Unmanned aerial vehicle real-time abnormal state detection and prediction method, equipment, medium and product

Abstract

The invention provides an unmanned aerial vehicle real-time abnormal state detection and prediction method and device, a medium and a product, and the method comprises the steps: dynamically generating a normal flight state digital model based on historical successful flight data through employing a flight data recording and playback system; acquiring real-time data streams by using a flight data recording and playback system, and matching the real-time data streams based on the normal flight state digital model; and completing abnormal state detection and prediction according to a matching result. According to the invention, the capability of a flight data recording and playback system is expanded from offline post playback to online real-time accompanying flight, and the sensitivity and accuracy of anomaly detection can be significantly improved. Moreover, weak anomalies can be captured through dynamic generation of the model and multivariate correlation analysis, and the method has practical application value and can be widely applied to unmanned aerial vehicle scenes such as military reconnaissance, agricultural monitoring or logistics distribution. Meanwhile, real-time recording and playback functions of a flight data recording and playback system are reserved, and data integrity and system compatibility are ensured.

Description

Methods, equipment, media and products for real-time abnormal state detection and prediction of unmanned aerial vehicles (UAVs) Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and more specifically, to a method, device, medium, and product for real-time abnormal state detection and prediction of UAVs. This method is applicable to multi-rotor UAVs, fixed-wing UAVs, or other unmanned aerial vehicle systems. Background Technology

[0002] Existing UAV systems primarily rely on two methods for anomaly detection: one is real-time sensor alarms based on preset thresholds, such as triggering an alarm when the engine temperature exceeds a fixed upper limit; the other is post-flight offline data analysis, using flight logs to review potential problems. Existing flight data recording and playback systems (such as tools known in the industry as "time capsules") typically function as offline data processing tools. Their core functions include real-time recording of sensor data and environmental parameters generated by the UAV and ground command and control station during mission execution, writing them to files in specific file formats (such as binary or custom structured formats), and reconstructing the real-time state from these files for playback analysis afterward. However, their application is currently limited to after the flight mission has ended and cannot provide online real-time support for flight data recording and playback systems.

[0003] For example, in the prior art, patent CN119045515A discloses a UAV flight status monitoring and anomaly handling system, including a signal connection establishment module and a flight status monitoring module, used to determine and handle anomalies, but it does not involve real-time matching and prediction of templates dynamically generated based on historical data. Patent CN107145158A discloses a rotorcraft UAV abnormal landing handling device, which activates a buffer by judging abnormal conditions, but lacks early detection and prediction of weak anomalies. Patent WO2021244544A1 discloses a UAV fault detection method, which judges faults through flight status information, but does not extend the software to online real-time flight escort.

[0004] These existing technologies have significant shortcomings: threshold alarm methods are too simplistic and fail to capture the complex interrelationships between parameters, leading to the neglect of many subtle anomalies (such as parameter combination deviations), potentially causing delays and failures; while post-flight analysis can utilize historical big data, it lacks real-time capability and cannot provide decision-making guidance during flight. Existing recording and playback systems, although powerful, are limited to offline mode and cannot directly intervene in anomaly detection during missions. With the increasing complexity of UAV missions (such as long-distance cruises and reconnaissance in complex environments), these problems are becoming increasingly prominent, potentially leading to mission failures, safety incidents, or equipment damage. For example, during high-altitude cruises, when a single parameter (such as engine speed) is within the normal range, but its correlation with fuel consumption rate deviates from historical patterns, existing methods struggle to predict potential power system failures in advance. Summary of the Invention

[0005] This invention aims to provide a method, device, medium, and product for real-time abnormal state detection and prediction of unmanned aerial vehicles (UAVs), thereby extending the powerful data processing capabilities of existing flight data recording and playback systems from offline post-event playback to online real-time flight escort, realizing real-time abnormal detection and prediction based on historical data models, and improving the robustness and intelligence level of UAV systems.

[0006] In a first aspect, the present invention provides a method for real-time abnormal state detection and prediction of unmanned aerial vehicles (UAVs), comprising: using a flight data recording and playback system to dynamically generate a digital template of normal flight state based on historical successful flight data; using the flight data recording and playback system to acquire real-time data streams, and matching the real-time data streams based on the digital template of normal flight state; and completing abnormal state detection and prediction based on the matching results.

