Aircraft aerodynamic parameter identification method based on improved dual-innovation EKF

Abstract

The invention discloses an aircraft aerodynamic parameter identification method based on an improved dual-innovation EKF, and the method comprises the steps: carrying out the wind tunnel test or numerical simulation of an aircraft, obtaining a test estimation value of a to-be-identified aerodynamic parameter, and obtaining an aerodynamic model; selecting a state variable, an input variable and an observation variable in the identification process, wherein the state variable comprises all to-be-identified aerodynamic parameters; in a balancing state, a dual-innovation extended Kalman filtering method based on a forgetting factor is adopted to carry out aerodynamic parameter identification, and an identification estimation value of a to-be-identified aerodynamic parameter is obtained; and judging whether the identification estimation value is superior to the test estimation value or not in another balancing state. The method is applied to the field of parameter identification, the historical measurement data information is fully utilized, error accumulation interference is reduced by introducing the forgetting factor, accurate identification of the aerodynamic parameters of the aircraft can be achieved without depending on the initial value, the problem that the initial value is sensitive in the identification process through a traditional method is solved, and the identification precision is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of parameter identification, and in particular to an aircraft aerodynamic parameter identification method based on an improved double-innovation EKF. Background Art

[0002] The accurate construction of the aerodynamic model of UAV is the premise and basis for achieving its high-precision control, and the construction of this aerodynamic model depends on the determination of the aerodynamic parameters of the UAV. At present, the determination of the aerodynamic parameters of UAV is mainly carried out through the following three methods: wind tunnel test, numerical calculation and flight identification test. For many years, wind tunnel test has been the main method for studying the aerodynamic characteristics of UAVs, which can provide aerodynamic parameter data for newly developed UAVs. However, the wind tunnel test itself has inherent limitations such as tunnel wall interference, bracket interference and scale similarity. It cannot fully simulate the real flight environment of UAVs and there are non-negligible differences. In addition, wind tunnel tests are expensive, have a long cycle, and require multiple corrections.

[0003] The numerical calculation method is based on computer technology and aerodynamic theory. This method has a short calculation time and low cost. It can intuitively represent the flow field characteristics of the UAV during flight and is mainly used in the aerodynamic layout design stage. However, there is an error between the mathematical model established by this method and the actual model, and its calculation results usually need to be tested.

[0004] Parameter identification (or parameter estimation) refers to the process of calculating or correcting certain parameters in the model through system input and output data based on the existing mathematical model of the system, so as to correct the mathematical model. The flight test method based on aerodynamic parameter identification can be mutually verified with the results obtained from wind tunnel tests or numerical calculations, thereby providing an accurate aerodynamic parameter model for the UAV design process. For the determination of the aerodynamic parameter model of small and medium-sized UAVs, the flight test method based on aerodynamic parameter identification has the advantages of low test cost, high identification accuracy, and good applicability.

[0005] At present, the main methods for UAV aerodynamic parameter identification include maximum likelihood method, Kalman filter and its improved algorithm, intelligent optimization algorithm (such as particle swarm optimization, genetic algorithm, etc.), and parameter identification method based on neural network developed in recent years. Intelligent optimization algorithm and parameter identification method based on neural network have the advantages of high identification accuracy and insensitivity to measurement noise. However, the parameter identification research using these methods is mainly carried out by simulation calculation, and has not been verified by combining the real flight data of UAV, and is rarely used in engineering practice.

[0006] Kalman filtering and its improved algorithms are a type of algorithm that is mature and widely used in practical engineering. At present, research on Kalman filtering aerodynamic parameter identification algorithms mainly focuses on algorithm improvement and identification accuracy improvement. The most common Kalman filtering algorithm in UAV aerodynamic parameter identification is the Extended Kalman Filter (EKF) algorithm. However, the traditional EKF algorithm is very sensitive to the initial value selection of the parameters to be identified, which usually leads to problems such as low identification accuracy and limited application scenarios. Summary of the invention

[0007] In view of the above-mentioned deficiencies in the prior art, the present invention provides an aircraft aerodynamic parameter identification method based on an improved double-innovation EKF, which, compared with the traditional extended Kalman filter method, is independent of initial values ​​and has higher identification accuracy.

