reduces the technical reliance on professionals, enabling non-technical personnel to perform high-quality maintenance operations while also shortening the time required for fault diagnosis and maintenance implementation.

[0044] The present application further proposes that surface fault data is obtained by identifying the appearance image of the drone through an image recognition algorithm, and the surface fault data includes fuselage cracks, fuselage wear and propeller deformation data.

[0045] Image recognition algorithms refer to automated analysis technologies based on computer vision. Specifically, convolutional neural network models can be used to extract and classify features from drone exterior images, and trained models can be used to identify areas with abnormal textures, such as cracks and wear. Fuselage cracks are linear fractures on the drone's exterior caused by external impact or material fatigue. Edge detection algorithms can capture these geometric features. Fuselage wear refers to the loss of surface coatings or structural wear caused by long-term use. The extent of wear can be quantified using color contrast analysis and texture recognition algorithms. Propeller deformation data includes blade bending angles and symmetry deviation parameters. Three-dimensional contour reconstruction algorithms can accurately measure these deformations.

[0046] Specifically, images of the drone's exterior are fed into a pre-trained deep learning model. The model extracts local features from the image through multi-layer convolution operations and, combined with fully connected layers, locates and classifies cracks and worn areas. For propeller deformation detection, a multi-view image acquisition device is used to acquire stereoscopic image data of the blades. A point cloud registration algorithm is then used to reconstruct a 3D model, which is then compared with a standard model to generate deformation parameters. During the detection process, the algorithm can identify fuselage cracks with submillimeter accuracy and maintain a measurement error of ±0.5 degrees for propeller bend angles.

[0047] Compared to existing technologies, traditional drone surface inspection relies on manual visual inspection or handheld measuring tools, which are inefficient and highly subjective. For example, manual inspection of propeller deformation requires disassembly of the blades and measurement with calipers, which can take over 10 minutes. However, this solution, using a 3D reconstruction algorithm, can complete non-contact measurement in just 30 seconds. Furthermore, manual inspection can easily overlook fine cracks, while image recognition algorithms can detect cracks as small as 0.1 mm in width, reducing the missed detection rate to less than 5%.

[0048] Through the above-mentioned technical solution, this application achieves automated detection of drone surface faults, increasing detection speed to over 10 times that of manual operation while avoiding misjudgments due to inexperience. For critical faults such as propeller deformation, detection accuracy meets flight safety standards, effectively preventing flight accidents caused by blade imbalance. Quantified data on fuselage wear provides an objective basis for determining maintenance priorities, avoiding over- or under-repair.

[0049] The present application further proposes that the internal parameters of the drone are obtained through a wireless communication module that can establish a communication connection with the drone, and the internal parameters include flight control system parameters and power system parameters.

[0050] The wireless communication module refers to the hardware unit used to establish a wireless data transmission channel. This can be implemented using a Bluetooth module, Wi-Fi module, or ZigBee module, establishing a two-way data exchange channel with the drone via a wireless communication protocol. Flight control system parameters refer to a dynamic data set reflecting the operating status of the flight control system. These parameters may include real-time flight status information such as three-axis acceleration, gyroscope angle, and GPS positioning coordinates. Power system parameters refer to physical quantities that characterize the operating performance of the power unit. These parameters may include periodically updated operating indicators such as brushless motor speed, remaining lithium battery charge, and electronically controlled output power.

[0051] Specifically, when the drone is in standby or operational mode, the wireless communication module initiates a handshake signal, establishing a data link with the drone via a pre-set communication protocol. Flight control system parameters are transmitted to the maintenance terminal via a wireless channel at a fixed frequency, reflecting the flight control unit's operating status in real time. Power system parameters are continuously collected during drone operation and synchronized to the maintenance system via the wireless communication module. If a parameter anomaly is detected, the maintenance system immediately triggers an alarm and generates appropriate maintenance instructions based on the degree of parameter deviation.