[0007] In a preferred embodiment, the step of dynamically generating a digital template of normal flight status based on historical successful flight data using the flight data recording and playback system includes: processing historical successful flight data using the offline data processing capabilities of the flight data recording and playback system; writing the historical successful flight data into a data recording file and storing it according to the required file format through the real-time recording mechanism of the flight data recording and playback system; extracting parameters from the historical successful flight data and dynamically generating a digital template of normal flight status based on the mission profile.

[0008] In a preferred embodiment, the step of extracting parameters from historical successful flight data and dynamically generating a digital template of normal flight status based on the mission profile includes: using the restoration function of the flight data recording and playback system to read historical successful flight data from the data recording file; using principal component analysis to reduce the dimensionality of the read historical successful flight data and extract parameters; for the extracted parameters, using statistical methods to calculate the mean μ and standard deviation σ, and forming a confidence interval [μ-3σ, μ+3σ], adjusting the normal fluctuation range of each parameter based on the confidence interval and the mission profile; for the extracted parameters, applying a multidimensional parameter correlation analysis method to obtain the parameter correlation relationship; the multidimensional parameter correlation analysis method includes Pearson correlation coefficient, multiple regression model, Gaussian mixture model, or neural network.

[0009] The generated normal flight state digital template defines the normal fluctuation range of individual parameters and parameter correlations for each stage of the mission profile. Cross-validation is used to ensure the generalization ability of the normal flight state digital template.

[0010] In a preferred embodiment, the step of acquiring real-time data streams using a flight data recording and playback system and matching the real-time data streams based on a normal flight state digital template includes: when the UAV is performing a mission, the flight data recording and playback system acquires real-time data streams through bypass access, samples the real-time data streams to obtain real-time data of parameters; based on the real-time data of parameters, a matching algorithm is used to calculate the deviation of each parameter from the normal flight state digital template.

[0011] In a preferred embodiment, the step of using a matching algorithm to calculate the deviation of each parameter from the normal flight state digital template includes: quantifying the single-parameter deviation of each parameter from the normal flight state digital template using standardized Z-scores; quantifying the overall deviation of each parameter from the normal flight state digital template using Mahalanobis distance or cosine similarity; and setting dynamic thresholds for single-parameter deviation and overall deviation to generate single-threshold alarms.

[0012] In a preferred embodiment, the dynamic threshold can be adaptively adjusted using an online learning algorithm.

[0013] In a preferred embodiment, the step of completing the abnormal state detection and prediction based on the matching results includes: performing abnormal state detection on the joint parameter correlation between single parameter deviation and overall deviation; using a time series model to predict the future state based on the current single parameter deviation and overall deviation combined with historical data; and outputting the abnormal state detection result and the future state prediction result.

[0014] In a second aspect, the present invention provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the at least one processor performs the above-described method by executing the instructions stored in the memory.

[0015] Thirdly, the present invention provides a computer-readable storage medium for storing instructions that, when executed, enable the above-described method to be implemented.

[0016] Fourthly, the present invention provides a computer program product that, when invoked by a computer, causes the computer to execute the above-described method.

[0017] This invention extends the capabilities of flight data recording and playback systems from offline post-event playback to online real-time flight tracking, significantly improving the sensitivity and accuracy of anomaly detection. Compared to existing technologies, this invention's dynamic template generation and multivariate correlation analysis can capture subtle anomalies, possessing practical application value and wide applicability in UAV scenarios such as military reconnaissance, agricultural monitoring, or logistics delivery. Simultaneously, it retains real-time recording and playback functions, ensuring data integrity and system compatibility. Therefore, the beneficial effects of this invention include: 1. Improved detection accuracy; 2. Increased early warning time; 3. Reduced computational overhead; 4. Enhanced system security and support for scalability (such as integrating more sensors). Through a file writing mechanism, all real-time data and prediction results can be persistently stored, facilitating auditing and optimization of the flight data recording and playback system. Attached Figure Description

[0018] Figure 1 is a flowchart of a real-time abnormal state detection and prediction method based on a historical data model provided by an embodiment of the present invention.