[0008] To achieve the above object, the present invention provides an aircraft aerodynamic parameter identification method based on an improved double-innovation EKF, comprising the following steps:

[0009] Step 1: Perform a wind tunnel test or numerical simulation on the aircraft to obtain test estimated values ​​of all aerodynamic parameters to be identified of the aircraft, and obtain an aerodynamic model for aerodynamic parameter identification of the aircraft;

[0010] Step 2, selecting state variables, input variables and observation variables in the identification process, and initializing the state variables, input variables and estimation error covariance matrix, wherein the state variables at least include all the aerodynamic parameters to be identified;

[0011] Step 3, according to the aerodynamic model and the initial values ​​of the state variables, and the actual measured values ​​of the observed variables, in a trim state of the aircraft, a double-innovation extended Kalman filter method based on a forgetting factor is used to perform aerodynamic parameter identification to obtain an identification estimate of the aerodynamic parameter to be identified;

[0012] Step 4: determining whether the identified estimated value is better than the tested estimated value in another trim state of the aircraft:

[0013] If so, outputting the identification estimate;

[0014] Otherwise, reselect the state variables, input variables and observation variables in the identification process and return to step 2.

[0015] In one embodiment, in step 3, during the process of aerodynamic parameter identification using the double-innovation extended Kalman filter method based on the forgetting factor, the prediction process is:

[0016]

[0017]

[0018] in, represents the prior estimate of the state variable at time k, f(·) is the state transfer function, represents the posterior estimate of the state variable at time k-1, u k-1 represents the input value of the input variable at time k-1, P k - represents the prior estimated covariance matrix at time k, P k-1 represents the posterior estimated covariance matrix at time k-1, Φ Ak represents the state transfer matrix at time k, Q k-1 represents the process noise covariance matrix at time k-1, and T represents the transpose of the matrix.

[0019] In one embodiment, in step 3, during the process of aerodynamic parameter identification using the double-innovation extended Kalman filter method based on the forgetting factor, the updating process is:

[0020]

[0021]

[0022]

[0023] Among them, K k and H k They represent the Kalman gain matrix and observation matrix at time k, R k represents the observation noise covariance matrix at time k, x k represents the state vector at time k, λ1 and λ2 represent the element values ​​in the forgetting factor matrix, z k and z k-1 denote the observation vectors at time k and time k-1 respectively, h(·) denotes the observation function, K k-1 represents the Kalman gain matrix at time k-1, and I represents the identity matrix.

[0024] In one embodiment, in step 2, during the process of initializing the state variables, the test estimated value is used as the initial value of the aerodynamic parameter to be identified in the state vector.

[0025] In one embodiment, in step 2, during the process of initializing the state variables, a random number is collected as the initial value of the aerodynamic parameter to be identified in the state vector; or

[0026] The initial value of the aerodynamic parameter to be identified in the state vector is calculated based on the test estimated value.

[0027] In one embodiment, in step 3, the process of obtaining the actual measured value of the observed variable is:

[0028] A flight test is performed on the aircraft, and test data corresponding to the observed variables are recorded. The actual measured values ​​are obtained after preprocessing the test data.

[0029] In one embodiment, the preprocessing includes but is not limited to outlier removal and compatibility checking.

[0030] Compared with the prior art, the present invention has the following beneficial technical effects:

[0031] In the process of aerodynamic parameter identification of an aircraft, the present invention makes full use of historical measurement data information and introduces a forgetting factor to reduce error accumulation interference. It can realize accurate identification of aerodynamic parameters of an aircraft without relying on initial values, solves the initial value sensitivity problem in the identification process using the traditional extended Kalman filter method, improves the identification accuracy, and can be effectively applied to aerodynamic parameter identification in situations such as uncertain aircraft aerodynamic models and verification of wind tunnel tests or numerical calculation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0033] Figure 1 Schematic diagram of the identification process of the aircraft aerodynamic parameter identification method in an embodiment of the present invention;

[0034] Figure 2 This is the schematic diagram of the traditional EKF algorithm;

[0035] Figure 3 Schematic diagram of the FDI-EKF method in an embodiment of the present invention;

[0036] Figure 4 This is a schematic diagram of the structure of a fixed-wing UAV in an example of an embodiment of the present invention;