[0052] Compared to existing technologies, traditional maintenance methods require manual disassembly of the outer shell and connection of physical interfaces to read internal parameters. This solution, however, utilizes a wireless communication module for contactless data collection, avoiding maintenance interruptions caused by physical interface incompatibilities. Existing technologies only capture static parameters and suffer from data lags. This solution, through real-time transmission of flight control and powertrain parameters, enables the maintenance system to detect transient abnormalities.

[0053] Through the above technical solution, this application achieves automated collection and real-time monitoring of drone internal operating parameters, enabling diagnosis of the electronic system status of multiple drone types without human intervention, effectively shortening fault detection time. The simultaneous acquisition of flight control system parameters and power system parameters provides a data foundation for correlation analysis of complex faults, enabling maintenance plans to cover coordinated anomalies of the mechanical structure and electronic control system.

[0054] This application further proposes that the flight control system parameters include attitude data, flight speed and flight altitude, and the power system parameters include motor speed, battery voltage and current.

[0055] Attitude data refers to the drone's pitch, roll, and yaw angles in three-dimensional space. It can be collected using an inertial measurement unit (IMU) and used to identify any abnormalities in flight attitude stability. Flight speed refers to the drone's real-time velocity in the air. It can be measured using a GPS module or airspeed meter and used to detect deviations in flight control logic. Flight altitude refers to the drone's vertical distance from the ground. It can be measured using a barometer or ultrasonic sensor and used to determine the operating status of the altitude control system. Motor speed refers to the rotational speed of the motor driving the propeller. It can be collected using a Hall effect sensor or photoelectric encoder and used to identify abnormal power output. Battery voltage refers to the output voltage of the drone's power system and can be monitored in real time using a voltage sensor to assess the remaining battery capacity. Current refers to the total current consumption of the drone during operation and can be collected using a current sensor to detect short circuits or overloads.

[0056] Specifically, by simultaneously collecting attitude data, flight speed, and flight altitude, a multi-dimensional operating status model of the flight control system can be established. For example, when mismatched fluctuations occur in flight speed and attitude angle data, it can be determined that there is a logical error in the flight control algorithm; when the correlation between flight altitude data and motor speed deviates from the preset threshold, abnormal propeller power output can be identified. Among the power system parameters, the synchronous monitoring of motor speed and battery voltage and current forms a cross-validation mechanism. For example, if the battery voltage is normal but the current suddenly increases, it can be determined that the motor has a mechanical jam fault. This multi-parameter combination monitoring method can distinguish the root causes of mechanical failures from electronic control system failures through correlation analysis between data.

[0057] Compared to existing technologies, traditional drone maintenance methods typically only collect a single parameter, such as monitoring battery voltage or flight altitude. These methods are unable to determine whether voltage anomalies are caused by battery aging or a sudden change in current due to motor overload. This solution simultaneously acquires key parameters from both the flight control system and the powertrain, establishing a dynamic correlation model between these parameters. This enables the system to identify the causal relationships between complex faults. For example, in the event of an abnormal flight attitude, the system can combine motor speed data to determine whether the cause is propeller deformation or a flight control sensor failure.

[0058] Through the above technical solution, this application can accurately distinguish the physical causes of drone failures from control system anomalies, avoiding misjudgments caused by insufficient parameter collection dimensions. For example, when a battery voltage drop is detected, by synchronously analyzing current data and motor speed, it can accurately determine whether it is a normal voltage drop caused by battery capacity decay or an abnormally high current discharge caused by motor stalling. This multi-source data fusion mechanism significantly improves the accuracy of fault diagnosis and provides a reliable basis for the generation of subsequent maintenance plans.

[0059] This application further proposes a database including a knowledge base and a 3D model library of drones. The knowledge base is constructed by using AI large model technology to collect technical documents, maintenance manuals and fault case data of drones of different brands and models, and the data is cleaned, classified and labeled before being input into deep learning model training; the 3D model library is constructed by using 3D modeling software to establish accurate 3D models of various types of drones based on design drawings or actual measurement data. The models have full-scale display, zooming in, zooming out and rotation functions.

