Undercarriage monitoring system and method based on AI visual perception

Abstract

The invention discloses an undercarriage monitoring system and method based on AI visual perception. The system comprises a visual acquisition module used for collecting image data of an aircraft landing gear folding and unfolding mechanism in real time; the image preprocessing module is used for preprocessing the collected image data, and preprocessing comprises denoising, enhancing and segmenting the image; the visual perception module is internally provided with a trained visual perception AI model, and position information of key points and motion characteristics of the undercarriage are recognized in real time through images; the health assessment module is internally provided with a trained health assessment AI model and is used for analyzing the real-time motion state of the undercarriage according to the extracted motion characteristics and judging the health state of the undercarriage; and the data feedback module is used for feeding back the health state information to the flight crew in real time and storing related data for maintenance and analysis.

Description

Technical Field

[0001] This invention relates to the field of aircraft landing gear status monitoring technology, and in particular to a real-time landing gear monitoring system and method based on artificial intelligence visual perception. Background Technology

[0002] The aircraft landing gear system is a critical functional system ensuring the safety of aircraft takeoff and landing. Statistics show that approximately 90% of air crashes occur during takeoff or landing, and landing gear system failures account for a significant proportion of all aircraft malfunctions, with some failures potentially leading to serious safety incidents. For example, on May 9, 2024, a Boeing 767 freighter made an emergency belly landing because its nose landing gear failed to deploy safely; on December 29 of the same year, a Boeing 737-800 passenger plane crashed due to factors including its landing gear failing to deploy, resulting in significant casualties.

[0003] Currently, most civil aircraft landing gear monitoring systems employ a solution consisting of proximity sensors, a control unit, and a display unit. The sensors determine the landing gear's position by detecting its proximity and transmit the signal to the control unit for further analysis and alerts. However, this solution has the following drawbacks: the sensors are susceptible to interference from environmental factors such as temperature, humidity, dust accumulation, and vibration, leading to data distortion or system malfunction; it only provides a binary status indicator of whether the landing gear is in place, failing to monitor the movement process in real time; it lacks fault warning and health assessment capabilities, requiring post-incident troubleshooting for maintenance, resulting in low efficiency; and the system has low redundancy, failing to provide backup monitoring methods when sensors fail.

[0004] With the development of artificial intelligence and computer vision technologies, AI visual perception has been widely used in fields such as industrial inspection and medical imaging. Its high precision, real-time performance, and non-contact monitoring characteristics provide a new technological path for landing gear monitoring. Therefore, it is necessary to develop an intelligent landing gear monitoring system that combines AI visual perception to improve monitoring reliability, maintenance efficiency, and flight safety. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a real-time landing gear monitoring solution based on artificial intelligence visual perception.

[0006] According to one aspect of the present invention, an AI-based visual perception landing gear monitoring system is provided. The system includes: a visual acquisition module for real-time acquisition of image data of an aircraft landing gear retraction mechanism; an image preprocessing module for preprocessing the acquired image data, wherein the preprocessing includes denoising, enhancement, and segmentation of the image; a visual perception module with a built-in trained visual perception AI model for real-time identification of key point location information and landing gear motion characteristics through images; a health assessment module with a built-in trained health assessment AI model for analyzing the real-time motion state of the landing gear based on extracted motion characteristics and determining the landing gear health status; and a data feedback module for real-time feedback of health status information to the flight crew and storing relevant data for maintenance analysis.

[0007] According to a further embodiment of the present invention, the vision acquisition module includes at least one camera deployed in the landing gear bay, the field of view of which covers at least three key points on the tires, support rods and actuation mechanisms during the landing gear retraction and extension process.

[0008] According to a further embodiment of the present invention, the camera is a non-contact camera and is equipped with a light source and protective structure adapted to the cabin environment.

[0009] According to a further embodiment of the present invention, the preprocessing performed by the image preprocessing module includes at least: denoising, enhancing, and segmenting the image, and normalizing the image data to eliminate the influence of dimensions.

[0010] According to a further embodiment of the present invention, the health assessment AI model is a convolutional neural network, which is trained on a labeled image dataset containing normal and various abnormal landing gear motion states, and is able to identify at least one fault mode among mechanical jamming, trajectory deviation, and abnormal vibration.