[0037] Figure 5 Schematic diagram of the change of input quantity over time in an example of an embodiment of the present invention, wherein: (a) is a schematic diagram of the change of elevator deflection angle, (b) is a schematic diagram of the change of engine thrust;

[0038] Figure 6 This is a schematic diagram of the flight trajectory of a drone in an example of an embodiment of the present invention;

[0039] Figure 7 This is a schematic diagram of the flight altitude of a drone in an example of an embodiment of the present invention;

[0040] Figure 8 It is a schematic diagram of measured data of observation data in an example of an embodiment of the present invention;

[0041] Fig. 9 This is a schematic diagram of observed quantity data after processing in an example of an embodiment of the present invention;

[0042] Fig.10 is the aerodynamic coefficient C in the embodiment of the present invention L0 Schematic diagram of the identification results;

[0043] Fig.11 is the aerodynamic coefficient C in the embodiment of the present invention Lα Schematic diagram of the identification results;

[0044] Fig.12 is the aerodynamic coefficient C in the embodiment of the present invention Lδe Schematic diagram of the identification results;

[0045] Fig.13 is the aerodynamic coefficient C in the embodiment of the present invention D0 Schematic diagram of the identification results;

[0046] Fig.14 is the aerodynamic coefficient C in the embodiment of the present invention m0 Schematic diagram of the identification results;

[0047] Fig.15 is the aerodynamic coefficient C in the embodiment of the present invention mα Schematic diagram of the identification results;

[0048] Fig.16 is the aerodynamic coefficient C in the embodiment of the present invention mδe Schematic diagram of the identification results;

[0049] Fig.17 is the aerodynamic coefficient C in the embodiment of the present invention mq Schematic diagram of the identification results;

[0050] Fig.18 Statistical box plots of the error ranges of aerodynamic parameter identification under EKF and FDI-EKF in the examples of the embodiments of the present invention, wherein: (a) is a statistical box plot of the error ranges of aerodynamic parameter identification under EKF, and (b) is a statistical box plot of the error ranges of aerodynamic parameter identification under FDI-EKF;

[0051] Fig.19 It is a schematic diagram showing the comparison of the error ranges of the aerodynamic parameters identified under EKF and FDI-EKF in an example of an embodiment of the present invention;

[0052] Fig. 20 Schematic diagram of comparison of observation output before and after aerodynamic parameter identification in an example of an embodiment of the present invention.

[0053] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0055] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0056] In addition, in the present invention, descriptions such as "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0057] In the present invention, unless otherwise clearly specified and limited, the terms "connection", "fixation", etc. should be understood in a broad sense. For example, "fixation" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, an electrical connection, a physical connection, or a wireless communication connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0058] In addition, the technical solutions between the various embodiments of the present invention can be combined with each other, but it must be based on the fact that ordinary technicians in the field can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0059] like Figure 1The figure shows a method for identifying aerodynamic parameters of an aircraft based on an improved dual-innovation EKF disclosed in this embodiment, which mainly includes the following steps:

[0060] Step 1, selecting an aerodynamic model for aircraft aerodynamic parameter identification based on the information before identification visual inspection and designing an aircraft flight test, while conducting a wind tunnel test or numerical simulation on the aircraft to obtain test estimated values ​​of all aerodynamic parameters to be identified of the aircraft;

[0061] Step 2, selecting state variables, input variables and observation variables in the identification process, and initializing the state variables, input variables and estimation error covariance matrix, wherein the state variables at least include all aerodynamic parameters to be identified;

[0062] Step 3, according to the aerodynamic model and the initial values ​​of the state variables, and the actual measured values ​​of the observed variables obtained from the designed flight test, in a trim state of the aircraft, a double-innovation extended Kalman filter method based on the forgetting factor is used to perform aerodynamic parameter identification to obtain the identification estimation value of the aerodynamic parameter to be identified;

[0063] Step 4: Verify the identification estimate in another trim state of the aircraft, that is, determine whether the identification estimate is better than the test estimate:

[0064] If so, output the identification estimate;

[0065] Otherwise, reselect the state variables, input variables and observation variables in the identification process and return to step 2.