[0019] Figure 2 is a flowchart of generating a digital template of normal flight status in an embodiment of the present invention.

[0020] Figure 3 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0022] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0023] This invention provides a real-time abnormal state detection and prediction method based on a historical data model, applicable to unmanned aerial vehicle (UAV) systems, such as multi-rotor UAVs, fixed-wing UAVs, or other unmanned aerial vehicle systems. As shown in Figure 1, the real-time abnormal state detection and prediction method based on a historical data model includes: S100, using a flight data recording and playback system to dynamically generate a digital template of normal flight state based on historical successful flight data; S200, using the flight data recording and playback system to acquire real-time data streams and matching the real-time data streams based on the digital template of normal flight state; S300, completing abnormal state detection and prediction based on the matching results.

[0024] This method dynamically generates a digital template of normal flight status (i.e., a data model describing normal flight) and monitors data deviation (i.e., the degree of difference between the data and the model) in real time, enabling early identification and trend prediction of minor anomalies. This transforms the flight data recording and playback system from post-flight playback to real-time escort mode, providing efficient decision support. Specifically, the real-time recording function of the flight data recording and playback system is extended to allow online access to real-time data streams while retaining its file writing and state restoration capabilities. This ensures data integrity and traceability; that is, it can write various UAV data generated by the UAV and the ground command and control station into a data log file according to the required file format (such as a structured binary file containing timestamps, parameter IDs, and values) while the UAV is performing a mission. It also supports restoring the real-time recorded state of the UAV data from the data log file, achieving both real-time recording and post-flight playback functions.

[0025] The following details the specific implementation of the real-time anomaly detection and prediction method based on historical data models.

[0026] As shown in Figure 1, the real-time abnormal state detection and prediction method based on historical data model includes: S100, using a flight data recording and playback system to dynamically generate a digital template of normal flight state based on historical successful flight data.

[0027] S101, utilizing the offline data processing capabilities of the flight data recording and playback system, processes historical successful flight data. This historical successful flight data is written into a data log file and stored according to the required file format (such as a structured binary file containing timestamps, parameter IDs, and values) through the real-time recording mechanism of the flight data recording and playback system. The historical successful flight data originates from sensor data collected during historical missions, including hundreds of parameters such as GPS position, altitude, speed, acceleration, engine temperature, RPM, fuel consumption rate, battery voltage, wind speed, and ambient temperature. For example, in 1000 successful cruise missions, this data is recorded as a time series for subsequent analysis.

[0028] S102 extracts parameters from historical successful flight data and dynamically generates a digital template of normal flight status based on the mission profile (such as takeoff, cruise, reconnaissance, or landing phases). The specific generation process includes: data cleaning: using the restoration function of the flight data recording and playback system to read historical successful flight data from the data recording file and remove noise and outliers (such as invalid data caused by sudden sensor failures).

[0029] Parameter Extraction: Principal Component Analysis (PCA) was used to reduce the dimensionality of the retrieved historical successful flight data and extract parameters. PCA is a data compression method that can reduce hundreds of parameters to 20 principal components while retaining 95% of the original information. For example, "flight trajectory stability" was extracted as a new parameter from GPS position and altitude data.

[0030] Statistical modeling: For the extracted parameters, statistical methods are used to calculate the mean μ and standard deviation σ (data dispersion), forming confidence intervals [μ-3σ, μ+3σ] (covering the range of 99.7% of normal data). Based on the confidence intervals and the task profile, the normal fluctuation range of each parameter is adjusted. For example, in a high-wind-speed environment, the normal range of the wind speed parameter is adjusted to 5 m / s to 15 m / s.

[0031] Relationship Analysis: For the extracted parameters, multidimensional parameter correlation analysis methods are applied to obtain the parameter correlation relationships. These methods include Pearson correlation coefficient (used to calculate the linear correlation strength between two parameters, e.g., a correlation coefficient r=0.85 indicates a strong correlation), multiple regression models (used to establish functional relationships between multiple parameters, e.g., Y=aX+b, where Y is fuel consumption rate, X is engine speed, and a and b are fitting coefficients), Gaussian Mixture Model (GMM, used to handle complex distributed data), or neural networks (such as Long Short-Term Memory (LSTM) networks, used to capture time-series dependencies, such as the delayed response of temperature after a change in engine speed). For example, the correlation between temperature and engine speed is calculated to be r=0.85 using the Pearson correlation coefficient; another example is using a LSTM network to input the parameter sequence of the past 10 seconds to predict the current temperature value.