[0060] Among them, the knowledge base refers to a structured knowledge system formed by integrating multi-source data. Specifically, natural language processing technology can be used to perform semantic analysis on technical documents, convolutional neural networks can be used to extract features from fault cases, data cleaning can be used to eliminate duplicate or erroneous information, and standardized data sets can be formed after classification and annotation for training deep learning models to generate fault diagnosis rules. Its role is to eliminate the defect of incomplete data coverage in a single manual in traditional maintenance. Among them, the 3D model library refers to a digital twin built based on the original design data. Specifically, SolidWorks or AutoCAD software can be used to import drone design drawings, and actual measurement data can be obtained through point cloud scanning to generate a three-dimensional model with a hierarchical structure. The model supports perspective transformation and local zoom operations. Its role is to break through the observation limitations of two-dimensional drawings and realize the visual positioning of hidden fault points.

[0061] Specifically, during the knowledge base construction process, AI large model technology is used to capture public technical documents and maintenance cases. For example, a crawler program is used to obtain PDF manuals from the manufacturer's official website, and OCR technology is used to convert image documents into editable text. Regular expressions are used to filter unstructured data in the cleaning stage, and an index directory is established based on the fault type label in the classification and labeling stage. The trained deep learning model can automatically match the association rules between current fault data and historical cases. When the 3D model library is constructed, the design drawings are disassembled into components such as motors, propellers, and circuit boards, and the assembly relationship of each component is realized through parametric modeling. Maintenance personnel can rotate the model through the touch screen to observe the internal structure and zoom in on specific areas to check the status of seams or solder joints.

[0062] Compared to existing technologies, traditional maintenance relies on manual experience and paper manuals, which can lead to data lags and difficulty understanding 3D structures. This solution integrates dynamically updated multi-brand data through a knowledge base and combines it with a 3D model library to provide a three-dimensional structural display, transforming fault analysis from a single empirical judgment to data-driven intelligent diagnosis. While existing maintenance manuals only provide textual descriptions, this solution's 3D models support interactive disassembly demonstrations, directly locating the spatial position of faulty components.

[0063] Through the above technical solution, this application realizes intelligent data support for drone fault diagnosis. The knowledge base improves the fault matching accuracy through structured data processing, and the 3D model library reduces the difficulty of identifying complex structural faults through visual interaction. The two work together to make the generation of maintenance plans no longer rely on manual experience, while shortening the time for troubleshooting hidden faults.

[0064] This application further proposes that after fault analysis and data comparison, the maintenance decision module generates a drone maintenance plan based on the results of fault analysis and data comparison, combined with the structural characteristics of the 3D model library and the technical documents, maintenance manuals and fault case data in the knowledge base; the content of the drone maintenance plan includes maintenance steps, maintenance tools and parts replacement recommendations.

[0065] Among them, the structural feature of the 3D model library refers to the spatial position relationship data of drone components established through three-dimensional modeling technology. Specifically, it can be achieved by using parametric modeling software to reverse engineer the drone assembly structure. This feature is used to intuitively display the connection relationship between damaged parts and adjacent components when locating faults. The technical documents in the knowledge base refer to structured data containing drone technical parameters and maintenance processes. Specifically, it can be achieved by using natural language processing technology to semantically parse the original maintenance manual and then construct a knowledge graph. This feature provides a standardized operational basis for the generation of maintenance steps. Fault case data refers to the associated data of fault types and solutions accumulated in historical maintenance records. Specifically, it can be achieved by using machine learning algorithms to extract features and match patterns in maintenance logs. This feature provides empirical verification support for component replacement recommendations.

[0066] Specifically, when an abnormal motor speed is detected in a drone, the maintenance decision module first calls the motor installation structure data for the corresponding model in the 3D model library to determine that the disassembly path must avoid the propeller linkage mechanism. It also extracts the standard operating procedures for motor replacement from the knowledge base. Combined with historical case records of bearing wear caused by unstable voltage in the same model motor, it generates a maintenance plan that includes steps for using an insulated wrench, a motor disassembly sequence, and recommendations for compatible model replacements. During the maintenance step generation process, the spatial topology data of the 3D model is converted into robot arm motion trajectory parameters, the torque standards in the technical documentation are quantified into tool operating parameters, and component compatibility data from historical cases is filtered into a spare parts list.