[0066] In step 2, the state variables may also include other parameters such as the speed and acceleration of the aircraft. In the process of initializing the state variables, the preferred implementation is to use the test estimate as the initial value of the aerodynamic parameter to be identified in the state vector, or to use a random number as the initial value of the aerodynamic parameter to be identified in the state vector, or to calculate the initial value of the aerodynamic parameter to be identified in the state vector based on the test estimate, for example, multiplying the test estimate by a preset coefficient as the initial value. Input variables may include thrust, elevator deflection, etc., and the input variables may be set based on experience or randomly during initialization.

[0067] In step 3, the actual measured values ​​of the observed variables are obtained by performing a flight test on the aircraft, recording the test data corresponding to the observed variables, and preprocessing the test data to obtain the actual measured values. The preprocessing methods for the test data include outlier elimination, compatibility testing, etc.

[0068] In step 4, the process of judging whether the identification estimate is better than the test estimate is as follows: after substituting the identification estimate and the test estimate into the aerodynamic model respectively, aerodynamic performance simulation is performed. If the simulation result of the identification estimate is better than the simulation result of the test estimate, the identification estimate is judged to be better than the test estimate. Otherwise, the parameter type or initial value of the state variable, input variable and observation variable is reset and the parameter identification is performed again. This verifies the reliability of the identification method in this embodiment under different trim flight conditions and the effectiveness of the identification model.

[0069] refer to Figure 2 , in the traditional EKF algorithm, for the nonlinear system under study, the state equation and observation equation can be expressed as:

[0070]

[0071] Among them, x k represents the system state vector at time k, and f(·) is the state transfer function, which passes through the system state vector x at time k-1. k-1 and the input vector u k-1 represents the state vector x at time k k ;z k represents the system observation vector at time k; h(·) is the observation function, which is calculated by the state vector x at time k. k Represents the observation vector z k , w k 、v k represent the process noise and measurement noise of the system respectively.

[0072] The prediction process of the traditional EKF algorithm can be expressed as:

[0073]

[0074]

[0075] in, represents the prior estimate of the state variable at time k, represents the posterior estimate of the state variable at time k-1, u k-1 represents the input value of the input variable at time k-1, represents the prior estimated covariance matrix at time k, P k-1 represents the posterior estimated covariance matrix at time k-1, Φ Ak represents the state transfer matrix at time k, Q k-1 represents the process noise covariance matrix at time k-1, and T represents the transpose of the matrix.

[0076] The update process of the traditional EKF algorithm can be expressed as:

[0077]

[0078]

[0079]

[0080] Among them, K k and H k They represent the Kalman gain matrix and observation matrix at time k, R k represents the observation noise covariance matrix at time k, and is defined as is the new information, and I represents the identity matrix.

[0081] When solving nonlinear problems, the traditional EKF algorithm generally linearizes the system by Taylor expansion. This process is easy to implement and efficient. However, the disadvantage is that the accuracy is not high when using Taylor expansion approximation, which easily causes error accumulation. At the same time, the algorithm is sensitive to initial value errors. The aircraft dynamics system has strong nonlinearity, and the aerodynamic parameter errors obtained by using the traditional EKF algorithm are generally large. Therefore, it is necessary to improve the traditional EKF algorithm to improve the adaptability of the algorithm when used to solve strong nonlinear problems, thereby improving the accuracy of aircraft aerodynamic parameter identification based on the algorithm.

[0082] It can be seen from formula (5) that when solving the state posterior estimation value at time k, the traditional EKF algorithm mainly uses the single innovation at time k for estimation. In order to make full use of the measurement data information at the historical moments, combined with the multi-innovation theory, the single innovation in the state posterior estimation process is used as Expanded to multiple innovations E(p,k), that is:

[0083]

[0084] At the same time, the Kalman gain matrix K k Expanded to K(p,k), it can be expressed as:

[0085] K(p,k)=[K k K k-1 … K k-p+1 ] (8)

[0086] In the process of aerodynamic parameter identification of aircraft, new measurement data has more reference value for state estimation, while historical measurement data has less reference value and is prone to cumulative interference, resulting in a large deviation between the estimated value and the true value. Therefore, the forgetting factor is introduced to weaken the correction effect of historical measurement data in the recursive solution process, thereby reducing the error accumulation interference and improving the accuracy of aerodynamic parameter identification. The forgetting factor matrix σ is multiplied by the Kalman gain matrix K(p,k) after multi-new information expansion. The weight of the new and old measurement data for the estimation process is adjusted by modifying the forgetting factor matrix σ. σ can be expressed as:

[0087]

[0088] As the number of new information increases, the algorithm calculation amount and calculation time will also increase, and too much historical data will lead to error accumulation interference problems. Therefore, in the process of aircraft aerodynamic parameter identification, this embodiment uses new information at time k and k-1, that is, double-new information extended Kalman filtering based on forgetting factor.

