A method, system, device and medium for establishing a drone fault diagnosis system

Abstract

The embodiments of the present application provide a method, system, device, and medium for establishing a drone fault diagnosis system, and relate to the field of drone technology. The method includes: constructing a data-driven diagnostic signal based on the first control data of the actuator and the first detection data of the sensor under normal drone operation; constructing various fault diagnosability indicators based on the second control data of the actuator and the second detection data of the sensor under various drone fault conditions; determining the data-driven residual and its evaluation function based on the data-driven diagnostic signal and the various fault diagnosability indicators; performing eigenmode decomposition on the data-driven residual, extracting the frequency domain features of the eigenmode decomposition signal, and extracting the time domain features of the evaluation function; establishing a fault diagnosis decision unit based on the frequency domain features and the time domain features, so as to establish a drone fault diagnosis system using the fault diagnosis decision unit. The embodiments of the present application can improve the accuracy of drone fault diagnosis.

Description

Technical Field

[0001] The present application relates to the field of drone technology, and more specifically, to a method, system, device, and medium for establishing a drone fault diagnosis system. Background Art

[0002] In recent years, drones have been widely used to perform various tasks such as military reconnaissance, environmental monitoring, facility inspection, and logistics distribution. The complex and changeable flight environment and various mission requirements have led to a continuous increase in the failure rate of drone onboard components, especially actuators and sensors in drone flight control systems.

[0003] Currently, the main methods for UAV fault diagnosis include model-based fault diagnosis methods and data-driven fault diagnosis methods. Model-based fault diagnosis methods require the establishment of an accurate mathematical model of the UAV and simulation analysis of the model to diagnose UAV faults. Data-driven fault diagnosis methods mainly diagnose UAV faults by analyzing the UAV's historical flight data to identify fault characteristics. Since the UAV's flight process is nonlinear, and the aerodynamic parameters change significantly with the flight state, coupled with problems such as hysteresis in the transmission, it is difficult to establish an accurate mathematical model of the UAV, so data-driven fault diagnosis methods are usually used. However, existing data-driven fault diagnosis methods mostly use time domain features to achieve fault diagnosis. The characteristic signals of minor faults that appear early in the UAV flight process are often submerged in noise, making it difficult to diagnose minor faults in the early stages. The accuracy of UAV fault diagnosis still needs to be improved. Summary of the Invention

[0004] The purpose of the embodiments of the present application is to provide a method, system, device and medium for establishing a drone fault diagnosis system, so as to achieve the technical effect of improving the accuracy of drone fault diagnosis.

[0005] In a first aspect, an embodiment of the present application provides a method for establishing a fault diagnosis system for a drone, wherein the drone is equipped with an actuator and a sensor; the method comprises:

[0006] constructing a data-driven diagnostic signal according to the first control data of the actuator and the first detection data of the sensor when the drone is operating normally;

[0007] constructing respective fault diagnosability indicators according to the second control data of the actuator and the second detection data of the sensor when respective faults of the drone occur;

[0008] Determining a data-driven residual and an evaluation function of the data-driven residual according to the data-driven diagnostic signal and the respective fault diagnosability indicators;

[0009] Performing eigenmode decomposition on the data-driven residual, extracting frequency domain features of the eigenmode decomposition signal, and extracting time domain features of the evaluation function;

[0010] Based on the frequency domain features and the time domain features, a fault diagnosis decision unit is established, and a fault diagnosis system of the UAV is established by using the fault diagnosis decision unit.

[0011] In the above implementation process, through the algorithm idea based on the data-driven algorithm, the first control data of the actuator and the first detection data of the sensor under normal operation of the UAV, as well as the second control data of the actuator and the second detection data of the sensor under various faults of the UAV are processed to construct data-driven diagnostic signals and various fault diagnosability indicators. According to the data-driven diagnostic signals and various fault diagnosability indicators, the data-driven residual and the evaluation function of the data-driven residual are determined, and based on the frequency domain characteristics of the characteristic mode decomposition signal of the data-driven residual and the time domain characteristics of the evaluation function, a fault diagnosis decision unit is established. By using the fault diagnosis decision unit to establish a fault diagnosis system for the UAV, the fault diagnosis system of the UAV can be enabled to comprehensively analyze the time and frequency domain characteristics of the fault to perform fault diagnosis including minor faults, thereby helping to improve the accuracy of UAV fault diagnosis.

[0012] Furthermore, constructing a data-driven diagnostic signal according to the first control data of the actuator and the first detection data of the sensor when the drone is operating normally includes:

[0013] constructing a data driving matrix according to the first control data and the first detection data;

[0014] Decomposing a lower triangular matrix from the data-driven matrix;

[0015] Performing singular value decomposition on a target submatrix of the lower triangular matrix to determine a target parameter vector;

[0016] The data-driven diagnostic signal is constructed according to the target parameter vector, the first control data, and the first detection data.

[0017] In the above implementation process, by adopting the methods of orthogonal triangular decomposition and singular value decomposition, a data-driven diagnostic signal is constructed according to the first control data of the actuator and the first detection data of the sensor under normal operation of the drone. The data-driven diagnostic signal can be constructed by comprehensively considering the correlation information between the drone actuator control data and the sensor detection data, which is beneficial to improving the drone fault diagnosis performance.

[0018] Furthermore, constructing each fault diagnosability index based on the second control data of the actuator and the second detection data of the sensor in each case where the drone has a fault includes:

[0019] The respective fault diagnosability indicators are constructed according to a target parameter vector, the second control data and the second detection data; wherein the target parameter vector is determined according to the first control data and the first detection data.

[0020] In the above implementation process, each fault diagnosability index is constructed by constructing a target parameter vector based on the second control data of the actuator and the second detection data of the sensor when each fault occurs in the UAV, and based on the first control data of the actuator and the first detection data of the sensor when the UAV is operating normally. The target parameter vector can be used to accurately associate the data under normal operating conditions of the UAV with the data under various fault conditions of the UAV, thereby ensuring the subsequent accurate determination of the data-driven residual and its evaluation function.

[0021] Furthermore, before determining the data-driven residual and the evaluation function of the data-driven residual based on the data-driven diagnostic signal and the various fault diagnosability indicators, the method further includes:

[0022] Determine whether the respective fault diagnosability indicators corresponding to the data-driven diagnostic signal meet a value condition.