[0067] Compared to existing technologies, traditional methods rely solely on maintenance manuals to guide manual operation, failing to tailor maintenance procedures to the device's actual structural characteristics. For example, patent application number 202210381187.9 only provides static maintenance instructions. This solution, however, leverages collaborative analysis of 3D models and a knowledge base to dynamically generate disassembly and assembly plans tailored to the internal structure of different drones, avoiding repair errors caused by structural differences. While existing technologies rely on maintenance personnel's experience to determine component replacement plans, this solution automatically recommends verified compatible parts by matching historical case data.

[0068] Through the above technical solution, this application achieves intelligent generation of drone maintenance plans. Through multi-dimensional cross-validation of 3D structural data and knowledge bases, it ensures that maintenance steps fully match the actual assembly structure of the equipment, resolving the discrepancy between graphic instructions and physical operation in traditional methods. At the same time, component replacement recommendations based on historical cases effectively avoid the compatibility risks associated with manually selecting replacement parts, significantly improving the operability and reliability of maintenance plans.

[0069] This application further proposes that the drone maintenance plan is executed by an execution module, which includes a multi-axis robotic arm system. The end of the multi-axis robotic arm system is used to be equipped with maintenance tools. The multi-axis robotic arm system plans the motion path and operation steps according to the drone maintenance plan to perform repair or maintenance operations on the drone.

[0070] Among them, the multi-axis robotic arm system refers to a robotic arm structure with multiple rotational joints, which can be specifically implemented by a six-degree-of-freedom robotic arm. The coordinated movement of multiple joints can cover any posture in three-dimensional space. The end-equipped maintenance tool refers to the integration of a replaceable maintenance tool interface on the end effector of the robotic arm. Specifically, it can be implemented by a quick-change fixture system to facilitate the automatic switching of screwdrivers, grippers or detection probes according to the maintenance plan. Motion path planning refers to generating the spatial movement trajectory of the tool at the end of the robotic arm according to the operating steps in the maintenance plan. Specifically, it can be implemented by a path planning algorithm based on inverse kinematics to ensure that the tool moves along the predetermined trajectory. Operation step planning refers to decomposing the maintenance actions in the maintenance plan into a motion sequence of the robotic arm joints. Specifically, it can be implemented by a time-optimal trajectory planning algorithm to optimize the movement speed and acceleration of each joint.

[0071] Specifically, after the maintenance plan is generated, the execution module receives information containing maintenance steps and tool requirements, and the multi-axis robotic arm system automatically loads the corresponding maintenance tools through the quick-change fixture. The path planning algorithm calculates the target position and posture that the end tool of the robotic arm needs to reach based on the positioning data of the drone on the maintenance platform, and generates a collision-free motion trajectory. The operation step planning decomposes the maintenance action into joint motion instructions that are executed sequentially, such as moving to the screw hole position first, and then performing a rotational tightening action. The robotic arm control system drives the motors of each joint according to the planned trajectory and steps, so that the end tool can accurately complete screw replacement, component disassembly or sensor calibration operations. During the maintenance process, the force feedback sensor of the robotic arm monitors the contact force in real time, and automatically pauses and reports an error message when abnormal resistance is detected.

[0072] Compared with existing technologies, traditional drone maintenance relies on manual operation of tools to complete each maintenance step. There are problems such as operating speed being limited by personnel proficiency and low repetition accuracy of complex maintenance actions. However, this solution can reproduce the standard maintenance process with millimeter-level positioning accuracy through the automated execution of a multi-axis robotic arm, eliminating movement deviations caused by human factors. In existing technologies, switching of maintenance tools needs to be done manually. This solution realizes automatic tool replacement through a quick-change fixture system, shortening the process connection time. Existing maintenance platforms lack motion trajectory planning capabilities. This solution generates a three-dimensional space path through an inverse kinematics algorithm, which can adapt to the structural differences of different types of drones.