[0089] refer to Figure 3 , the prediction process of the double innovation extended Kalman filter method based on the forgetting factor (hereinafter referred to as "FDI-EKF") in this embodiment is:

[0090]

[0091]

[0092] The updating process can be expressed as:

[0093]

[0094]

[0095]

[0096] The FDI-EKF method in this embodiment makes full use of historical measurement data information and introduces a forgetting factor to reduce error accumulation interference. It can achieve accurate identification of aircraft aerodynamic parameters without relying on initial values, and solves the initial value sensitivity problem in the identification process using the traditional EKF algorithm, thereby improving the identification accuracy of aircraft aerodynamic parameters.

[0097] The following further describes the aircraft aerodynamic parameter identification method and the FDI-EKF method in this embodiment with reference to specific examples.

[0098] by Figure 4 The fixed-wing UAV shown in the figure is the research object. The fixed-wing UAV adopts a simple fuselage design. The trapezoidal fuselage integrates the swept main wing, V-shaped tail and the upper air intake. Its relevant structural parameters are shown in Table 1:

[0099] Table 1 Structural parameters

[0100]

[0101] Based on the longitudinal dynamics model of the UAV, aerodynamic parameter identification is performed to identify the aerodynamic parameters related to the lift coefficient, drag coefficient, and pitch moment coefficient of the fixed-wing UAV. In the body coordinate system, the dynamic model of the longitudinal plane of the UAV can be expressed as:

[0102]

[0103] in:

[0104]

[0105] In equations (15) and (16), u and w are the x- and z-axis components of the UAV velocity in the body coordinate system, q is the pitch velocity, θ and α are the pitch angle and angle of attack, h and m are the height and mass of the aircraft, respectively. V a is the airspeed, J y is the moment of inertia of the y-axis, S and c are the ratio of the wing area and the average chord length, C L , C D and C m are lift coefficient, drag coefficient and pitch moment coefficient respectively, C Lα , C Dα , C mα is the stability derivative, C Lq , C Dq , C mq is the damping derivative, C Lδe , C Dδe , C mδe is the main control derivative;

[0106] Since the lift coefficient and drag coefficient of the designed UAV satisfy the following relationship:

[0107]

[0108] Therefore, in this example, the aerodynamic parameters C L0 , C Lα , C Lδe , C D0 , C m0 , C mα , C mq , C mα To identify.

[0109] In the identification process, the state variables include the longitudinal motion state of the aircraft and the aerodynamic parameters to be estimated, which can be expressed as:

[0110]

[0111] in:

[0112] x=[uw θ q] T (19)

[0113] p=[C L0 C Lα C Lδe C D0 C m0 Cmα C mδe C mq ] T (20)

[0114] Assume that the aerodynamic parameters to be estimated do not change with time and motion state, so Combining the dynamic model of formula (15), the state model in the parameter identification process can be obtained as follows:

[0115]

[0116] It can be further expressed as:

[0117]

[0118] Therefore, in the prediction equations of EKF and FDI-EKF, the prior estimate of the state quantity at the kth moment is It can be expressed as:

[0119]

[0120] in, It represents the posterior estimate of the state quantity at the k-1th moment, and dt represents the discrete time step.

