A method for evaluating online tuning of control laws based on impulse response identification

Abstract

The invention relates to an online control law adjustment evaluation method based on pulse response identification. The method comprises the following steps: selecting a state point to be identified of an aircraft and pulse excitation and response data acquired in real time, fitting a fitted pulse time domain response and a fitted pulse frequency domain response, and forming an identification model with fitting parameters to perform online fitting identification. Then, a closed-loop transfer function with a controller is derived according to the automatic control principle, and expected closed-loop poles are selected according to flight quality requirements of similar aircraft models. The expected poles are built into an onboard online design program. The controller's undetermined gain is calculated according to an undetermined coefficient method and applied to the controller to obtain a closed-loop control system based on an open-loop identification model and a controller. Finally, a step excitation is added to the closed-loop control system of the aircraft based on the control law model to obtain a step response of the closed-loop system, and various indicators are calculated and compared with expected indicators to evaluate whether the controller meets the design requirements, thereby ensuring that the aircraft still has good stability, safety and high efficiency in the identification test.

Description

Technical Field

[0001] The invention belongs to the technical field of unmanned aerial vehicle flight control, and in particular relates to an online adjustment participation evaluation method of a control law based on impulse response identification. Background Art

[0002] Identification technology has the advantages of short cycle and low cost in obtaining the dynamic model and parameters of the controlled object, and has been widely used. The identification technology is summarized as follows: (1) According to the application scenario, it can be divided into system identification and parameter identification. The main difference lies in whether the model structure needs to be identified. For helicopters and fixed-wing aircraft, their dynamic model structure is relatively fixed, and the identification problem degenerates from system identification to parameter identification; (2) According to the data type classification of identification matching, identification technology can be divided into frequency domain identification and time domain identification. Among them, frequency domain identification is completed by matching the predicted frequency response and the measured frequency response. In the field of aircraft identification, Mark B. Tischler and his CIFER have very good practicality and are widely used by rotorcraft researchers; time domain identification is completed by matching the predicted time domain response and the measured time domain response. The main method used is the maximum likelihood method, and the Newton-Simpson method is often used for solution. It is widely used in the fields of fixed-wing aircraft flight simulation data packet acquisition, test flight data verification, etc. (3) According to whether it is carried out in real time, it can be divided into online identification and offline identification. Among them, online identification performs real-time data collection and online identification solutions, which can be combined with adaptive control technology and has great application prospects in scenarios such as fault-tolerant control, but the stability theory still needs to be improved. Offline identification refers to obtaining flight test data through flight tests, and then performing model frequency domain / time domain model identification after the test. Its advantage is that it can collect sufficient data through multiple tests, and there is ample time to adjust the model parameters to approximate the collected data and identify a model with higher accuracy. At present, most identification practices belong to offline identification.

[0003] The current identification has the following problems:

[0004] (1) The identification excitation needs to be determined in combination with the aircraft modal characteristics. However, for some verification aircraft, their characteristics themselves hope to be determined through identification. This requires a large number of preliminary tests and identifications to determine the excitation type and excitation parameters, which is inefficient.

[0005] (2) Commonly used identification excitation signals, such as multi-pole square waves and swept frequency signals, take a long time to collect data. During the collection process, the aircraft is prone to deviate from the predetermined identification point, resulting in a large identification error.

[0006] (3) Identification technology is often used in the research and development of wing-rotor composite configurations and multi-mode aircraft such as tilt-rotor, tilt-propeller, and tilt-wing to reduce costs and speed up progress. However, these new configuration verification aircraft have poor stability due to factors such as aerodynamic coupling and modal conversion. Conventional offline identification and offline design may face the risk of poor stability or even crash during aircraft identification tests. Online identification and parameter adjustment are required to improve the safety and efficiency of the test. Summary of the invention

[0007] In order to overcome the deficiencies of the prior art, the object of the present invention is to provide a method for realizing online adjustment and evaluation of control laws based on impulse response identification, so as to improve test safety and test efficiency.