[0023] In the above implementation process, by first determining whether the various fault diagnosability indicators corresponding to the data-driven diagnostic signal meet the value conditions, and then determining the data-driven residual and its evaluation function based on the data-driven diagnostic information and the various fault diagnosability indicators, it is possible to ensure that the data-driven residual and its evaluation function are effectively determined.

[0024] Furthermore, performing eigenmode decomposition on the data-driven residual to extract frequency domain features of the eigenmode decomposition signal includes:

[0025] Performing eigenmode decomposition on the data-driven residual using an eigenmode decomposition method based on correlation kurtosis to obtain the eigenmode decomposition signal;

[0026] The frequency domain features are extracted from the eigenmode decomposition signal.

[0027] In the above implementation process, by adopting the eigenmode decomposition method based on correlation kurtosis, the data-driven residual is subjected to eigenmode decomposition to obtain the eigenmode decomposition signal, and the frequency domain features are extracted from the eigenmode decomposition signal. This can accurately extract the frequency domain features of each fault, which is conducive to further improving the accuracy of UAV fault diagnosis.

[0028] Furthermore, the UAV includes a fixed-wing UAV;

[0029] The actuators include ailerons, elevators, rudders and throttles:

[0030] The sensor includes one or more of an airspeed sensor, an angle of attack sensor, a sideslip angle sensor, a roll gyroscope, a pitch gyroscope, a yaw gyroscope, an inertial measurement unit, an acceleration sensor, a three-axis gyroscope, a global navigation satellite system, and an altitude sensor.

[0031] In a second aspect, an embodiment of the present application provides a device for establishing a fault diagnosis system for a drone, wherein the drone is equipped with an actuator and a sensor; the device comprises:

[0032] a data-driven diagnostic signal construction module, configured to construct a data-driven diagnostic signal based on first control data of the actuator and first detection data of the sensor when the drone is operating normally;

[0033] a fault diagnosability index construction module, configured to construct various fault diagnosability indicators based on the second control data of the actuator and the second detection data of the sensor when various faults occur in the UAV;

[0034] A residual and its evaluation function determination module is used to determine the data-driven residual and the evaluation function of the data-driven residual based on the data-driven diagnosis signal and various fault diagnosability indicators;

[0035] The time-frequency domain feature extraction module is used to perform eigenmode decomposition on the data-driven residual, extract the frequency domain features of the eigenmode decomposition signal, and extract the time domain features of the evaluation function;

[0036] The fault diagnosis system establishment module is used to establish a fault diagnosis decision unit based on frequency domain characteristics and time domain characteristics, so as to establish a fault diagnosis system for the UAV using the fault diagnosis decision unit.

[0037] In a third aspect, an embodiment of the present application provides a drone fault diagnosis system, which is established according to the drone fault diagnosis system establishment method as described above.

[0038] In a fourth aspect, an embodiment of the present application provides a computer program product, which includes instructions. When the instructions are executed by a computer, the computer implements the method as described above.

[0039] In a fifth aspect, an embodiment of the present application provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor; when the processor executes the computer program, the method described above is implemented.

[0040] In a sixth aspect, an embodiment of the present application provides a computer-readable storage medium, which includes a stored computer program; wherein, when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method described above. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0042] Figure 1 A flowchart of a method for establishing a UAV fault diagnosis system provided in the first embodiment of the present application;

[0043] Figure 2 A schematic diagram of the structure of a device for establishing a UAV fault diagnosis system according to the second embodiment of the present application;

[0044] Figure 3 A schematic structural diagram of a UAV fault diagnosis system provided in the third embodiment of the present application;

[0045] Figure 4 A schematic structural diagram of an electronic device provided in the third embodiment of the present application. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0047] It should be noted that in the description of this application, the terms "first" and "second" are used only to distinguish descriptions and should not be understood to indicate or imply relative importance. Furthermore, the step numbers herein are used only to facilitate the explanation of the embodiments of this application and do not limit the order in which the steps are to be executed.

[0048] In recent years, drones have been widely used to perform various tasks such as military reconnaissance, environmental monitoring, facility inspection, and logistics distribution. The complex and changeable flight environment and various mission requirements have led to a continuous increase in the failure rate of drone onboard components, especially actuators and sensors in drone flight control systems.

[0049] In the related art, the main methods for UAV fault diagnosis include model-based fault diagnosis methods and data-driven fault diagnosis methods. Model-based fault diagnosis methods require the establishment of an accurate mathematical model of the UAV and simulation analysis of the model to diagnose the UAV fault. Data-driven fault diagnosis methods mainly diagnose UAV faults by analyzing the UAV's historical flight data to identify fault characteristics. Since the UAV's flight process is nonlinear and the aerodynamic parameters change significantly with the flight state, and there are problems such as hysteresis in the transmission, it is difficult to establish an accurate mathematical model of the UAV, so a data-driven fault diagnosis method is usually used. However, most of the existing data-driven fault diagnosis methods use time domain features to achieve fault diagnosis. The characteristic signals of minor faults that appear early in the UAV flight process are often submerged in noise, making it difficult to diagnose minor faults in the early stages. The accuracy of UAV fault diagnosis still needs to be improved.

[0050] To this end, an embodiment of the present application provides a method for establishing a drone fault diagnosis system. Through an algorithmic idea based on a data-driven algorithm, the first control data of the actuator and the first detection data of the sensor under normal operation of the drone, as well as the second control data of the actuator and the second detection data of the sensor under various faults of the drone are processed to construct a data-driven diagnostic signal and various fault diagnosability indicators. According to the data-driven diagnostic signal and various fault diagnosability indicators, the data-driven residual and the evaluation function of the data-driven residual are determined, and based on the frequency domain characteristics of the characteristic mode decomposition signal of the data-driven residual and the time domain characteristics of the evaluation function, a fault diagnosis decision unit is established. By using the fault diagnosis decision unit to establish a drone fault diagnosis system, the drone fault diagnosis system can be enabled to comprehensively analyze the time-frequency domain characteristics of the fault to perform fault diagnosis including minor faults, thereby helping to improve the accuracy of drone fault diagnosis.

[0051] The method provided in the embodiment of the present application can be executed by a relevant terminal device, and the following description will be given using a user terminal as an example of the execution subject.