[0032] Template Optimization: A digital template for normal flight status is generated through statistical modeling and relational analysis. It can be seen that this digital template is a multi-dimensional mathematical model that defines the normal fluctuation range of parameters and the correlation between parameters at each stage of the mission profile. The digital template for normal flight status needs to be cross-validated (e.g., dividing the data into training and test sets to verify the model's generalization ability) to ensure its generalization capability.

[0033] Template Storage: The normal flight status digital template is stored in JSON or database format for easy real-time retrieval. The generation process ensures the high accuracy and adaptability of the normal flight status digital template, while also being compatible with the file format requirements of flight data recording and playback systems for easy integration.

[0034] S200: Acquire real-time data streams using the flight data recording and playback system, and match the real-time data streams based on a digital template of normal flight conditions; S201: When the UAV is performing a mission, the flight data recording and playback system acquires real-time data streams through bypass access (such as an API interface or data bus), samples the real-time data streams (at a frequency of one second or higher) to obtain real-time data of parameters, which is time-series data (such as continuous rotational speed values). Simultaneously, the flight data recording and playback system writes this real-time data to a data log file according to its original functions, supporting post-mission playback.

[0035] S202, based on real-time data of parameters, a matching algorithm is used to calculate the deviation of parameters from the normal flight state digital template, including: (1) Single parameter deviation calculation: the deviation of each parameter from the normal flight state digital template is quantified using a standardized Z-score (Z=(x-μ) / σ), where x is the current real-time parameter value (e.g., 5000 rpm), μ is the mean of historical data, and σ is the standard deviation. For example, Z>2 indicates a deviation from the normal range, which may trigger an alarm.

[0036] (2) Overall Deviation Monitoring: The overall deviation of each parameter from the normal flight state numerical template is quantified using Mahalanobis distance or cosine similarity. Mahalanobis distance considers the covariance matrix between parameters (a matrix describing the mutual influence of parameters) to capture correlations. For example, a Mahalanobis distance > 3 indicates an overall anomaly. Cosine similarity calculates the consistency of the directions of two vectors, with a value close to 1 indicating normality.

[0037] (3) Threshold setting: Dynamic thresholds are set for single-parameter deviation and overall deviation to trigger threshold alarms. The initial dynamic threshold is set based on historical data (e.g., a low-level alarm is triggered when the deviation is >2, and a high-level alarm is triggered when the deviation is >5), and can be adaptively adjusted using online learning algorithms (such as the Kalman filter, an adaptive filtering method used to adjust the threshold in real time to adapt to environmental changes).

[0038] The matching algorithm described above is optimized in resource-constrained UAV edge computing environments to achieve low-latency (e.g., <100ms) processing, ensuring no interference with the main flight control system and seamless integration with the real-time recording of the flight data recording and playback system.

[0039] S300: Complete abnormal state detection and prediction based on the matching results.

[0040] (1) Abnormal State Detection: Abnormal state detection is performed by combining the parameter correlation between single parameter deviation and overall deviation. In this embodiment of the invention, a hierarchical cascade approach is adopted to perform abnormal detection by combining the parameter correlation between single parameter deviation and overall deviation, so as to cover the full spectrum of identification from significant faults to minor anomalies: First level (single parameter threshold alarm): First, it is determined whether the single parameter deviation of each parameter exceeds the dynamic threshold. If a single parameter (such as the battery voltage dropping sharply below the dynamic threshold) significantly exceeds the limit, an instantaneous alarm is immediately triggered.

[0041] The second level (overall deviation monitoring): If no alarm is triggered by a single parameter, the overall deviation of multiple current parameters from the normal flight state digital template is further evaluated using Mahalanobis distance or cosine similarity. If the overall deviation exceeds the dynamic threshold, it indicates that although the individual parameters are still within acceptable range, the overall system state has deviated from the "normal flight path," which is considered a systemic anomaly.