[0011] According to a further embodiment of the present invention, the health assessment AI model adopts the Transformer architecture, and its input is the landing gear key point time series and the actuator motion command time series. The health assessment AI model is trained by including the landing gear key point time series, actuator motion command time series and corresponding state labels under normal and various abnormal states, and can identify at least one fault mode among mechanical jamming, trajectory deviation and abnormal vibration.

[0012] According to a further embodiment of the present invention, the landing gear monitoring system based on AI visual perception is set up in parallel with the aircraft's original proximity sensor-based monitoring system to form a dual-redundant monitoring channel.

[0013] According to a further embodiment of the present invention, the landing gear monitoring system provides landing gear status information to the landing gear controller, which compares the landing gear status information provided by the landing gear monitoring system with the landing gear status information provided by the proximity sensor, wherein when the proximity sensor fails, the landing gear status information provided by the landing gear monitoring system shall prevail.

[0014] According to another aspect of the present invention, a landing gear monitoring method based on AI visual perception is provided. The method includes: acquiring image data of the landing gear retraction and extension process in real time through a camera; preprocessing the acquired image data and extracting motion features of key landing gear components through a visual perception AI model; inputting the extracted motion features into a trained health assessment AI model, which analyzes the real-time motion state of the landing gear and determines the health status of the landing gear; feeding back the health status information to the flight crew in real time and storing the relevant data for maintenance analysis.

[0015] According to a further embodiment of the present invention, the health assessment AI model adopts the Transformer architecture, and its input is the landing gear key point time series and the actuator motion command time series. The health assessment AI model is trained by including the landing gear key point time series, actuator motion command time series and corresponding state labels under normal and various abnormal states, and can identify at least one fault mode among mechanical jamming, trajectory deviation and abnormal vibration.

[0016] Compared with existing technologies, the landing gear monitoring solution based on AI visual perception provided by this invention has at least the following advantages: 1. It achieves continuous visual monitoring of the entire process from folding up and locking to putting down and touching the ground, breaking through the limitations of traditional "point" detection.

[0017] 2. High precision and non-contact: It adopts a visual perception method, avoiding direct contact between physical sensors and mechanical structures and the installation stress, wear and interference problems caused by it. It has high monitoring accuracy and good reliability.

[0018] 3. Intelligent early warning and diagnosis: Based on the AI ​​model, it can not only judge the abnormal status, but also perform preliminary diagnosis and location of fault types (such as jamming, loosening, deformation), supporting predictive maintenance.

[0019] 4. Enhanced system redundancy: As a new monitoring channel independent of the traditional sensor system, it forms a dual guarantee with the original system, significantly improving the overall safety margin of landing gear monitoring.

[0020] 5. Improve operational efficiency: Detailed image data and health history records provide ground staff with intuitive and accurate troubleshooting information, which can significantly shorten maintenance time and reduce operating costs.

[0021] These and other features and advantages will become apparent from the following detailed description and with reference to the accompanying drawings. It should be understood that the foregoing general description and the following detailed description are illustrative only and do not limit the scope of the claims. Attached Figure Description

[0022] Figure 1 This is a schematic block diagram of a landing gear monitoring system based on AI visual perception according to an embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram of the system architecture for integrating an AI landing gear monitoring system with a conventional landing gear control system according to an embodiment of the present invention.

[0024] Figure 3 This is a schematic diagram illustrating the camera's field of view and key components of the landing gear in one embodiment of the present invention.

[0025] Figure 4 This is a schematic diagram of the data processing and AI model training process according to an embodiment of the present invention.

[0026] Figure 5 This is a flowchart of a landing gear monitoring method based on AI visual perception according to an embodiment of the present invention. Detailed Implementation

[0027] The present invention will now be described in detail with reference to the accompanying drawings, and its features will become further apparent from the following specific description. In this detailed description, numerous specific details are set forth to provide a thorough understanding of the exemplary embodiments described. However, it will be apparent to those skilled in the art that the described embodiments may be practiced without some or all of these specific details. In other exemplary embodiments, well-known structures or processing steps have not been described in detail to avoid unnecessarily obscuring the concepts of this disclosure.