[0052] Please see Figure 1 , Figure 1 This is a flowchart of a method for establishing a UAV fault diagnosis system provided in the first embodiment of the present application. The first embodiment of the present application provides a method for establishing a UAV fault diagnosis system, wherein the UAV is equipped with actuators and sensors; the method comprises steps S101 to S105:

[0053] S101, constructing a data-driven diagnostic signal based on first control data of an actuator and first detection data of a sensor when the drone is operating normally;

[0054] S102, constructing each fault diagnosability index based on the second control data of the actuator and the second detection data of the sensor when each fault occurs in the drone;

[0055] S103, determining a data-driven residual and an evaluation function of the data-driven residual based on the data-driven diagnostic signal and each fault diagnosability index;

[0056] S104, performing eigenmode decomposition on the data-driven residual, extracting frequency domain features of the eigenmode decomposition signal, and extracting time domain features of the evaluation function;

[0057] S105. Establish a fault diagnosis decision unit based on the frequency domain characteristics and the time domain characteristics, and establish a fault diagnosis system for the UAV using the fault diagnosis decision unit.

[0058] As an example, according to actual application requirements, a certain type of drone is selected, and the drone is equipped with actuators and sensors.

[0059] The first control data of the actuator and the first detection data of the sensor are obtained when the drone is operating normally, and the second control data of the actuator and the second detection data of the sensor are obtained when various faults occur in the drone.

[0060] In an optional implementation of an embodiment of the present application, the drone includes a fixed-wing drone; the actuators include ailerons, elevators, rudders and throttles; the sensors include one or more of an airspeed sensor, an angle of attack sensor, a sideslip angle sensor, a roll gyroscope, a pitch gyroscope, a yaw gyroscope, an inertial measurement unit, an acceleration sensor, a three-axis gyroscope, a global navigation satellite system and an altitude sensor.

[0061] As an example, if the selected UAV is a small fixed-wing UAV, the actuators configured for the small fixed-wing UAV usually include ailerons, elevators, rudders and throttles, and the sensors configured usually include one or more of an airspeed sensor, an angle of attack sensor, a sideslip angle sensor, a roll gyroscope, a pitch gyroscope, a yaw gyroscope, an inertial measurement unit, an acceleration sensor, a three-axis gyroscope, a global navigation satellite system and an altitude sensor. Then, the first control data of the actuator under normal operation of the UAV may include the control data of each actuator collected at t sampling moments during the normal operation of the UAV, and the first detection data of the sensor under normal operation of the UAV may include the detection data of each sensor collected at t sampling moments during the normal operation of the UAV.

[0062] For example, the first control data of the actuator under normal operation of the UAV is shown in formula (1):

[0063] (1);

[0064] In formula (1), is the control data of each actuator collected at the kth sampling moment during the normal operation of the UAV; is the offset of the aileron control stick collected at the kth sampling moment during the normal operation of the UAV; is the offset of the elevator control stick collected at the kth sampling moment during the normal operation of the UAV; is the offset of the rudder control stick collected at the kth sampling moment during the normal operation of the UAV; is the offset of the throttle control lever collected at the kth sampling moment during the normal operation of the UAV; , t is the total number of sampling moments; Represents the transpose of a matrix.

[0065] The first detection data of the sensor under normal operation of the drone is shown in formula (2):

[0066] (2);

[0067] In formula (2), is the control data of each sensor collected at the kth sampling moment during the normal operation of the UAV; is the airspeed collected at the kth sampling moment during the normal operation of the UAV, detected by the airspeed sensor; is the angle of attack collected at the kth sampling moment during the normal operation of the UAV, detected by the angle of attack sensor; is the sideslip angle collected at the kth sampling moment during the normal operation of the UAV, detected by the sideslip angle sensor; is the roll angular velocity collected at the kth sampling moment during the normal operation of the UAV, detected by the roll gyroscope; is the pitch angular velocity collected at the kth sampling moment during the normal operation of the UAV, detected by the pitch gyroscope; is the yaw angular velocity collected at the kth sampling moment during the normal operation of the UAV, detected by the yaw gyroscope; The roll angle is the roll angle collected at the kth sampling moment during the normal operation of the UAV, which is detected by the Inertial Measurement Unit (IMU) or determined based on the detection results of the acceleration sensor and the three-axis gyroscope; The pitch angle is the pitch angle collected at the kth sampling moment during the normal operation of the UAV, which is detected by the inertial measurement unit or determined based on the detection results of the acceleration sensor and the three-axis gyroscope; The yaw angle is the value collected at the kth sampling moment during the normal operation of the UAV, which is detected by the inertial measurement unit or determined based on the detection results of the acceleration sensor and the three-axis gyroscope; 、 are the eastward displacement and northward displacement collected at the kth sampling moment during the normal operation of the UAV, detected by the Global Navigation Satellite System (GNSS). Since the negative number of the eastward displacement can represent the westward displacement, and the negative number of the northward displacement can represent the southward displacement, in practical applications, displacements in other directions can also be selected, such as westward displacement and northward displacement, or westward displacement and southward displacement; is the height collected at the kth sampling moment during the normal operation of the UAV, detected by the altitude sensor; , t is the total number of sampling moments; Represents the transpose of a matrix.

[0068] In practical applications, the actuators and sensors configured on a drone often depend on the drone's system architecture and the mission being performed. For large fixed-wing drones, actuators may include flaps, elevators, rudders, and throttles, along with airspeed sensors, angle of attack sensors, sideslip angle sensors, roll gyros, pitch gyros, yaw gyros, inertial measurement units (IMUs), accelerometers, three-axis gyros, global navigation satellite systems (GNSS), and altitude sensors, as well as cameras. For rotary-wing drones, actuators typically include motors for each rotor, and sensors are similar to those used on fixed-wing drones.

[0069] The first control data of the actuator and the first detection data of the sensor under normal operation of the drone can be acquired using the above-mentioned acquisition method. The second control data of the actuator and the second detection data of the sensor under various drone failure conditions can be acquired using the acquisition method. Alternatively, the second control data of the actuator and the second detection data of the sensor under various drone failure conditions can be simulated based on common drone failures and failure analysis results.

[0070] For example, the second control data of the actuator under various failure conditions of the drone is shown in formula (3):

[0071] (3);

[0072] In formula (3), is the control data of each actuator collected at the kth sampling moment during the process of the i-th fault of the UAV; is the offset of the aileron control stick collected at the kth sampling moment during the i-th fault of the UAV; is the offset of the elevator control stick collected at the kth sampling moment during the i-th fault of the UAV; is the offset of the rudder control stick collected at the kth sampling moment during the i-th fault of the UAV; is the offset of the throttle control lever collected at the kth sampling moment during the i-th fault of the UAV; , t is the total number of sampling moments; , is the total number of failures; Represents the transpose of a matrix.