[0073] Through the above-mentioned technical solution, this application realizes the automation of drone maintenance operations. The robotic arm system accurately completes maintenance actions according to pre-planned paths and steps, solving the problems of low efficiency and insufficient precision of manual operation. The combination of a quick-change fixture system and a path planning algorithm enables a single robotic arm to adapt to a variety of maintenance scenarios, reducing the reliance on the operator's technical expertise. The force feedback monitoring mechanism ensures maintenance accuracy while avoiding component damage, improving the success rate of complex maintenance tasks.

[0074] This application further proposes a multi-type drone maintenance system, including a visual recognition module, a wireless communication module, a data processing module, a maintenance decision module and an execution module.

[0075] The visual recognition module captures images of the drone's exterior using multi-angle image acquisition equipment and analyzes surface fault data using image recognition algorithms. This can be achieved using a high-resolution camera combined with a convolutional neural network algorithm. It automatically identifies surface defects such as fuselage cracks and propeller deformation, replacing manual visual inspection. The wireless communication module establishes a data connection with the drone via a wireless protocol. It can be implemented using Wi-Fi or Bluetooth. This module acquires real-time data on the flight control system's attitude and internal operating parameters such as the powertrain's motor speed, avoiding the limitations of relying solely on external detection. The data processing module analyzes surface fault data alongside internal parameters. This can be achieved using a data fusion algorithm combined with standard parameter thresholds in a database. This allows for a comprehensive assessment of fault type and severity, improving diagnostic comprehensiveness. The maintenance decision module generates maintenance plans based on maintenance cases in a knowledge base and structural features in a 3D model library. This can be achieved using a decision tree algorithm combined with 3D model dynamic simulation to generate customized plans that include repair steps, tools, and component replacement recommendations. Among them, the execution module refers to the module that performs maintenance operations through a multi-axis robotic arm system. Specifically, it can be implemented by using a six-degree-of-freedom robotic arm equipped with a clamping tool or a screwdriver assembly. It is used to complete the disassembly, replacement or maintenance of parts according to the planned path, replacing manual operations.

[0076] Specifically, the visual recognition module collects images of the drone's appearance through a surround camera array, and after image preprocessing, it inputs the trained crack detection model to output the coordinates and size data of the fuselage cracks; the wireless communication module obtains the real-time attitude offset data of the flight control system and the battery voltage anomaly information of the power system by parsing the drone's communication protocol; the data processing module associates the crack data with the voltage anomaly data and determines that the fuselage material is deformed due to battery overheating; the maintenance decision module calls the battery compartment structure model of the corresponding drone model in the 3D model library, and combines the battery replacement operation specifications in the knowledge base to generate maintenance steps for removing the battery cover, replacing the battery and detecting the contact points; the execution module controls the end tool of the robotic arm to complete the screw removal, battery replacement and sealant application operations according to the path planning, forming a closed loop of detection, decision-making and execution.

[0077] Compared with existing technologies, existing solutions rely on maintenance personnel to manually troubleshoot faults based on manual instructions, lack correlation analysis between internal parameters and surface faults, and require manual maintenance. This solution uses dual data collection from vision and communication modules, combined with intelligent decision-making based on a knowledge base and 3D model library, to automate fault location and repair plan generation. Furthermore, a robotic arm execution module reduces manual intervention, addressing the low efficiency, poor accuracy, and high reliance on specialized personnel of traditional methods.

[0078] Through the above technical solution, this application can automatically identify internal and external faults of drones and generate targeted maintenance plans. It can accurately execute maintenance steps through robotic arms to avoid misjudgments or operational errors caused by insufficient human experience. At the same time, it reduces the requirements for the professional skills of maintenance personnel and significantly improves maintenance efficiency and consistency.

[0079] The present application further proposes an electronic device comprising at least one processor and at least one memory; the memory is used to store one or more program instructions; and the processor is used to run one or more program instructions to execute a drone maintenance method.