[0028] Existing civil aircraft landing gear monitoring systems generally consist of proximity sensors, control units, display units, and power supplies. When the aircraft landing gear is being extended or retracted, the proximity sensors monitor the landing gear's proximity in real time and transmit relevant signals to the control unit. After processing these signals, the control unit determines whether the landing gear has been correctly extended or retracted. If the landing gear is not fully extended or retracted, the control unit will issue a warning signal, and the display unit will simultaneously display the corresponding status information to the pilot or ground personnel.

[0029] However, sensors are susceptible to environmental interference, such as changes in temperature and humidity, sensor aging, dust accumulation, and vibration, leading to inaccurate data or system malfunctions. Once a malfunction occurs, troubleshooting is difficult, especially during flight and landing, when ground crew and the flight crew cannot obtain the true status of the landing gear in a timely manner. Furthermore, the system can only provide basic landing gear status information and cannot provide fault warnings or proactive maintenance. To optimize the maintainability, control, and energy efficiency of existing technologies, and improve operational economy and product competitiveness, this invention proposes a landing gear monitoring system for future intelligent civil aircraft that overcomes the shortcomings of difficult maintenance and troubleshooting and the lack of fault warnings.

[0030] Figure 1 A schematic block diagram of an AI-based visual perception landing gear monitoring system 100 according to an embodiment of the present invention is shown. Although Figure 1 The boxes in the diagram are interpreted as different components, but the functions described above for these boxes can be implemented by a single hardware, software, or combination of components or a combination of various components. They can be mounted on the same or different airborne systems, implemented by an integrated module controller, or concentrated in a specific functional module.

[0031] like Figure 1 As shown, in some aspects, the landing gear monitoring system 100 may include a vision acquisition module 110, which can be configured to acquire image data of the aircraft landing gear retraction mechanism in real time. In one example, the vision acquisition module 110 may include at least one camera deployed in the landing gear bay, the camera's field of view covering at least three key points on the tires, support rods, and actuation mechanisms during landing gear retraction. In another example, the vision acquisition module 110 may be a non-contact camera equipped with a light source and protective structure adapted to the cabin environment. As a non-limiting example, without affecting all functions of the aircraft landing gear, a perception camera may be installed on the upper side of the landing gear bay (including the nose and main landing gear), and combined with AI visual perception technology, to monitor the movement status of key landing gear points in real time.

[0032] like Figure 1As shown, in some aspects, the landing gear monitoring system 100 may include an image preprocessing module 120, which can be configured to preprocess acquired image data, wherein the preprocessing includes denoising, enhancement, and segmentation of the image. In one example, the key components of the landing gear may be composed of tires, support rods, and actuation mechanisms. In another example, the preprocessing performed by the image preprocessing module 120 includes at least: denoising, enhancing, and segmenting the image, and normalizing the image data to eliminate the influence of dimensions. As a non-limiting example, a deep learning model can be used to perform target detection and segmentation of the key components of the landing gear, which may include the following steps: image annotation (key point annotation), dataset partitioning (training set and test set), model training and optimization (merging categories, cropping edges, data augmentation).

[0033] like Figure 1 As shown, in some aspects, the landing gear monitoring system 100 may include a vision perception module 130, which has a built-in trained vision perception AI model and can be configured to identify the location information of key points and the motion characteristics of the landing gear in real time through images. In one example, the vision perception AI model is a convolutional neural network, which is trained on a labeled image dataset containing key point location information and is capable of identifying key point locations and landing gear motion state information.

[0034] like Figure 1 As shown, in some aspects, the landing gear monitoring system 100 may include a health assessment module 140, which has a built-in trained health assessment AI model and can be configured to analyze the real-time motion state of the landing gear based on extracted motion features to determine the health status of the landing gear. In one example, the health assessment AI model adopts a Transformer architecture, and its inputs are the landing gear key point time series and the actuator motion command time series. The health assessment AI model is trained by including the landing gear key point time series, actuator motion command time series, and corresponding state labels under normal and various abnormal states, and can identify at least one fault mode among mechanical jamming, trajectory deviation, and abnormal vibration. As a non-limiting example, the health assessment AI model can be used to fit the actual motion of the landing gear and the actuator commands to accurately determine whether there is a fault in the landing gear. In another example, the landing gear monitoring system 100 may provide landing gear status information to the landing gear controller (e.g., via ARINC429 communication), which compares the landing gear status information provided by the landing gear monitoring system 100 with the landing gear status information provided by the proximity sensor, wherein when the proximity sensor fails, the landing gear status information provided by the landing gear monitoring system 100 shall prevail.