[0073] The second detection data of the sensor under various failure conditions of the drone is shown in formula (4):

[0074] (4);

[0075] In formula (4), is the control data of each sensor collected at the kth sampling moment during the process of the i-th fault of the UAV; is the airspeed collected at the kth sampling moment during the i-th fault of the UAV; is the angle of attack collected at the kth sampling moment during the process of the i-th fault of the UAV; is the sideslip angle collected at the kth sampling moment during the i-th fault of the UAV; is the roll angular velocity collected at the kth sampling moment during the i-th fault of the UAV; is the pitch angular velocity collected at the kth sampling moment during the i-th fault of the UAV; is the yaw angular velocity collected at the kth sampling moment during the i-th fault of the UAV; is the roll angle collected at the kth sampling moment during the i-th fault of the UAV; is the pitch angle collected at the kth sampling moment during the i-th fault of the UAV; is the yaw angle collected at the kth sampling moment during the i-th fault of the UAV; are the eastward displacement and northward displacement collected at the kth sampling moment during the process of the i-th fault of the UAV; is the height collected at the kth sampling moment during the process of the i-th fault of the UAV, detected by the altitude sensor; , t is the total number of sampling moments; , is the total number of failures; Represents the transpose of a matrix.

[0076] A data-driven diagnostic signal is constructed based on the first control data of the actuator and the first detection data of the sensor when the drone is operating normally. Furthermore, a fault diagnosability index corresponding to each fault is constructed based on the second control data of the actuator and the second detection data of the sensor when the drone fails, thereby obtaining each fault diagnosability index.

[0077] After obtaining the data-driven diagnostic signal and each fault diagnosability index, the data-driven residual and the evaluation function of the data-driven residual are determined according to the data-driven diagnostic signal and each fault diagnosability index.

[0078] If a drone fault occurs and the corresponding fault diagnosability index value exceeds the performance threshold, the data-driven diagnostic signal will change. By determining the data-driven residual and its evaluation function based on the data-driven diagnostic signal and each fault diagnosability index, the data-driven residual and its evaluation function can be used to accurately measure the sensitivity of the data-driven diagnostic signal to diagnosing various faults.

[0079] The data-driven residual is subjected to eigenmode decomposition, the frequency domain features of the eigenmode decomposition signal are extracted, and the time domain features of the evaluation function are extracted.

[0080] Based on the frequency domain characteristics and time domain characteristics, a fault diagnosis decision unit is established, and a fault diagnosis system for the UAV is established using the fault diagnosis decision unit.

[0081] By combining the frequency domain characteristics of the characteristic mode decomposition signal of the data-driven residual and the time domain characteristics of the evaluation function to establish a UAV fault diagnosis system, the UAV fault diagnosis system can perform fault diagnosis including minor faults based on the time and frequency domain characteristics of the fault, ensuring that the fault diagnosis results are more accurate.

[0082] The embodiment of the present application, through an algorithmic idea based on a data-driven algorithm, processes the first control data of the actuator and the first detection data of the sensor under normal operation of the drone, as well as the second control data of the actuator and the second detection data of the sensor under various fault conditions of the drone, to construct a data-driven diagnostic signal and various fault diagnosability indicators. According to the data-driven diagnostic signal and various fault diagnosability indicators, the data-driven residual and the evaluation function of the data-driven residual are determined, and based on the frequency domain characteristics of the characteristic mode decomposition signal of the data-driven residual and the time domain characteristics of the evaluation function, a fault diagnosis decision unit is established. By using the fault diagnosis decision unit to establish a fault diagnosis system for the drone, the fault diagnosis system of the drone can be enabled to comprehensively analyze the time-frequency domain characteristics of the fault to perform fault diagnosis including minor faults, thereby helping to improve the accuracy of drone fault diagnosis.

[0083] In an optional embodiment, the data-driven diagnostic signal is constructed based on the first control data of the actuator and the first detection data of the sensor under normal operation of the drone, including: constructing a data-driven matrix based on the first control data and the first detection data; decomposing a lower triangular matrix from the data-driven matrix; performing singular value decomposition on the target submatrix of the lower triangular matrix to determine the target parameter vector; and constructing the data-driven diagnostic signal based on the target parameter vector, the first control data and the first detection data.

[0084] As an example, after obtaining the first control data of the actuator and the first detection data of the sensor under normal operation of the drone, a data driving matrix is ​​constructed according to the first control data of the actuator and the first detection data of the sensor under normal operation of the drone.

[0085] In an optional implementation manner of the embodiment of the present application, constructing a data driving matrix based on the first control data and the first detection data includes: constructing a first input vector based on the first control data; constructing a second input vector based on the first detection data; and constructing a data driving matrix based on the first input vector and the second input vector.

[0086] As an example, assuming that the first control data of the actuator under normal operation of the drone is as shown in formula (1), and the first detection data of the sensor under normal operation of the drone is as shown in formula (2), then the first data driving parameter n is set, and the first input vector constructed according to the first control data of the actuator under normal operation of the drone is as shown in formula (5):

[0087] , (5);

[0088] The second input vector constructed based on the first detection data of the sensor under normal operation of the drone is shown in formula (6):

[0089] , (6);

[0090] The second data driving parameter N is set, and the data driving matrix constructed according to the first input vector and the second input vector is shown in formula (7):

[0091] (7);

[0092] In formula (7), , ,

[0093] ,

[0094] .

[0095] After obtaining the data-driven matrix, the orthogonal triangular decomposition method is used to decompose the lower triangular matrix from the data-driven matrix, and the singular value decomposition method is used to perform singular value decomposition on the target submatrix of the lower triangular matrix to determine the target parameter vector.

[0096] It should be noted that the target sub-matrix of the lower triangular matrix contains the first control data of the actuator and the first detection data of the sensor under normal operation of the drone, and the association information between the first control data of the actuator and the second detection data of the sensor under various fault conditions of the drone.

[0097] For example, the data-driven matrix Z shown in formula (7) is decomposed into lower triangular form, and the result is divided into blocks, as shown in formula (8):

[0098] (8);

[0099] In formula (8), L and Q are the lower triangular matrix and orthogonal matrix obtained after decomposition, respectively; 、 、…、 's blocks, and 、 、…、 The blocks are divided into 、 、 and The number of rows is determined; represents the zero matrix.