[0121] State transition matrix Φ Ak It can be approximated by Taylor expansion:

[0122] Φ Ak =I+A k ·dt (24)

[0123] Among them, A k The Jacobean matrix representing the state model at the kth moment can be expressed as:

[0124]

[0125] The input variables mainly include elevator deflection angle and engine thrust, which can be expressed as:

[0126] in=[δ e δ T ] T (26)

[0127] The observed quantities are the x-axis velocity, z-axis velocity, pitch angle, pitch angular velocity, x-axis acceleration, z-axis acceleration, and pitch angular acceleration under the body axis system, that is:

[0128]

[0129] The observation equation can be expressed as:

[0130]

[0131] Therefore, the Jacobean matrix of the observation model at time k is:

[0132]

[0133] A k and H k The coefficients in the matrix are shown in Table 2:

[0134] Table 2 Expressions of coefficients

[0135]

[0136]

[0137] Taking the aerodynamic coefficients calculated by CFD software as their true values, they can be expressed as:

[0138]

[0139] A posteriori initial values ​​for the FDI-EKF method The deviation between the initial value and the true value of the aerodynamic parameters during the simulation is represented by η, that is, Movement status Determine the trim status of the aircraft in straight and level flight conditions.

[0140] When the deviation between the initial value of the aerodynamic parameter and the actual value η = 100%, Under the aerodynamic parameter conditions, the motion state when the flight speed is 34m / s is obtained, that is, u0=51.01m / s, w0=1.88m / s, θ0=0.04rad, q0=0. Therefore, the initial value of the posterior estimate is It can be expressed as:

[0141]

[0142] The initial value of the estimated error covariance matrix represents the deviation between the initial value of the state estimate and the true value. To speed up the convergence of the estimated value, it can be set to:

[0143] P0=diag[10;10;10;10;10;10;10;10;10;10;10;10]×10 (32)

[0144] Forgetting factor λ1=1,λ2=0.1.

[0145] The observation data is obtained from the measurement data in the UAV flight test after preprocessing. Among them, the north velocity v n and ground speed v dMeasured by the GNSS module, the measurement error is 0.1m / s, the pitch angle θ and pitch angular velocity q are measured by the gyroscope, the measurement errors are 0.1rad and 0.01rad / s respectively, the motion acceleration a xm and a zm Measured by accelerometer, the measurement error is 0.01m / s 2 , according to the measurement errors of each sensor, the measurement noise covariance matrix R can be determined, that is:

[0146] R=diag[100;100;100;1;1;1;1]×10 -4 (33)

[0147] Q is the process noise covariance matrix, which represents the error caused by various approximate calculations during the simulation calculation process. Since the error is small, it can be set to:

[0148]

[0149] The input variables mainly include elevator deflection angle and engine thrust. The initial value of elevator deflection angle is -2.98° obtained by trimming in the above-mentioned fixed straight and level flight state. The input form of elevator deflection is generally a "3211" periodic signal. Compared with the "3211" periodic signal, the "211" form has the advantages of high efficiency and strong anti-interference. This example designs a "211" periodic signal with an amplitude of 0.5° and a step time interval of 1s. The initial value of the engine thrust is 41.01N through trim calculation. The thrust size remains unchanged during the simulation process. The input quantity changes with time as shown in the figure. Figure 5 shown.

[0150] The use of FDI-EKF identification method for UAV longitudinal aerodynamic parameter identification requires the combination of actual flight test measurement data. Therefore, it is necessary to design a UAV flight test and select measurement data of the corresponding time period, which are then converted into observation data required for the parameter identification process after preprocessing.

[0151] The test system includes a test prototype, a digital radio, a ground control station, etc. The test prototype control system uses the RX-0501 autopilot, which integrates sensors such as a three-axis gyroscope, a three-axis accelerometer, a three-axis magnetometer, an airspeed meter, a barometric altimeter, and a GNSS navigation module. The onboard processor outputs corresponding pulse width modulation signals based on the information from each sensor to drive the onboard servo and control the flight of the aircraft. The digital radio is used to provide a digital transmission link between the control system and the ground control station, so that the ground control station can monitor the status of the aircraft in real time and send corresponding control instructions.

[0152] The test process is as follows: the test prototype first takes off along the center of the runway, then climbs to an altitude of 140m and maintains a level flight along the predetermined route. During this period, the V-shaped tail rudder deflection command is set so that the equivalent elevator deflection angle is Figure 5 (a) shows the variation (the flight measurement data in this time period will be used for parameter identification), followed by changes in flight altitude and level flight state, and finally landing and recovery of the UAV. In this process, the flight trajectory and altitude changes of the UAV in the time period of t = 140-200s are shown in the figure below: Figure 6 , Figure 7 shown.