[0035] like Figure 1As shown, in some aspects, the landing gear monitoring system 100 may include a data feedback module 150, which can be configured to provide real-time health status information to the flight crew and store relevant data for maintenance analysis. In one example, monitoring images and fault information can be fed back to the pilot through a human-machine interface, ensuring that flight operators are promptly informed of the equipment status, and automatically recording and storing relevant monitoring data for subsequent analysis and fault tracing.

[0036] Figure 2 A schematic diagram of a system architecture 200 integrating an AI landing gear monitoring system with a conventional landing gear control system according to an embodiment of the present invention is shown. Figure 2 As shown, the AI ​​landing gear monitoring system includes non-contact cameras, an eagle-eye monitoring system, and monitoring and fault information storage, operating independently of the traditional landing gear control system. The monitoring results of both systems can be compared and verified. When a traditional sensor falsely reports a fault, the real-time image provided by the AI ​​landing gear monitoring system can serve as the final decision-making basis; conversely, the AI ​​landing gear monitoring system can also utilize command signals from traditional sensors to assist in analysis. For example, the landing gear controller can receive landing gear status information from the AI ​​landing gear monitoring system (e.g., via ARINC429 communication) and landing gear status information from traditional proximity sensors; the landing gear controller compares these two sets of information, and when a proximity sensor malfunctions, the status information from the AI ​​landing gear monitoring system takes precedence. This design achieves functional redundancy, greatly improving the reliability of the overall monitoring system.

[0037] Figure 3 The image shows the motion trajectory of the landing gear retraction mechanism within the effective field of view captured by a non-contact camera. The dashed line 301 represents the effective field of view; 311, 312, and 313 are three key points centrally labeled in the data set; 321 represents the tire, 322 the aircraft landing gear, and 323 the support rod; 331 represents the non-contact camera and light source. Figure 3 As shown, a high-resolution industrial camera 331 is installed in a location that is not easily obstructed and has a good field of view, located above the interior of the landing gear bay. This camera must be vibration-resistant, waterproof, resistant to high and low temperatures, and resistant to electromagnetic interference, and be equipped with a supplementary light to ensure image quality in the dimly lit environment inside the bay.

[0038] Figure 4 A schematic diagram illustrating the data processing and AI model training process 400 according to an embodiment of the present invention is shown. Figure 4As shown, during system operation, the camera continuously acquires image and video data. The data processing module first performs target detection on the images, selecting key components such as tires, landing gear, and support rods, and annotating key points to extract geometric features. Subsequently, data augmentation is performed using methods such as rotation, scaling, and adding noise to increase sample diversity. Next, the image data is normalized, for example, by calculating the standard deviation using the following formula:

[0039] Normalization is achieved using the following formula to eliminate dimensional differences:

[0040] Finally, the dataset is divided into a training set, a validation set, and a test set, which are used for model training, parameter tuning, and performance evaluation, respectively.

[0041] Figure 5 A flowchart of a landing gear monitoring method 500 based on AI visual perception according to an embodiment of the present invention is shown. Method 500 is a landing gear monitoring system (e.g., as described above). Figure 1 and Figure 2 An example of a landing gear monitoring system (in the context of a landing gear monitoring system) implementing an AI-based visual perception-based landing gear monitoring method.

[0042] like Figure 5 As shown, in some aspects, method 500 may include: acquiring image data of the landing gear retraction and extension process in real time via a camera (box 510). In one example, at least one camera may be deployed in the landing gear bay, the camera's field of view covering at least three key points on the tires, support rods, and actuation mechanisms during the landing gear retraction and extension process.

[0043] like Figure 5 As further shown, in some aspects, method 500 may include: preprocessing the acquired image data and extracting motion features of key landing gear components using a visual perception AI model (box 520). In one example, preprocessing may include: denoising, enhancing, and segmenting the image, and normalizing the image data to eliminate dimensional effects. In another example, the key landing gear components may be composed of tires, support rods, and actuation mechanisms.