[0100] For the target submatrix in the lower triangular matrix L Perform singular value decomposition, as shown in formula (9):

[0101] (9);

[0102] In formula (9), 、 are the two unitary matrices obtained after decomposition, , , represents the identity matrix, ; is the singular value matrix, , .

[0103] Setting the singular value threshold , filter out values ​​greater than the singular value threshold from the singular value matrix The singular value of , the screening result is shown in formula (10):

[0104] (10);

[0105] According to the above screening results, at least one feasible parameter vector is determined, and each feasible parameter vector is .

[0106] For each feasible parameter vector The target parameter vectors corresponding to each feasible parameter vector are obtained by transposing the vector into blocks, as shown in formula (11):

[0107] (11);

[0108] In formula (11), is the hth feasible parameter vector; and is the target parameter vector corresponding to the hth feasible parameter vector, for Before List, for Except The remaining columns of .

[0109] According to the target parameter vector corresponding to each feasible parameter vector, the first input vector shown in formula (5), and the second input vector shown in formula (6), the data-driven diagnostic signal corresponding to each feasible parameter vector is constructed, as shown in formula (12):

[0110] , (12);

[0111] In formula (12), is the data-driven diagnostic signal corresponding to the hth feasible parameter vector at the kth sampling moment.

[0112] The embodiment of the present application adopts the methods of orthogonal triangular decomposition and singular value decomposition to construct a data-driven diagnostic signal based on the first control data of the actuator and the first detection data of the sensor under normal operation of the drone. It can comprehensively consider the correlation information between the drone actuator control data and the sensor detection data to construct the data-driven diagnostic signal, which is beneficial to improving the drone fault diagnosis performance.

[0113] In an optional embodiment, each fault diagnosability index is constructed based on the second control data of the actuator and the second detection data of the sensor in each fault situation of the drone, including: constructing each fault diagnosability index based on the target parameter vector, the second control data and the second detection data; wherein the target parameter vector is determined based on the first control data and the first detection data.

[0114] As an example, after obtaining the second control data of the actuator and the first detection data of the sensor under various fault conditions of the drone, and determining the target parameter vector based on the first control data of the actuator and the first detection data of the sensor under normal operation of the drone, the fault diagnosability index of the fault is constructed based on the target parameter vector and the second control data of the actuator and the second detection data of the sensor under each fault condition of the drone, thereby constructing various fault diagnosability indicators.

[0115] For example, after obtaining the target parameter vectors corresponding to the feasible parameter vectors shown in formula (11), based on the set first data driving parameter n, the third input vector is constructed according to the second control data of the actuator under various failure conditions of the UAV shown in formula (3). , according to the second detection data of the sensor in each failure case of the UAV shown in formula (4), the fourth input vector is constructed .

[0116] For each feasible parameter vector, the fault diagnosability index of each fault corresponding to the hth feasible parameter vector is constructed according to the target parameter vector corresponding to the feasible parameter vector and the third input vector and the fourth input vector corresponding to each fault, as shown in Equation (13):

[0117] (13);

[0118] In formula (13), is the fault diagnosability index of the i-th fault corresponding to the h-th feasible parameter vector; is based on The amount of data within is determined.

[0119] The embodiment of the present application constructs various fault diagnosability indicators by determining the target parameter vector based on the second control data of the actuator and the second detection data of the sensor under various fault conditions of the drone, and the first control data of the actuator and the first detection data of the sensor under normal operation of the drone. The target parameter vector can be used to accurately associate the data under normal operation of the drone with the data under various fault conditions of the drone, thereby ensuring the subsequent accurate determination of the data-driven residual and its evaluation function.

[0120] In an optional embodiment, before determining the data-driven residual and the evaluation function of the data-driven residual based on the data-driven diagnostic signal and each fault diagnosability index, it includes: determining whether each fault diagnosability index corresponding to the data-driven diagnostic signal meets the value conditions.

[0121] As an example, when a certain fault occurs in a drone, if the fault diagnosability index value corresponding to the fault exceeds the performance threshold, it will cause the data-driven diagnostic signal to change, indicating that the data-driven diagnostic signal is sensitive to diagnosing the fault and the data-driven diagnostic signal is effective.

[0122] Since multiple data-driven diagnostic signals may be obtained in actual applications, in order to determine the data-driven residual and its evaluation function based on the effective data-driven diagnostic signal, it is possible to first determine whether the various fault diagnosability indicators corresponding to the data-driven diagnostic signal meet the pre-set value conditions. If it is determined that the various fault diagnosability indicators corresponding to the data-driven diagnostic signal meet the value conditions, then continue to determine the data-driven residual and its evaluation function based on the data-driven diagnostic signal and the various fault diagnosability indicators.

[0123] For example, setting performance thresholds for fault diagnosis signals , for the i-th fault, if , then the data drives the diagnostic signal Insensitive to the i-th fault; if , it is believed that the i-th fault of the UAV will cause a data-driven diagnosis signal Change has occurred.

[0124] If the data drives the diagnostic signal For any fault, , at this time the data drives the diagnostic signal Insensitive to any fault, using data-driven diagnostic signals If fault diagnosis is difficult to achieve, remove the data-driven diagnostic signal , the data-driven residual and its evaluation function are no longer determined based on this signal.

[0125] Drive diagnostic signals based on retained data The data-driven residuals determined by the various fault diagnosability indicators are shown in formula (14):

[0126] , (14);

[0127] In formula (14), is the mth data-driven residual at the kth sampling moment; 、 is the target parameter vector corresponding to the mth feasible parameter vector; , represents the total number of data-driven residuals, Less than The total number of .

[0128] Distributed data-driven residuals , In fact, it is a set of data-driven diagnostic signals. This set of data-driven diagnostic signals satisfies: for any two faults, such as the i-th fault and the j-th fault, , there is always at least one fault diagnosability index that satisfies formula (15) or formula (16):

[0129] and (15);

[0130] and (16).

[0131] If it is difficult to find a set of data-driven diagnostic signals that satisfy equation (15) or equation (16), then for at least one pair of faults that satisfy equation (17) or equation (18), such as the i-th fault and the j-th fault, all data-driven diagnostic signals that are sensitive to this at least one pair of faults are:

[0132] (17);

[0133] (18).

[0134] Drive residuals based on distributed data , , the constructed data-driven residual evaluation function is shown in formula (19):

[0135] (19);

[0136] In formula (19), is the mth data-driven residual evaluation function at the kth sampling moment, is the smoothing parameter of the evaluation function, indicating the recent NJ +1 moment sum of squares of residuals, is the evaluation function coefficient, that is, from i J =0 (corresponding to ) superimposed on (correspond ).