[0008] To solve the above problems, the technical solution adopted by the present invention is as follows:

[0009] The present invention provides a control law online adjustment participation evaluation method based on impulse response identification, comprising:

[0010] S1. Aircraft pulse excitation: select the state point of the aircraft to be identified, and obtain the pulse excitation and response data in real time through the pulse excitation test;

[0011] S2. Online pulse response identification: Based on the pulse excitation and response data obtained in step S1, a fitted pulse time domain response and a fitted pulse frequency domain response are obtained, which include basis functions and parameters to be identified; then, according to the identification model and the real characteristics of the aircraft, the pulse response fitting time length, the number of fitting points, the fitting basis function, and the fitting system state are selected for online fitting identification, and the key fitting parameters are solved to obtain the fitted pulse frequency domain response. According to the principle of automatic control, the fitted pulse frequency domain response is the transfer function of the controlled object;

[0012] S3. Online control gain design: Design the controller using the undetermined coefficient method according to the expected dynamic performance; Derived the closed-loop transfer function with the controller according to the identification model of step S2, the control law model of the undetermined control gain and the aircraft automatic control principle. According to the undetermined coefficient method, the denominator of the closed-loop transfer function can be made equal to the denominator of the expected closed-loop transfer function obtained according to the flight quality requirements by reasonably selecting the control gain; Thus, the controller gain is calculated and applied to the controlled object to obtain a closed-loop control system based on the open-loop identification model and the controller;

[0013] S4. Closed-loop step response and evaluation: Based on the control law model in step S3, a step excitation is applied to the closed-loop control system of the aircraft to obtain the step response of the closed-loop system. Then, based on the step response of the closed-loop system, various indicators are calculated and compared with the expected indicators to evaluate whether the controller meets the design requirements.

[0014] Furthermore, the control law online adjustment participation evaluation method based on impulse response identification also includes: repeatedly executing steps S1-S4 multiple times, carrying out pulse identification and control parameter solution (ie S3) multiple times, and making a preferential selection or averaging according to the evaluation results, that is: averaging the gain of the control law, or the parameters.

[0015] Furthermore, during the execution of steps S1-S4, the pulse excitation and response data, the step response and identification model of the closed-loop system, the control parameters, and the evaluation results are transmitted back to the ground station for evaluation and decision-making by ground personnel.

[0016] Furthermore, in step S2, "on the basis of the pulse excitation and response data obtained in step S1, a fitted pulse time domain response and a fitted pulse frequency domain response are fitted, and an identification model is formed with the fitting parameters", specifically: first, the fitted pulse time domain response is calculated by the basis function fitting method according to the acquired pulse excitation and response data and the Euler formula, and then the fitted pulse time domain response is subjected to a pulse excitation time-frequency transformation to obtain a fitted pulse frequency domain response, and finally, the fitting parameters are obtained by minimizing the optimization function, and an identification model is formed with the fitted pulse time domain response and the fitted pulse frequency domain response.

[0017] Furthermore, the “calculating the fitted pulse time domain response by the basis function fitting method and according to the acquired pulse excitation and response data and the Euler formula, and then performing pulse excitation time-frequency transformation on the fitted pulse time domain response to obtain the fitted pulse frequency domain response” specifically comprises the following steps:

[0018] S2-1. Definition of pulse excitation time domain basis functions

[0019] The basis function fitting method is used to identify the impulse response of the model and approximate the real impulse response y in the acquired impulse excitation and response data. impl (t), and the fitted pulse time domain response is obtained

[0020]

[0021] Among them, the form of the basis function φ is θ=[θ1,θ2,θ3,θ4] is the key parameter of the basis function, θ1>0,θ2>0,θ3>0, and 0≤θ4≤2π;

[0022] According to Euler's formula, we can get

[0023]

[0024] S2-2. Pulse excitation time-frequency transformation

[0025] Performing Laplace transform on the above equation, we get the complex frequency domain basis function φ(s,θ):

[0026]

[0027] Get the fitted pulse frequency domain response

[0028]

[0029] Furthermore, the gradient descent method is used to solve θ i =[θ i1 ,θ i2 ,θ i3 ,θ i4 ], i = 1, 2, 3, ..., N, and is solved by the following formula:

[0030]

[0031] Determine θ i =[θ i1 ,θ i2 ,θ i3 ,θ i4 ], i = 1, 2, 3, ..., N, and get the fitted pulse frequency domain response

[0032] Furthermore, the “obtaining the fitting parameters by minimizing the optimization function” is specifically: obtaining the fitting parameters θ according to the following minimizing optimization function J i =[θ i1 ,θ i2 ,θ i3 ,θ i4 ], i = 1, 2, 3, ..., N;

[0033]

[0034] where w j is the weight coefficient, j=1, 2, 3, ..., l is the impulse response sampling point, and sampling is performed within the impulse response approximation time period.