[0044] like Figure 5As further shown, in some aspects, method 500 may include: inputting the extracted motion features into a trained health assessment AI model, which analyzes the real-time motion state of the landing gear to determine the health status of the landing gear (box 530). In one example, the health assessment AI model adopts a Transformer architecture, and its input is the landing gear key point time series and actuator motion command time series; the health assessment AI model is trained by including landing gear key point time series, actuator motion command time series and corresponding state labels under normal and various abnormal states, and can identify at least one fault mode among mechanical jamming, trajectory deviation, and abnormal vibration. As a non-limiting example, during training, a large number of labeled normal and abnormal landing gear motion image sequences are used as a dataset, and the model learns to identify the motion patterns of key points from the image sequences; after deployment, the model receives the processed image sequences in real time, and determines whether the current motion trajectory deviates from the normal mode by analyzing the displacement and velocity changes of key points; if jamming (key points have no displacement for a long time), abnormal trajectory (key point motion path does not match the command), or excessive vibration is detected, it is determined to be a fault.

[0045] like Figure 5 As further shown, in some aspects, method 500 may include: providing real-time feedback of health status information to the flight crew and storing relevant data for maintenance analysis (box 540). In one example, health status (such as "normal," "warning," "fault") and specific diagnostic information derived from a health assessment AI model are output through a feedback control module. In the cockpit, the information can be presented to the pilot in graphical (such as real-time video with overlaid trajectory lines) and textual form via a dedicated display or integrated into existing flight displays. Simultaneously, this information is transmitted to the flight data recorder and maintenance computer via an onboard bus such as ARINC 429. All monitoring data, including raw images, analysis results, and timestamps, is packaged and automatically transmitted to the airline's ground data center cloud platform after landing or during flight via the aircraft communications addressing and reporting system or a ground wireless network for remote analysis and maintenance planning by maintenance personnel.

[0046] This invention utilizes a high-precision non-contact camera to capture real-time images of the landing gear's motion mechanism. Combined with AI visual perception algorithms, it monitors multi-dimensional data such as the landing gear's trajectory, speed, and angle, promptly identifying potential anomalies during movement, such as mechanical jamming, excessive vibration, or deviation from the normal trajectory. This vision-based monitoring method is more flexible and has a wider coverage than traditional physical sensors. A large-scale AI visual perception model tracks, records, and intelligently analyzes the landing gear's motion status in real time, and uses a health assessment AI model to determine the landing gear's health status, avoiding the blind spots of traditional proximity sensors and providing more accurate health assessments. The large-scale AI visual perception model can then provide real-time feedback of health monitoring information to the flight crew through data analysis. The landing gear health diagnostic information provided by the system helps the crew understand the landing gear's operating status in a timely manner, identify and locate potential faults in advance, improve flight safety and maintenance efficiency, and implement the concept of applying artificial intelligence in the civil aviation field. The landing gear monitoring system adopts a dual-monitoring model, primarily using AI visual perception and supplemented by traditional functional methods, improving the coverage and accuracy of monitoring, as well as the overall reliability and ease of maintenance of the landing gear system.

[0047] The various steps and modules of the methods, apparatus, and systems described above can be implemented in hardware, software, or a combination thereof. If implemented in hardware, the various illustrative steps, modules, and circuits described in connection with this disclosure can be implemented or executed using a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic components, hardware components, or any combination thereof. A general-purpose processor can be a processor, microprocessor, controller, microcontroller, or state machine, etc. If implemented in software, the various illustrative steps and modules described in connection with this disclosure can be stored as one or more instructions or codes on a computer-readable medium or transmitted. Software modules implementing the various operations of this disclosure can reside in storage media such as RAM, flash memory, ROM, EPROM, EEPROM, registers, hard disks, removable disks, CD-ROMs, cloud storage, etc. The storage medium can be coupled to a processor so that the processor can read and write information from / to the storage medium and execute corresponding program modules to implement the various steps of this disclosure. Moreover, software-based embodiments can be uploaded, downloaded, or remotely accessed through appropriate communication means. Such appropriate means of communication include, for example, the Internet, the World Wide Web, intranets, software applications, cables (including fiber optic cables), magnetic communication, electromagnetic communication (including RF, microwave and infrared communication), electronic communication, or other such means of communication.

[0048] The numerical values ​​given in the various embodiments are merely examples and are not intended to limit the scope of the invention. Furthermore, as a whole, there are other components or steps not listed in the claims or specification of this invention. Moreover, a single name for a component does not preclude other names for that component.