[0137] The embodiment of the present application first determines whether the various fault diagnosability indicators corresponding to the data-driven diagnostic signal meet the value conditions, and then determines the data-driven residual and its evaluation function based on the data-driven diagnostic information and the various fault diagnosability indicators, thereby ensuring the effective determination of the data-driven residual and its evaluation function.

[0138] In an optional embodiment, the method of performing eigenmode decomposition on the data-driven residual and extracting the frequency domain features of the eigenmode decomposition signal includes: performing eigenmode decomposition on the data-driven residual using a eigenmode decomposition method based on correlation kurtosis to obtain the eigenmode decomposition signal; and extracting the frequency domain features from the eigenmode decomposition signal.

[0139] As an example, the Correlation Kurtosis-based Mode Decomposition (CKMD) method combines correlation kurtosis and modal decomposition techniques. Its basic idea is to use correlation kurtosis as an optimization criterion to extract the modal components with the maximum correlation kurtosis from a complex signal, thereby better capturing the sudden changes in the signal. The CKMD method was originally applied to fault diagnosis of rotating machinery, such as gears.

[0140] By applying the CKMD method to the UAV fault diagnosis problem, the CKMD method is used to perform eigenmode decomposition on the data-driven residual to obtain the eigenmode decomposition signal, and the frequency domain features are extracted from the eigenmode decomposition signal, so that the frequency domain features of each fault can be accurately extracted.

[0141] For example, suppose the residual modal number is , the filter length is A, the initial number of filters is , the data-driven residual shown in (14) The length is B, and the maximum number of iterations is .

[0142] Initialization iteration count , initialize the frequency band of the data-driven residual to Segment, use a filter, such as a Hanning window smoothing filter, to calculate the Kth segment filter Upper cutoff frequency and lower cutoff frequency , as shown in formula (20):

[0143] (20);

[0144] In formula (20), ; is the sampling frequency of the original signal, the data-driven residual.

[0145] Use convolution operation to obtain the filtered signal , that is, The decomposition mode of the iteration is shown in formula (21):

[0146] (twenty one);

[0147] In formula (21), ; ; , For the The number of decomposition modes of the iteration; * indicates the convolution operation.

[0148] According to the original signal , decomposition mode , update the filter coefficients by optimizing the correlation kurtosis shown in Equation (22):

[0149] (twenty two);

[0150] In formula (22), is the optimized filter function; For the The estimated period of iterations; 、 For the use of The filter used in the iteration The filtered signal, , ; , For the The number of decomposed modes for the iteration.

[0151] After crossing zero The autocorrelation spectrum of At the point where the local maximum is reached, the estimation period is determined :

[0152] (twenty three);

[0153] In formula (23), is the delay used by the next iteration filter, let the autocorrelation spectrum The point that reaches the local maximum corresponds to That is the Estimated period of iterations .

[0154] Iterate 、 and Until , use the current filter coefficients to obtain the mode .

[0155] Calculate according to formula (24) Any two modes and Correlation coefficient , and build The correlation coefficient matrix of

[0156] (twenty four);

[0157] In formula (24), , ; ; and Mode and The mean of .

[0158] Select The two modes with the largest values ​​are calculated based on the estimated period, as shown in Equations (25) and (26):

[0159] (25);

[0160] (26);

[0161] For two modes and , remove the one with smaller correlation kurtosis ( 、 )'s modality (actually the modality and Choose one of the two and select the mode with more significant features). Set . Repeat the above operation to get 、 、 、 、 and , until , and obtain the data-driven residual The characteristic mode decomposition signal of , until the eigenmode decomposition signals of all data-driven residuals are obtained, and the frequency domain features are extracted from the eigenmode decomposition signals.

[0162] And extract the time domain features of the evaluation function. After obtaining the frequency domain features and time domain features, a fault diagnosis decision unit is established based on the frequency domain features and time domain features.

[0163] For example, using the fault diagnosability index ,With the help of the correlation matrix information between different actuator and sensor fault scenarios and data-driven residuals, combined with the value range of the evaluation function under different fault scenarios, the time domain characteristics of the fault are extracted;

[0164] Fault diagnosability index Explain the evaluation function For faults More sensitive. Rudder failure and yaw gyro failure As an example, assuming the number of residual signals is , respectively , according to the fault diagnosability index Establishing Fault Scenarios and Data-Driven Residuals The correlation matrix information between them is shown in Table 1:

[0165] Table 1

[0166]

[0167] According to Table 1, for the fault The fault separation decision is set as follows: If the evaluation function satisfies formula (27), then the rudder fault Alarm. Due to fault diagnosability indicators Indicates a fault Can cause the evaluation function changes, but , For faults Not sensitive, so for fault The fault separation decision is set as follows: If the evaluation function satisfies formula (28), then the yaw gyro fault Call the police.

[0168] (27);

[0169] (28);

[0170] In formulas (27) and (28), Corresponding to the evaluation function under normal operation The maximum value of is used as the fault diagnosis threshold.

[0171] According to the above example, the evaluation function under different fault conditions There may be some differences in the magnitude of and data-driven residuals Satisfying formula (19), that is The amplitude difference of the evaluation function can be used as the time domain feature of the residual under different fault conditions. According to the judgment result of whether each evaluation function exceeds the threshold, the time domain feature of the residual can be extracted.

[0172] Considering that the number of UAV actuators and sensors is large, it may be difficult to achieve fault separation using a single time domain feature. Therefore, the time domain features based on the evaluation function and the frequency domain features based on the eigenmode decomposition signal are used to integrate the design of the fault diagnosis decision unit.

[0173] Since the data-driven residual , Each has The characteristic mode decomposition signal is used to extract the frequency domain characteristics of the fault using the residual mode characteristics under different fault situations.

[0174] For example, if the yaw gyro fails Aileron failure scenario Both cause residual Significant changes and residuals No significant change, set the residual The signal is decomposed using four characteristic modes and recorded as , then use the fault condition and Dataset ,as well as Get the characteristic mode decomposition signal The difference is taken as the frequency domain feature of the residual.

[0175] Assumptions No significant difference, but the failure scenario Down satisfy , fault scenario Down satisfy ,in, The value range of and The value range of There is no overlapping area.