[0042] Level 3 (Weak Anomalies and Correlation Analysis): Many early fault characteristics manifest as a "break in the coupling relationship" between parameters, rather than numerical out-of-bounds errors. Therefore, for latent weak anomalies not detected in Levels 1 and 2, parameter correlation analysis is needed to identify these "weak anomalies." The parameter correlation can be achieved using a multidimensional parameter correlation analysis method that generates parameter correlation relationships in a normal flight state digital template. For example, under a specific flight attitude, the UAV's battery output voltage (e.g., 22.5V) and rotor speed (e.g., 4000rpm) are both within normal fluctuation ranges when viewed individually, meeting Level 1 requirements; and the deviation is within the dynamic threshold, also meeting Level 2 requirements. However, based on historical patterns in the normal flight state digital template, when the rotor speed is maintained at 4000rpm and in the climb phase, the battery voltage should exhibit a specific linear decreasing slope or maintain a specific functional relationship with the discharge current (expected correlation coefficient r≈0.98). If real-time monitoring detects a weakening of the correlation between the two (e.g., the actual correlation coefficient drops to r=0.75), or the response lag time of voltage changes with current exceeds the tolerance defined in the digital template for normal flight conditions, the system determines that there is a "latent weak anomaly" (such as an abnormal increase in battery internal resistance or a precursor to aging of the electronic speed controller).

[0043] (2) Future State Prediction: Based on the current single-parameter deviation and overall deviation, combined with historical data, a time series model (such as the Autoregressive Integrated Moving Average (ARIMA) model, used to predict trends, where p=1 represents the autoregressive order, d=1 represents the difference order, and q=1 represents the moving average order; or the Prophet prediction model, used to handle seasonality and trends) is used to predict the future state. For example, the ARIMA model can be used to predict that a power system failure may occur within 3-5 minutes, and the confidence probability can be calculated (an alarm is issued when the probability is >80%). To improve prediction accuracy, the Runge-Kutta fourth-order method can be used to solve the difference equations for incremental calculations. For example, the position update formula is x(t+Δt)=x(t)+vx*Δt+0.5*ax*Δt², where Δt is the time step (i.e., the simulation time increment, usually 0.01 seconds, used for step-by-step calculation of the future state), vx is the velocity in the x-direction (affected by wind speed), and ax is the acceleration in the x-direction (affected by flight commands). This method improves accuracy through fourth-order approximation, avoiding the error accumulation of lower-order methods.

[0044] (3) Output of results: Output the abnormal state detection results and future state prediction results, including the abnormality type, severity, prediction time window and suggested actions (such as deceleration or return to base), and provide feedback to the operator through human-machine interface or wireless transmission. At the same time, abnormal events can be written to a dedicated file through the flight data recording and playback system for easy post-event playback and further analysis.

[0045] The following is an example of the method described above.

[0046] The S100 utilizes a flight data recording and playback system to dynamically generate a digital template of normal flight status based on historical successful flight data. Specifically, the system processes historical successful flight data, such as sensor logs from 1000 cruise missions. This historical successful flight data is written to a specified file format and retrieved via the system's recovery function. For reconnaissance mission profiles, parameters such as altitude, speed, temperature, engine speed, and fuel consumption rate are extracted. Through statistical modeling (e.g., mean and variance calculations) and machine learning correlation analysis (e.g., multiple regression), a digital template of normal flight status is generated. This template defines the normal fluctuation range of parameters (e.g., temperature fluctuation ±5°C during cruise) and their correlations (e.g., the linear ratio of engine speed to fuel consumption) for each stage of the mission profile.

[0047] The S200 utilizes a flight data recording and playback system to acquire real-time data streams, which are then matched against a normal flight state digital template. Specifically, during a UAV flight mission, the flight data recording and playback system bypasses the real-time data stream (e.g., acquiring it once per second via an API interface) and writes the data to a file. A matching algorithm (such as Euclidean distance or cosine similarity calculation) is used to compare the deviation of the current parameters from the normal flight state digital template. If the deviation exceeds a dynamic threshold (e.g., 10%), monitoring is triggered.