[0153] In order to reduce the test error and improve the identification accuracy, we first need to select data with a relatively stable and small lateral heading and a certain range of longitudinal variation during the flight of the drone to avoid the coupling effect of the longitudinal and lateral heading channels. In this example, the longitudinal channel data of the drone in level flight during the experiment are selected, which are the north velocity v measured by the GNSS module and the north velocity v measured by the GNSS module. n and ground speed v d , the pitch angle θ and pitch angular velocity q measured by the gyroscope, and the motion acceleration a measured by the accelerometer xm and a zm After preprocessing the data by removing outliers and performing compatibility tests, we get Figure 8 Measured data shown.

[0154] The measured data is converted into the observed data required for the identification process, where the experimentally measured velocity v n 、v d It is the data in the reference coordinate system, which needs to be converted into the velocity u and w in the body coordinate system; acceleration a xm 、a zm Although the accelerometer is measured in the body coordinate system, the initial calibration does not take into account the gravity acceleration, so it is necessary to xm 、a zm Add the acceleration of gravity to get the actual motion acceleration a in the body coordinate system. x 、a z ; The attitude angle and angular velocity θ, q in the measured data are all obtained in the body coordinate system, and no conversion is required; since the pitch angular acceleration cannot be directly measured by the experimental equipment, the pitch angular velocity q needs to be calculated using the central difference method; finally, the observation data required for the identification process is obtained as follows Fig. 9 Show.

[0155] When the deviation between the initial value of the aerodynamic parameter and the true value η = 100%, the aerodynamic parameter identification results obtained by EKF and FDI-EKF are compared with the true value as shown in the figure. Figures 10 to 17 And as shown in Table 3.

[0156] Table 3 Aerodynamic parameter identification results

[0157]

[0158]

[0159] From the above simulation results, it can be seen that both EKF and FDI-EKF can achieve good identification of aerodynamic parameters of fixed-wing UAVs in the presence of process noise, measurement noise, and when the initial value deviation of the simulation is 100%. Using FDI-EKF for identification, the deviation of each aerodynamic parameter from the true value can be reduced from the initial 100% to less than 10%, where C L0 and C Lδe The identification errors of are relatively large, which are 5.170% and 8.865% respectively, while the identification errors of other aerodynamic parameters are all less than 5%; for the identification of the same aerodynamic parameters, the identification errors obtained using FDI-EKF are smaller than those obtained under traditional EKF.

[0160] Under the same elevator deflection angle and engine thrust input, white noise is randomly generated and Monte Carlo simulation analysis is performed 1000 times. When the initial value deviation of aerodynamic parameters η = 100%, the statistical box plots of the aerodynamic parameter identification error range under EKF and FDI-EKF are shown in Figure 2. Fig.18 shown.

[0161] During 1000 Monte Carlo simulations, the comparison of the error range of each aerodynamic parameter identification under EKF and FDI-EKF is shown in the figure below. Fig.19 As shown. Fig.19 It can be seen that in 1000 Monte Carlo simulations, except for C mδe Except for the larger identification error range of C, the identification error range of each aerodynamic parameter obtained by FDI-EKF is smaller than that obtained by EKF. L0 , C Lδe and C m0 The maximum errors of the identification results obtained by using EKF are 13.348%, 10.924% and 8.379%, respectively, while the maximum errors obtained by using FDI-EKF are 10.120%, 5.388% and 2.479%, respectively. Compared with the traditional EKF algorithm, the maximum errors of the three aerodynamic coefficients in the identification method of the present invention are reduced by 24.183%, 50.677% and 70.414%, respectively. This shows that the identification method of the present invention can significantly reduce the sensitivity of the identification process to the initial value deviation, thereby improving the identification accuracy.

[0162] In order to verify the applicability of the identification method of the present invention to any flight state of the UAV, other balancing motion states x0 are taken, such as x0 = [51.01; 1.88; 0.04; 0], that is, u0 = 51.01m / s, w0 = 1.88m / s, θ0 = 0.04rad, q0 = 0. In this motion state, the observed output results and the true values ​​before and after the aerodynamic parameter identification are compared, as shown in FIG. Fig. 20 As shown, from Fig. 20 It can be seen that compared with the model output before identification, the output results after identification have smaller deviations from the test measurement results, indicating that the identification model is applicable and effective for different flight states.