[0176] The above frequency domain features and 、 Both cause residual Significant changes and The time domain features without significant changes are combined to set the fault separation decision as follows: If the evaluation function and the characteristic mode decomposition signal satisfy formula (29), then the yaw gyro fault situation Alarm; If the evaluation function and the characteristic mode decomposition signal satisfy formula (30), then the aileron fault situation Call the police;

[0177] (29);

[0178] (30).

[0179] Extending to all fault scenarios, combining time domain features with frequency domain features, a distributed fault diagnosis decision unit is constructed as shown in Equation (31):

[0180] (31);

[0181] In formula (31), , is the number of failures. Faults can be obtained A fault diagnosis decision unit.

[0182] Finally, the acquired data-driven residual and its evaluation function, characteristic mode decomposition signal and fault diagnosis decision unit are used to establish a fault diagnosis system for UAVs.

[0183] The fault diagnosis system is applied to the real-time fault diagnosis of UAVs. The real-time actuator and sensor data are collected as input and output data. The residual signal is updated using formula (14), the evaluation function is updated using formula (19), the characteristic mode decomposition signal is updated using formula (21), and the fault diagnosis result is obtained using formula (31), i.e., the fault diagnosis decision unit.

[0184] The embodiment of the present application adopts an eigenmode decomposition method based on correlation kurtosis to perform eigenmode decomposition on the data-driven residual to obtain an eigenmode decomposition signal, and extracts frequency domain features from the eigenmode decomposition signal. This can accurately extract the frequency domain features of each fault, which is conducive to further improving the accuracy of drone fault diagnosis.

[0185] Please see Figure 2 , Figure 2A schematic diagram of the structure of a device for establishing a drone fault diagnosis system according to the second embodiment of the present application. The second embodiment of the present application provides a device for establishing a drone fault diagnosis system, wherein the drone is equipped with actuators and sensors. The device comprises: a data-driven diagnostic signal construction module 201 for constructing a data-driven diagnostic signal based on first control data of the actuator and first detection data of the sensor under normal drone operation; a fault diagnosability index construction module 202 for constructing various fault diagnosability indices based on second control data of the actuator and second detection data of the sensor under various drone fault conditions; a residual error and its evaluation function determination module 203 for determining the data-driven residual error and the evaluation function of the data-driven residual error based on the data-driven diagnostic signal and various fault diagnosability indices; a time-frequency domain feature extraction module 204 for performing eigenmode decomposition on the data-driven residual error, extracting frequency domain features of the eigenmode decomposition signal, and extracting time domain features of the evaluation function; and a fault diagnosis system establishment module 205 for establishing a fault diagnosis decision unit based on the frequency domain features and time domain features, thereby establishing a drone fault diagnosis system using the fault diagnosis decision unit.

[0186] In an optional embodiment, the data-driven diagnostic signal is constructed based on the first control data of the actuator and the first detection data of the sensor under normal operation of the drone, including: constructing a data-driven matrix based on the first control data and the first detection data; decomposing a lower triangular matrix from the data-driven matrix; performing singular value decomposition on the target submatrix of the lower triangular matrix to determine the target parameter vector; and constructing the data-driven diagnostic signal based on the target parameter vector, the first control data and the first detection data.

[0187] In an optional embodiment, each fault diagnosability index is constructed based on the second control data of the actuator and the second detection data of the sensor in each fault situation of the drone, including: constructing each fault diagnosability index based on the target parameter vector, the second control data and the second detection data; wherein the target parameter vector is determined based on the first control data and the first detection data.

[0188] In an optional embodiment, the residual and its evaluation function determination module 203 is also used to determine whether the various fault diagnosability indicators corresponding to the data-driven diagnostic signal meet the value conditions before determining the data-driven residual and the evaluation function of the data-driven residual based on the data-driven diagnostic signal and the various fault diagnosability indicators.

[0189] In an optional embodiment, the method of performing eigenmode decomposition on the data-driven residual and extracting the frequency domain features of the eigenmode decomposition signal includes: performing eigenmode decomposition on the data-driven residual using a eigenmode decomposition method based on correlation kurtosis to obtain the eigenmode decomposition signal; and extracting the frequency domain features from the eigenmode decomposition signal.

[0190] In an optional embodiment, the drone includes a fixed-wing drone; the actuators include ailerons, elevators, rudders and throttles; the sensors include one or more of an airspeed sensor, an angle of attack sensor, a sideslip angle sensor, a roll gyroscope, a pitch gyroscope, a yaw gyroscope, an inertial measurement unit, an acceleration sensor, a three-axis gyroscope, a global navigation satellite system and an altitude sensor.

[0191] The implementation process of the functions and effects of each module in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0192] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of a drone fault diagnosis system provided in the third embodiment of the present application. The third embodiment of the present application provides a drone fault diagnosis system, which is established according to the drone fault diagnosis system establishment method described in the first embodiment of the present application and can achieve the same beneficial effects as the first embodiment.

[0193] The fourth embodiment of the present application provides a computer program product, which includes instructions. When the instructions are executed by a computer, the computer implements the method described in the first embodiment of the present application and can achieve the same beneficial effects.

[0194] The method described in the first embodiment of the present application can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in each embodiment of the present application are executed in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user device, a core network device, an OAM (Open Application Model), or other programmable device.

[0195] A computer program or instruction can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, a computer program or instruction can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. A computer-readable storage medium can be any available medium that can be accessed by a computer, or a data storage device such as a server or data center that integrates one or more available media. Available media can be magnetic media, such as floppy disks, hard disks, or magnetic tapes; optical media, such as digital video disks; or semiconductor media, such as solid-state drives. The computer-readable storage medium can be volatile or non-volatile, or can include both volatile and non-volatile types of storage media.

[0196] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device provided in the fifth embodiment of the present application. The fifth embodiment of the present application provides an electronic device 40, comprising a processor 401, a memory 402, and a computer program stored in the memory 402 and configured to be executed by the processor 401; when the processor 401 executes the computer program, it implements the method described in the first embodiment of the present application and can achieve the same beneficial effects as described above.

[0197] In which, when the processor 401 reads the computer program from the memory 402 through the bus 403 and executes the computer program, it can implement the method of any embodiment included in the method described in the first embodiment of the present application.

[0198] Processor 401 can process digital signals and can include various computing architectures, such as a complex instruction set computer architecture, a reduced instruction set computer architecture, or an architecture that implements a combination of multiple instruction sets. In some examples, processor 401 can be a microprocessor.