[0048] The S300 performs anomaly detection and prediction based on matching results. Specifically, it performs correlation analysis on seemingly normal parameter combinations. For example, the engine temperature is at the normal value of 80°C, and the engine speed is at the normal value of 5000 rpm, but the correlation with fuel consumption rate deviates from the normal flight condition numerical template (expected consumption rate 0.5L / min, actual 0.7L / min). Potential faults are predicted 3 minutes in advance using a time series model, and an alert is issued to the operator. Anomaly data is written to a file through the flight data recording and playback system, supporting post-event playback.

[0049] Based on the same technical concept, embodiments of the present invention also provide an electronic device that can implement the real-time abnormal state detection and prediction method for unmanned aerial vehicles (UAVs) provided in the above embodiments of the present invention. In one embodiment, the electronic device can be a server, a terminal device, or other electronic devices. As shown in Figure 3, the electronic device may include: at least one processor and a memory connected to at least one processor. Embodiments of the present invention do not limit the specific connection medium between the processor and the memory; Figure 3 shows an example of a connection between the processor and the memory via a bus. The bus is represented by a thick line in Figure 3. The connection methods between other components are only illustrative and not intended to be limiting. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, only one thick line is used in Figure 3, but this does not mean that there is only one bus or one type of bus. Alternatively, the processor can also be called a controller; the name is not limited.

[0050] In this embodiment of the invention, the memory stores instructions that can be executed by at least one processor. By executing the instructions stored in the memory, at least one processor can execute the aforementioned method for real-time abnormal state detection and prediction of unmanned aerial vehicles.

[0051] The processor is the control center of the device. It can connect to various parts of the control device through various interfaces and lines. By running or executing instructions stored in memory and calling data stored in memory, it can monitor the device's various functions and process data, thereby enabling overall monitoring of the device.

[0052] In an alternative design, the processor may include one or more processing units. The processor may integrate an application processor and a modem processor, wherein the application processor primarily handles the operating system, user interface, and applications, while the modem processor primarily handles wireless communication. It is understood that the modem processor may also not be integrated into the processor. In some embodiments, the processor and memory may be implemented on the same chip; in some embodiments, they may also be implemented separately on separate chips.

[0053] The processor can be a general-purpose processor, such as a CPU, digital signal processor, application-specific integrated circuit, field-programmable gate array or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component, capable of implementing or executing the methods, steps, and logic block diagrams disclosed in the embodiments of this invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the real-time abnormal state detection and prediction method for unmanned aerial vehicles disclosed in the embodiments of this invention can be directly manifested as execution by a hardware processor, or execution by a combination of hardware and software modules within the processor.

[0054] Memory, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. Memory can include at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type memory, random access memory (RAM), static random access memory (SRAM), programmable read-only memory (PROM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), magnetic memory, magnetic disk, optical disk, etc. Memory is any other medium capable of carrying or storing desired program code in the form of instructions or data structures, and accessible by a computer, but is not limited thereto. In embodiments of the present invention, memory can also be a circuit or any other device capable of implementing storage functions, used to store program instructions and / or data.

[0055] By designing and programming the processor, the code corresponding to the real-time abnormal state detection and prediction method for UAVs described in the foregoing embodiments can be embedded into the chip, enabling the chip to execute the steps of the method described in the foregoing embodiments during runtime. How to design and program the processor is a technique well-known to those skilled in the art and will not be elaborated upon here.

[0056] Based on the same inventive concept, embodiments of the present invention also provide a storage medium storing computer instructions, which, when executed on a computer, cause the computer to perform a real-time abnormal state detection and prediction method for unmanned aerial vehicles (UAVs) as described above.

[0057] In some alternative embodiments, the present invention also provides a method for real-time abnormal state detection and prediction of unmanned aerial vehicles (UAVs), which can also be implemented as a program product including program code. When the program product is run on a device, the program code is used to cause the control device to perform the steps in the method for real-time abnormal state detection and prediction of unmanned aerial vehicles (UAVs) according to various exemplary embodiments of the present invention as described above.