[0035] Furthermore, the closed-loop transfer function of the controller in step S3 is Where K(s) is the control law model to be determined, To identify the model, Y c (s) is the controlled quantity instruction y c(t), Y(s) is the Laplace transform of the controlled output y(t);

[0036] According to the flight quality requirements, the poles of the expected closed-loop transfer function are obtained and built into the on-board online design program. K(s) is solved by the undetermined coefficient method. Specifically, the expected closed-loop poles p1, p2, ..., p are selected according to the flight quality requirements of similar aircraft models. n , and the expected poles p1,p2,...,p n Built into the on-board online design program, the undetermined coefficient method is used to obtain

[0037]

[0038] Wherein, den[] is used to obtain the denominator of the system transfer function, K(s) is the undetermined control law model, and K(s) is derived using the undetermined coefficient method.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] The identification method based on impulse response proposed in the present invention uses impulse input as excitation. This type of excitation has clear physical meaning and is easy to implement. The pilot or operator can adjust it at any time according to the time domain response. The time required for impulse response identification is short, which can effectively avoid the widely existing identification state deviation problems such as frequency sweep or multi-level square waves. The control law online adjustment and evaluation method based on impulse response proposed in the present invention can complete the impulse response identification and control gain design online in real time, ensuring that the aircraft still has good stability in the identification test, avoiding the risk of crash, and greatly improving the safety and efficiency of the test. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flow chart of the method for online adjustment and evaluation of control law based on impulse response identification according to the present invention;

[0042] Figure 2 It is a principle diagram of impulse response identification in the online adjustment participation evaluation method of the control law based on impulse response identification of the present invention;

[0043] Figure 3 It is a schematic diagram of the automatic control principle of an aircraft in the online adjustment participation evaluation method of the control law based on impulse response identification of the present invention;

[0044] Figure 4 It is a diagram showing the principle of controller evaluation based on closed-loop step response in the online adjustment participation evaluation method of the control law based on impulse response identification of the present invention;

[0045] Figure 5This is a schematic diagram of the short-period motion pulse excitation principle in an embodiment of the present invention;

[0046] Figure 6 is a diagram showing the acquisition points of the short-period impulse response of the aircraft body, the fitted impulse response and the basis function in an embodiment of the present invention;

[0047] Figure 7 is a comparison diagram of the short-period characteristics of the aircraft and the Bode diagram of the identification model in the embodiment of the present invention;

[0048] Figure 8 is a structural diagram of a controller for short-cycle control of an aircraft in an embodiment of the present invention;

[0049] Fig. 9 It is a comparison of the step tracking time domain curves of the closed-loop system in the embodiment of the present invention and the actual system. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0051] like Figure 1 As shown, the present invention provides a control law online adjustment participation evaluation method based on impulse response identification, comprising the following steps:

[0052] Step S1. Aircraft pulse excitation (aircraft body characteristic pulse excitation): select the state point of the aircraft to be identified, and obtain the pulse excitation and response data in real time through the pulse excitation test;

[0053] Step S2. Online pulse response identification: Based on the pulse excitation and response data obtained in step S1, the fitted pulse time domain response and the fitted pulse frequency domain response are fitted, and an identification model is formed with the fitting parameters; then, according to the identification model and the real characteristics of the aircraft, the pulse response fitting time length, the number of fitting points, the fitting basis function, and the fitting system state are selected for online fitting identification. The principle of pulse response identification is as follows: Figure 2 , G(s) in the figure is the object to be identified, To identify the model, is the minimization optimization function J, δ(t) is the pulse excitation, y impl (t) is the true impulse response, To fit the impulse response;

[0054] Step S3. Online control gain design: According to the expected dynamic performance, the controller is designed using the undetermined coefficients, and the identification model of step S2, the undetermined control law model and the aircraft automatic control principle (such as Figure 3As shown in the figure, K(s) is the control law model to be determined. To identify the model, y c (t) is the controlled quantity instruction, y(t) is the controlled quantity output), the closed-loop transfer function with the controller is derived, and then the poles of the desired closed-loop transfer function are obtained according to the flight quality requirements, and built into the on-board online design program, and K(s) is calculated using the undetermined coefficient method;

[0055] Step S4. Closed-loop step response and evaluation: Based on the control law model in step S3, a step excitation is applied to the closed-loop control system of the aircraft to obtain the step response of the closed-loop system; then, based on the step response of the closed-loop system, various indicators are calculated and compared with the expected indicators to evaluate whether the controller meets the design requirements, such as Figure 4 shown.