[0199] The memory 402 can be used to store instructions executed by the processor 401 or data related to the execution of instructions. These instructions and / or data may include code for implementing some or all functions of one or more modules described in the embodiments of this application. The processor 401 of the embodiment of the present disclosure can be used to execute the instructions in the memory 402 to implement the method described in the first embodiment of this application. The memory 402 includes dynamic random access memory, static random access memory, flash memory, optical storage, or other memory known to those skilled in the art.

[0200] The sixth embodiment of the present application provides a computer-readable storage medium, which includes a stored computer program; wherein, when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method described in the first embodiment of the present application, and can achieve the same beneficial effects as the method described in the first embodiment of the present application.

[0201] In summary, the embodiments of the present application provide a method, system, device and medium for establishing a drone fault diagnosis system. In the method for establishing a drone fault diagnosis system, the drone is configured with an actuator and a sensor; the method includes: constructing a data-driven diagnostic signal based on the first control data of the actuator and the first detection data of the sensor under normal operation of the drone; constructing various fault diagnosability indicators based on the second control data of the actuator and the second detection data of the sensor under various fault conditions of the drone; determining the data-driven residual and the evaluation function of the data-driven residual based on the data-driven diagnostic signal and the various fault diagnosability indicators; performing characteristic mode decomposition on the data-driven residual, extracting the frequency domain features of the characteristic mode decomposition signal, and extracting the time domain features of the evaluation function; establishing a fault diagnosis decision unit based on the frequency domain features and the time domain features, so as to establish a fault diagnosis system for the drone using the fault diagnosis decision unit. The embodiment of the present application, through an algorithmic idea based on a data-driven algorithm, processes the first control data of the actuator and the first detection data of the sensor under normal operation of the drone, as well as the second control data of the actuator and the second detection data of the sensor under various fault conditions of the drone, to construct a data-driven diagnostic signal and various fault diagnosability indicators. According to the data-driven diagnostic signal and various fault diagnosability indicators, the data-driven residual and the evaluation function of the data-driven residual are determined, and based on the frequency domain characteristics of the characteristic mode decomposition signal of the data-driven residual and the time domain characteristics of the evaluation function, a fault diagnosis decision unit is established. By using the fault diagnosis decision unit to establish a fault diagnosis system for the drone, the fault diagnosis system of the drone can be enabled to comprehensively analyze the time-frequency domain characteristics of the fault to perform fault diagnosis including minor faults, thereby helping to improve the accuracy of drone fault diagnosis.

[0202] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0203] In addition, the functional modules in each embodiment of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0204] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard drives, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.

[0205] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A method for establishing a UAV fault diagnosis system, characterized in that: The drone is configured with an actuator and a sensor; the method comprises: constructing a data-driven diagnostic signal according to the first control data of the actuator and the first detection data of the sensor when the drone is operating normally; constructing respective fault diagnosability indicators according to the second control data of the actuator and the second detection data of the sensor when respective faults of the drone occur; Determining a data-driven residual and an evaluation function of the data-driven residual according to the data-driven diagnostic signal and the respective fault diagnosability indicators; Performing eigenmode decomposition on the data-driven residual, extracting frequency domain features of the eigenmode decomposition signal, and extracting time domain features of the evaluation function; Establishing a fault diagnosis decision unit based on the frequency domain characteristics and the time domain characteristics, and establishing a fault diagnosis system for the UAV using the fault diagnosis decision unit; Wherein, the fault diagnosis decision unit is: ； 、 ,...are faults respectively The corresponding evaluation function of each data-driven residual; 、 ,...are the evaluation functions under normal operation 、 , ..., the fault diagnosis threshold, the time domain features are based on each evaluation function 、 , ...whether they exceed their respective fault diagnosis thresholds 、 ... the extracted statistical features, including amplitude features extracted from the evaluation function when any evaluation function exceeds the fault diagnosis threshold of the evaluation function itself; 、 ,...are faults respectively The frequency domain characteristics of the modal decomposition signals of the corresponding data-driven residuals; [ ]、[ ], ... respectively 、 , ... the corresponding value interval; , is the total number of the respective data-driven residuals; , is the number of failures.

2. The method according to claim 1, characterized in that The step of constructing a data-driven diagnostic signal according to the first control data of the actuator and the first detection data of the sensor when the drone is operating normally includes: constructing a data driving matrix according to the first control data and the first detection data; Decomposing a lower triangular matrix from the data-driven matrix; Performing singular value decomposition on a target submatrix of the lower triangular matrix to determine a target parameter vector; The data-driven diagnostic signal is constructed according to the target parameter vector, the first control data, and the first detection data.

3. The method according to claim 1, characterized in that The constructing of each fault diagnosability index according to the second control data of the actuator and the second detection data of the sensor in each case where the UAV has a fault includes: The respective fault diagnosability indicators are constructed according to a target parameter vector, the second control data and the second detection data; wherein the target parameter vector is determined according to the first control data and the first detection data.

4. The method according to claim 1, wherein Before determining the data-driven residual and the evaluation function of the data-driven residual according to the data-driven diagnostic signal and the various fault diagnosability indicators, the method includes: Determine whether the respective fault diagnosability indicators corresponding to the data-driven diagnostic signal meet a value condition.

5. The method according to claim 1, wherein The performing eigenmode decomposition on the data-driven residual to extract frequency domain features of the eigenmode decomposition signal includes: Performing eigenmode decomposition on the data-driven residual using an eigenmode decomposition method based on correlation kurtosis to obtain the eigenmode decomposition signal; The frequency domain features are extracted from the eigenmode decomposition signal.

6. The method according to any one of claims 1 to 5, characterized in that The UAV includes a fixed-wing UAV; The actuators include ailerons, elevators, rudders and throttles: The sensor includes one or more of an airspeed sensor, an angle of attack sensor, a sideslip angle sensor, a roll gyroscope, a pitch gyroscope, a yaw gyroscope, an inertial measurement unit, an acceleration sensor, a three-axis gyroscope, a global navigation satellite system, and an altitude sensor.

7. A UAV fault diagnosis system, characterized in that: The system is established according to the method for establishing a drone fault diagnosis system according to any one of claims 1 to 6.

8. A computer program product, characterized in that The computer program product comprises instructions which, when executed by a computer, cause the computer to implement the method according to any one of claims 1 to 6.

9. An electronic device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor; when the processor executes the computer program, the method according to any one of claims 1 to 6 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program; wherein, when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 6.

Images