[0058] It should be noted that although several units or sub-units of the apparatus have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of the invention, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units. Furthermore, although the operation of the method of the invention is described in a specific order in the drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0059] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0060] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a server, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more blocks of the flowchart illustrations and / or one or more blocks of the block diagrams.

[0061] Program code for performing the operations of this invention can be written using any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0062] In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0063] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0064] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0065] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for real-time abnormal state detection and prediction of unmanned aerial vehicles (UAVs), characterized in that, include: Using a flight data recording and playback system, a digital template of normal flight status is dynamically generated based on historical successful flight data; real-time data streams are acquired using the flight data recording and playback system, and the real-time data streams are matched based on the digital template of normal flight status; abnormal status detection and prediction are completed based on the matching results.

2. The method for real-time abnormal state detection and prediction of unmanned aerial vehicles according to claim 1, characterized in that, The method of dynamically generating a digital template of normal flight status based on historical successful flight data using the flight data recording and playback system includes: processing historical successful flight data using the offline data processing capability of the flight data recording and playback system; writing the historical successful flight data into a data recording file and storing it in the required file format through the real-time recording mechanism of the flight data recording and playback system; extracting parameters from the historical successful flight data and dynamically generating a digital template of normal flight status based on the mission profile.

3. The method for real-time abnormal state detection and prediction of unmanned aerial vehicles according to claim 2, characterized in that, The process of extracting parameters from historical successful flight data and dynamically generating a digital template of normal flight status based on the mission profile includes: using the restoration function of the flight data recording and playback system to read historical successful flight data from the data recording file; using principal component analysis to reduce the dimensionality of the read historical successful flight data and extract parameters; for the extracted parameters, using statistical methods to calculate the mean μ and standard deviation σ, and forming a confidence interval [μ-3σ, μ+3σ], adjusting the normal fluctuation range of each parameter based on the confidence interval and the mission profile; for the extracted parameters, applying multidimensional parameter correlation analysis methods to obtain parameter correlation relationships; the multidimensional parameter correlation analysis methods include Pearson correlation coefficient, multiple regression model, Gaussian mixture model, or neural network; the generated digital template of normal flight status defines the normal fluctuation range and parameter correlation relationships of individual parameters at each stage of the mission profile, and cross-validation is used to ensure the generalization ability of the digital template of normal flight status.

4. The method for real-time abnormal state detection and prediction of unmanned aerial vehicles according to claim 1, characterized in that, The process of acquiring real-time data streams using a flight data recording and playback system and matching the real-time data streams based on a normal flight state digital template includes: when the UAV is performing a mission, the flight data recording and playback system acquires real-time data streams through bypass access, samples the real-time data streams to obtain real-time data of parameters; based on the real-time data of parameters, a matching algorithm is used to calculate the deviation of each parameter from the normal flight state digital template.

5. The method for real-time abnormal state detection and prediction of unmanned aerial vehicles according to claim 4, characterized in that, The method of using a matching algorithm to calculate the deviation of each parameter from the digital template of normal flight status includes: quantifying the single-parameter deviation of each parameter from the digital template of normal flight status using standardized Z-scores; quantifying the overall deviation of each parameter from the digital template of normal flight status using Mahalanobis distance or cosine similarity; and setting dynamic thresholds for single-parameter deviation and overall deviation to generate single-threshold alarms.

6. The method for real-time abnormal state detection and prediction of unmanned aerial vehicles according to claim 5, characterized in that, The dynamic threshold can be adaptively adjusted using an online learning algorithm.

7. The method for real-time abnormal state detection and prediction of unmanned aerial vehicles according to claim 5, characterized in that, The process of detecting and predicting abnormal states based on matching results includes: detecting abnormal states by analyzing the correlation between single-parameter deviation and overall deviation parameters; using a time series model to predict future states by combining the current single-parameter deviation and overall deviation with historical data; and outputting the abnormal state detection results and future state prediction results.

8. An electronic device, characterized in that, include: At least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the at least one processor performs the method as described in any one of claims 1-7 by executing the instructions stored in the memory.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store instructions that, when executed, cause the method as described in any one of claims 1-7 to be implemented.

10. A computer program product, characterized in that, When the computer program product is invoked by a computer, it causes the computer to perform the method as described in any one of claims 1-7.

Images