[0056] The method of the present invention uses pulse input as excitation. This type of excitation has clear physical meaning and is easy to implement. The pilot or operator can adjust it at any time according to the time domain response. The pulse response identification takes a short time, which can effectively avoid the widely existing identification state deviation problems such as frequency sweep or multi-level square waves. The control law online adjustment and evaluation method based on pulse response proposed by the present invention can complete the pulse response identification and control gain design online in real time, ensuring that the aircraft still has good stability in the identification test, avoiding the risk of crash, and greatly improving the safety and efficiency of the test.

[0057] The control law online adjustment participation evaluation method based on impulse response identification of the present invention is further described below through embodiments.

[0058] Example:

[0059] This embodiment selects a certain type of aircraft at an altitude of 4000m and a speed of 130m / s as an embodiment, and provides a specific embodiment as follows based on longitudinal short-cycle identification, control law design and evaluation:

[0060] Step 1: Pulse excitation test and data acquisition (aircraft controlled object identification)

[0061] Select the state point of the aircraft to be identified and carry out pulse excitation test (such as Figure 5 The airborne system acquires pulse excitation and response data in real time.

[0062] Step 2: Online impulse response identification

[0063] The impulse response identifies the short-period characteristics of the aircraft. Based on the aircraft dynamics model established based on the known wind tunnel test data, the actual short-period characteristics of the aircraft are:

[0064]

[0065] After applying unit pulse elevator excitation to the aircraft in a short period, the impulse response data is collected. According to the method in step 1, the impulse response fitting time length is selected as 20s, the number of fitting points is 2000 points, the fitting basis function is 1, and the fitting system state is 2. The fitting results are as follows: Figure 6 and Figure 7 As shown, the result of fitting identification is:

[0066]

[0067] Compared with the real model, the accuracy of the identification results is shown in Table 1:

[0068] Table 1 Example identification model accuracy

[0069]

[0070] Step 3: Online Control Gain Design

[0071] Using conventional control structures such as Figure 8 As shown, in order to obtain the damping feedback coefficient kd and the proportional coefficient kp, the following design is performed:

[0072] In order to maintain generality, the identification results are abstracted as:

[0073]

[0074] Among them, b1 = 189.7776, b2 = 75.4068, a1 = 1, a2 = 1.676, a3 = 12.59; after closing the loop with damping feedback coefficient kd and proportional coefficient kp, according to the principle of automatic control, the closed-loop transfer function can be obtained as follows:

[0075]

[0076] The desired closed-loop transfer function poles are selected as p1 = -113, p2 = -1.8, and p3 = -0.36.

[0077] From this we can get:

[0078] s 3 +(a1+b1k d )s 2 +(a2+k d b2+b1k p )s+b2k p =(s-p1)(s-p2)(s-p3),

[0079] Solved

[0080] kp =-p1*p2*p3 / b2=0.97,

[0081] k p =-(p1+p2+p3+a1) / b1=0.6015,

[0082] The design of control parameters is now completed. After applying the controller, the closed-loop control system based on the open-loop identification model and the controller is:

[0083]

[0084] Step 4: Closed-loop step stimulus and response acquisition and evaluation

[0085] For the actual controlled object, use the controller in step 3 to form a closed-loop system, give a step command to obtain the closed-loop step response of the actual system, and the step curve is as follows: Fig. 9 For the convenience of comparison, Fig. 9 The ideal step curve of the closed-loop system with controller is also given in the embodiment. By comparison, it can be seen that in addition to the steady-state response amplitude, the rise time, settling time and overshoot can all be effectively evaluated through the step response. The difference in the steady-state response amplitude between the two is mainly due to the error between the identification model and the actual object.

[0086] The above-mentioned embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and substitutions made by technicians in this field on the basis of the present invention shall fall within the scope of protection required by the present invention.