[0163] In summary, the aerodynamic parameter identification method based on FDI-EKF in the present invention makes full use of historical measurement data information compared to the traditional EKF algorithm, and introduces the forgetting factor to reduce the error accumulation interference. By designing the UAV flight test, the aerodynamic parameter identification was performed using the FDI-EKF algorithm according to the test measurement data, and the precise identification of the longitudinal aerodynamic parameters of the fixed-wing UAV was achieved, solving the initial value sensitivity problem in the identification process using the traditional EKF algorithm, and testing the applicability and effectiveness of the FDI-EKF algorithm and identification model for different trim flight states.

[0164] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the inventive concept of the present invention, or directly / indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A method for identifying aerodynamic parameters of an aircraft based on an improved double-innovation EKF, characterized in that: The steps include: Step 1: Perform a wind tunnel test or numerical simulation on the aircraft to obtain test estimated values ​​of all aerodynamic parameters to be identified for the aircraft, and obtain an aerodynamic model for aerodynamic parameter identification of the aircraft; Step 2, selecting state variables, input variables and observation variables in the identification process, and initializing the state variables, input variables and estimation error covariance matrix, wherein the state variables at least include all the aerodynamic parameters to be identified; Step 3, according to the aerodynamic model and the initial values ​​of the state variables, and the actual measured values ​​of the observed variables, in a trim state of the aircraft, a double-innovation extended Kalman filter method based on a forgetting factor is used to perform aerodynamic parameter identification to obtain an identification estimate of the aerodynamic parameter to be identified; Step 4: determining whether the identified estimated value is better than the tested estimated value in another trim state of the aircraft: If so, outputting the identification estimate; Otherwise, reselect the state variables, input variables and observation variables in the identification process and return to step 2.

2. The method for identifying aerodynamic parameters of an aircraft based on an improved double-innovation EKF according to claim 1, characterized in that: In step 3, when the double-innovation extended Kalman filter method based on forgetting factor is used to identify the aerodynamic parameters, the prediction process is: in, represents the prior estimate of the state variable at time k, f(·) is the state transfer function, represents the posterior estimate of the state variable at time k-1, u k-1 represents the input value of the input variable at time k-1, represents the prior estimated covariance matrix at time k, P k-1 represents the posterior estimated covariance matrix at time k-1, Φ Ak represents the state transfer matrix at time k, Q k-1 represents the process noise covariance matrix at time k-1, and T represents the transpose of the matrix.

3. The method for identifying aerodynamic parameters of an aircraft based on an improved dual-innovation EKF according to claim 2, characterized in that: In step 3, when the double-innovation extended Kalman filter method based on forgetting factor is used to identify the aerodynamic parameters, the updating process is: Among them, K k and H k They represent the Kalman gain matrix and observation matrix at time k, R k represents the observation noise covariance matrix at time k, x k represents the state vector at time k, λ1 and λ2 represent the element values ​​in the forgetting factor matrix, z k and z k-1 denote the observation vectors at time k and time k-1 respectively, h(·) denotes the observation function, K k-1 represents the Kalman gain matrix at time k-1, and I represents the identity matrix.

4. The method for identifying aerodynamic parameters of an aircraft based on an improved double-innovation EKF according to claim 1, 2 or 3, characterized in that: In step 2, in the process of initializing the state variables, the test estimated value is used as the initial value of the aerodynamic parameter to be identified in the state vector.

5. The method for identifying aerodynamic parameters of an aircraft based on an improved dual-innovation EKF according to claim 1, 2 or 3, characterized in that: In step 2, in the process of initializing the state variables, random numbers are collected as initial values ​​of the aerodynamic parameters to be identified in the state vector; or the initial values ​​of the aerodynamic parameters to be identified in the state vector are calculated based on the test estimated values.

6. The method for identifying aerodynamic parameters of an aircraft based on an improved dual-innovation EKF according to claim 1, 2 or 3, characterized in that: In step 3, the process of obtaining the actual measured value of the observed variable is: A flight test is performed on the aircraft, and test data corresponding to the observed variables are recorded. The actual measured values ​​are obtained after preprocessing the test data.

7. The method for identifying aerodynamic parameters of an aircraft based on improved dual-innovation EKF according to claim 6, characterized in that: The preprocessing includes but is not limited to outlier elimination and compatibility testing.

Images