[0080] The processor is an arithmetic unit capable of executing program instructions, and can be implemented using a multi-core processor or distributed computing architecture. Its role is to coordinate the collaborative work of the visual recognition module, wireless communication module, and maintenance execution module to achieve real-time processing of fault analysis and maintenance decisions. Memory is a hardware unit that stores program instructions and data resources, and can be implemented using solid-state drives or cloud storage media. Its role is to ensure fast access to data such as knowledge bases and 3D model libraries, and to support the stable operation of deep learning models and robotic arm control algorithms.

[0081] Specifically, after program instructions are loaded into the processor, the device drives the drone's external image recognition, collects internal parameters, compares faults against a database, and generates a maintenance plan. Image recognition algorithms extract data on fuselage cracks or propeller deformation. This data, combined with flight control system attitude data or powertrain current parameters acquired through the wireless communication module, is fed into a knowledge base and 3D model library for intelligent analysis. Based on the analysis results, the processor generates a maintenance plan with repair steps and tool selection, and controls the multi-axis robotic arm system to perform the repair. Throughout this process, the device replaces manual diagnosis with automated processing, reducing human judgment errors.

[0082] In some specific embodiments, the processor can be configured to prioritize fault case data in the knowledge base for similarity matching. For example, when a battery voltage anomaly is detected, it can automatically associate it with a historical case study related to circuit board aging. The memory can be divided into multiple storage areas, storing structured data from the knowledge base and rendering files from the 3D model library, respectively, to improve data retrieval efficiency.

[0083] Compared with existing technologies, traditional methods rely on maintenance personnel to manually review technical documentation to diagnose faults, resulting in long diagnostic cycles and a reliance on experience. This solution uses program instructions to drive equipment to automatically complete data collection, intelligent analysis, and repair execution, reducing fault diagnosis time from hours of manual operation to minutes while also eliminating the risk of misdiagnosis due to differences in personnel skills.

[0084] Through the above technical solution, this application achieves a fully automated drone maintenance process, resolving the issues of low efficiency and inaccuracy in manual diagnosis. The device integrates image recognition, parameter acquisition, and mechanical control functions through program instructions, reducing reliance on professional expertise and ensuring the scientific nature and precision of maintenance plans.

[0085] The present application further proposes that a computer-readable storage medium contains one or more program instructions, and the program instructions are used to execute a multi-type drone maintenance method.

[0086] Computer-readable storage media refers to the physical medium used to persistently store program code, specifically solid-state drives or flash memory chips. Their data storage properties ensure the stability and reusability of program instructions. Program instructions refer to a collection of codes that can be parsed and executed by an electronic device processor. They can be written in machine language or a high-level programming language and compiled into a binary executable file. Their structured nature enables standardized packaging of maintenance method flows.

[0087] Specifically, the program instructions are configured to control the electronic device to sequentially execute appearance image acquisition, communication connection establishment, data transmission and analysis, maintenance plan generation, and execution operations. When the storage medium is loaded into the electronic device, the processor reads the program instructions to drive the visual recognition module to automatically capture multi-angle appearance images and calls the image recognition algorithm to extract fuselage crack data. At the same time, it triggers the wireless communication module to establish a data link with the drone to obtain the attitude data of the flight control system and the motor speed parameters of the power system in real time. The collected fault data and operating parameters are transmitted to the database, and by comparing the standard models in the knowledge base with historical fault cases, a maintenance plan containing component replacement recommendations is generated. Finally, the program instructions control the multi-axis robotic arm to complete the propeller replacement operation according to the preset path.

[0088] Compared to existing technologies, traditional maintenance methods rely on manual review of paper manuals to determine the fault type, which is subject to subjective judgment errors and operational delays. This solution, however, uses program-driven equipment to automatically complete the entire process of data collection, intelligent diagnosis, and mechanical maintenance, eliminating manual intervention. The physical carrier characteristics of the storage medium allow maintenance methods to be replicated across different devices, ensuring standardized maintenance operations.

[0089] Through the above technical solution, this application achieves fully automated operation of the drone fault diagnosis and repair process. The fixed execution mechanism of program instructions avoids fluctuations in repair quality caused by differences in manual experience, and the precise operation of the robotic arm reduces the risk of secondary damage during component replacement. Real-time data access from the knowledge base and 3D model library shortens fault analysis time, and the collaborative operation of multiple modules significantly improves repair efficiency.