Claims

1. A control law online tuning evaluation method based on impulse response identification, characterized in that: include: S1. Aircraft pulse excitation: select the state point of the aircraft to be identified, and obtain the pulse excitation and response data in real time through the pulse excitation test; S2. Online pulse response identification: Based on the pulse excitation and response data obtained in step S1, a fitted pulse time domain response and a fitted pulse frequency domain response are obtained, which include basis functions and parameters to be identified; then, according to the identification model and the real characteristics of the aircraft, the pulse response fitting time length, the number of fitting points, the fitting basis function, and the fitting system state are selected for online fitting identification, and the fitting key parameters are solved to obtain the fitted pulse frequency domain response. According to the principle of automatic control, the fitted pulse frequency domain response is the transfer function of the controlled object; S3. Online control gain design: Design the controller using the undetermined coefficient method according to the expected dynamic performance; derive the closed-loop transfer function with the controller according to the identification model of step S2, the control law model of the undetermined control gain and the aircraft automatic control principle. According to the undetermined coefficient method, the denominator of the closed-loop transfer function is made equal to the denominator of the expected closed-loop transfer function obtained according to the flight quality requirements by reasonably selecting the control gain; thus, the controller gain is calculated and applied to the controlled object to obtain a closed-loop control system based on the open-loop identification model and the controller; S4. Closed-loop step response and evaluation: Based on the control law model in step S3, a step excitation is applied to the closed-loop control system of the aircraft to obtain the step response of the closed-loop system; then, based on the step response of the closed-loop system, various indicators are calculated and compared with the expected indicators to evaluate whether the controller meets the design requirements; In step S2, "fitting the pulse time domain response and the pulse frequency domain response based on the pulse excitation and response data obtained in step S1, and forming an identification model with the fitting parameters" is specifically: firstly, by the basis function fitting method, and according to the acquired pulse excitation and response data and Euler's formula, the fitting pulse time domain response is calculated, and then the pulse excitation time-frequency transformation is performed on the fitting pulse time domain response to obtain the fitting pulse frequency domain response, and finally, the fitting parameters are obtained by minimizing the optimization function, and the identification model is formed with the fitting pulse time domain response and the fitting pulse frequency domain response; The "implementing the fitting model pulse time domain response by the basis function fitting method, simplifying it according to the Euler formula, and then performing pulse excitation time-frequency transformation on the fitting pulse time domain response to obtain the fitting pulse frequency domain response" specifically includes the following steps: S2-1. Definition of pulse excitation time domain basis functions In order to approximate the real impulse response obtained by the pulse excitation , use the basis function fitting method to construct the fitted pulse time domain response ; ， in, is the basis function number, which takes values ​​of 1, 2, 3, ... N, is time, with a value of 0~+∞, basis function The shape is , is the key parameter of the basis function, ; According to Euler's formula, we can get ； S2-2. Pulse excitation time-frequency transformation Perform Laplace transform on the above formula to obtain the complex frequency domain basis function : ； Then the fitted pulse frequency domain response is obtained : In the formula is the basis function number, which takes values ​​of 1, 2, 3, … N, and s is the Laplace complex variable.

2. The control law online tuning evaluation method based on impulse response identification according to claim 1 is characterized in that: Also includes: Steps S1-S4 are repeatedly executed multiple times, pulse identification and control gain solution are performed multiple times, and the optimal selection or averaging is performed based on the evaluation results.

3. The control law online tuning evaluation method based on impulse response identification according to claim 1 is characterized in that: During the execution of steps S1-S4, the pulse excitation and response data, the step response and identification model of the closed-loop system, the control parameters, and the evaluation results are transmitted back to the ground station for evaluation and decision-making by ground personnel.

4. The control law online tuning evaluation method based on impulse response identification according to claim 1, 2 or 3, characterized in that: Gradient descent method is used to solve , which is solved by the following formula: Sure , and get the fitted pulse frequency domain response .

5. The control law online tuning evaluation method based on impulse response identification according to claim 1 is characterized in that: The "obtaining fitting parameters by minimizing the optimization function" is specifically: according to the following minimization optimization function Find the fitting parameters ; ； in is the weight coefficient, is the impulse response sampling point, sampled during the impulse response approximation period.

6. The control law online tuning evaluation method based on impulse response identification according to claim 1, 2, 3 or 5, characterized in that: The closed-loop transfer function with the controller in step S3 is ,in is the control law model to be determined, is the vehicle body characteristic identification model, The command for the controlled quantity The Laplace transform of Output of the controlled quantity The Laplace transform of According to the flight quality requirements, the poles of the desired closed-loop transfer function are obtained. , and the expected extreme Built into the on-board online design program, get ； in, To obtain the denominator of the system transfer function, The control law model is to be determined. The control law model is derived by using the undetermined coefficient method. The specific parameter in is the control law gain.

Images