[0049] It should also be noted that these embodiments may be described as processes depicted as flowcharts, flow diagrams, structure diagrams, or block diagrams. Although a flowchart may describe the operations as sequential processes, many of these operations can be executed in parallel or concurrently. Furthermore, the order of these operations can be rearranged.

[0050] The disclosed methods, apparatuses, and systems should not be limited in any way. Rather, this disclosure covers all novel and non-obvious features and aspects of the various disclosed embodiments (individually and in various combinations and sub-combinations of each other). The disclosed methods, apparatuses, and systems are not limited to any particular aspect or feature or combination thereof, and no disclosed embodiment is required to have any one or more specific advantages or to solve any particular or all technical problems.

[0051] This invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other modifications based on the teachings of this invention without departing from the spirit and scope of the claims. All of these modifications are within the scope of protection of this invention.

Claims

1. A landing gear monitoring system based on AI visual perception, characterized in that, include: The vision acquisition module is used to acquire image data of the aircraft landing gear retraction mechanism in real time; The image preprocessing module is used to preprocess the acquired image data, wherein the preprocessing includes denoising, enhancement, and segmentation of the image; The visual perception module has a built-in trained visual perception AI model, which is used to identify the location information of key points and the motion characteristics of the landing gear in real time through images. The health assessment module has a built-in trained health assessment AI model, which is used to analyze the real-time motion status of the landing gear based on the extracted motion features and determine the health status of the landing gear. The data feedback module is used to provide real-time health status information to the flight crew and store relevant data for maintenance and analysis.

2. The landing gear monitoring system according to claim 1, characterized in that, The vision acquisition module includes at least one camera deployed in the landing gear bay, and the camera's field of view covers at least three key points on the tires, support rods, and actuation mechanisms during the landing gear retraction and extension process.

3. The landing gear monitoring system according to claim 2, characterized in that, The camera is a non-contact camera and is equipped with a light source and protective structure adapted to the cabin environment.

4. The landing gear monitoring system according to claim 1, characterized in that, The image preprocessing module performs at least the following preprocessing steps: denoising, enhancing, and segmenting the image, and normalizing the image data to eliminate the influence of dimensions.

5. The landing gear monitoring system according to claim 1, characterized in that, The visual perception AI model is a convolutional neural network, which is trained on a labeled image dataset containing key point location information and is able to identify key point locations and landing gear motion status information.

6. The landing gear monitoring system according to claim 1, characterized in that, The health assessment AI model adopts the Transformer architecture, and its inputs are the landing gear key point time series and the actuator motion command time series. The health assessment AI model is trained by including the landing gear key point time series, actuator motion command time series and corresponding state labels under normal and various abnormal states, and can identify at least one fault mode among mechanical jamming, trajectory deviation and abnormal vibration.

7. The landing gear monitoring system according to any one of claims 1 to 6, characterized in that, The landing gear monitoring system is set up in parallel with the aircraft's original proximity sensor-based monitoring system, together forming a dual-redundant monitoring channel.

8. The landing gear monitoring system according to claim 7, characterized in that, The landing gear monitoring system provides landing gear status information to the landing gear controller. The landing gear controller compares the landing gear status information provided by the landing gear monitoring system with the landing gear status information provided by the proximity sensor. When the proximity sensor malfunctions, the landing gear status information provided by the landing gear monitoring system shall prevail.

9. A landing gear monitoring method based on AI visual perception, characterized in that, include: The landing gear retraction and extension process is captured in real time by a camera; The collected image data is preprocessed, and the motion characteristics of key landing gear components are extracted using a visual perception AI model. The extracted motion features are input into a trained health assessment AI model, which analyzes the real-time motion state of the landing gear and determines the health status of the landing gear. Health status information is fed back to the flight crew in real time, and relevant data is stored for maintenance and analysis.

10. The method according to claim 9, characterized in that, The health assessment AI model adopts the Transformer architecture, and its inputs are the landing gear key point time series and the actuator motion command time series. The health assessment AI model is trained by including the landing gear key point time series, actuator motion command time series and corresponding state labels under normal and various abnormal states, and can identify at least one fault mode among mechanical jamming, trajectory deviation and abnormal vibration.

Images