[0090] In this document, the directions or positional relationships indicated by terms such as "up", "down", "front", "back", "left", "right", "top", "bottom", "inside", "outside", "vertical", and "horizontal" are based on the directions or positional relationships shown in the accompanying drawings and are only for the clarity of the technical solution and the convenience of description, and therefore should not be understood as limiting the present invention.

[0091] As used herein, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion of elements other than the listed elements and may also include additional elements not specifically listed.

[0092] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-type UAV maintenance method, characterized in that: The steps include: S1. Obtain surface fault data of the drone by identifying its appearance image; S2. Establish a communication connection with the drone to obtain the drone's internal parameters; S3. Send the drone's surface fault data and internal parameters to a pre-set database for fault analysis and data comparison, generating a drone maintenance plan. S4. Perform repair or maintenance operations on the drone according to the drone maintenance plan.

2. The multi-type UAV maintenance method according to claim 1, characterized in that: The surface fault data is obtained by identifying the appearance image of the UAV through an image recognition algorithm, and the surface fault data includes fuselage cracks, fuselage wear and propeller deformation data.

3. The multi-type UAV maintenance method according to claim 1, characterized in that: The internal parameters of the drone are obtained through a wireless communication module that can establish a communication connection with the drone. The internal parameters include flight control system parameters and power system parameters.

4. The multi-type UAV maintenance method according to claim 3, characterized in that: The flight control system parameters include attitude data, flight speed and flight altitude; the power system parameters include motor speed, battery voltage and current.

5. The multi-type UAV maintenance method according to claim 1, characterized in that: The database includes a knowledge base and a 3D model library of drones; The method for constructing the knowledge base is as follows: First, AI big model technology is used to collect technical documents, maintenance manuals, and fault case data for different brands and models of drones. The collected data is then cleaned, classified, and labeled, and input into a deep learning model for training to build a drone knowledge base. The method for constructing the 3D model library is as follows: Using 3D modeling software, based on the drone's design drawings or actual measurement data, an accurate 3D model is created for each type of drone to form a 3D model library; the 3D model has full-scale display, zoom in, zoom out and rotation functions, making it easy to view details during fault analysis.

6. The multi-type UAV maintenance method according to claim 5, characterized in that: After the fault analysis and data comparison, the maintenance decision module generates a drone maintenance plan based on the results of the fault analysis and data comparison, combined with the structural characteristics of the 3D model library and the technical documents, maintenance manuals and fault case data in the knowledge base; the content of the drone maintenance plan includes maintenance steps, maintenance tools and parts replacement suggestions.

7. The multi-type UAV maintenance method according to claim 1, characterized in that: The drone maintenance plan is executed by an execution module, which includes a multi-axis robotic arm system. The end of the multi-axis robotic arm system is used to be equipped with maintenance tools. The multi-axis robotic arm system plans the motion path and operation steps according to the drone maintenance plan to perform repair or maintenance operations on the drone.

8. A multi-type UAV maintenance system, characterized in that: include The visual recognition module is used to obtain the appearance images of the UAV from different angles and analyze the appearance images through image recognition algorithms to obtain the surface fault data of the UAV; Wireless communication module, used to obtain the internal parameters of the drone by establishing communication with the drone; The data processing module is used to perform fault analysis and data comparison on the surface fault data and internal parameters of the UAV with the standard data stored in the database; The maintenance decision module is used to generate a maintenance plan for the drone based on the results of fault analysis and data comparison, combined with the drone's knowledge base and 3D model library; The execution module is used to plan the motion path and operation steps according to the drone maintenance plan and perform repair or maintenance operations on the drone.

9. An electronic device, characterized in that: include: at least one processor and at least one memory; The memory is used to store one or more program instructions; The processor is configured to run one or more program instructions to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium includes one or more program instructions, and the one or more program instructions are used to execute the method according to any one of claims 1 to 7.

Images