Method for identifying structure of static pressure mathematical model of gate finger of temporary impact type wind tunnel
Through the identification of linear and nonlinear partial structures of the Wiener model, the problem of determining the structure of the middle gate finger-static pressure model of the 2.4-meter transsonic wind tunnel is solved, and high-precision control of the Mach number of the wind tunnel is achieved, and the accuracy of the static pressure control is improved. .
Patent Information
- Application Number
- CN202411215695.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-02
- Publication Date
- 2024-12-03
AI Technical Summary
The prior art is difficult to accurately determine the structure of the Wiener model, which leads to difficulty in precise control of the gate finger-static pressure model of the 2.4-meter transonic wind tunnel, affecting the high-precision control of the Mach number.
By establishing a Wiener model, using the information matrix to construct and estimate the linear partial order and the number of nonlinear partial basis functions of the model, the structure of the gate finger-static pressure model is determined, and high-precision control of the wind tunnel system is achieved.
The accuracy of Mach number control in the wind tunnel is improved, and reasonable estimation of the gate finger-static pressure model is provided, providing a foundation for high-precision control of static pressure and high-precision control of Mach number.
Smart Images

Figure CN119063960A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a structure identification method of a static pressure mathematical model of a temporary wind tunnel grid finger. Background Art
[0002] Wind tunnel is an important equipment for aerodynamic experiments on aircraft shape design. It is one of the most commonly used and effective tools for aerodynamic experiments, and it plays a decisive role in aircraft development. For high-speed transient wind tunnels, the dynamic performance of the finger mechanism directly affects the static pressure of the wind tunnel, and ultimately affects the Mach number control accuracy of the wind tunnel. In order to provide a high-quality wind tunnel flow field test environment for advanced aircraft, the Mach number of the wind tunnel system must be accurately controlled. However, the flow field control of the 2.4-meter transient transonic wind tunnel has the characteristics of nonlinearity, strong coupling, and large hysteresis, and it is very difficult to achieve high-precision control of the Mach number.
[0003]
[0004] Where Ma is the Mach number; P0 is the total pressure in the stable phase; P ct The static pressure in the room.
[0005] From the above Mach number calculation formula, it can be seen that the Mach number is mainly calculated by total pressure and static pressure. The 2.4-meter transonic wind tunnel mainly operates by stabilizing the total pressure and adjusting the static pressure. The control accuracy of the Mach number is directly reflected in the precise control of the static pressure. The 2.4-meter transonic wind tunnel's finger mechanism is mainly used for dynamic adjustment of static pressure. Therefore, the dynamic control of the finger mechanism is a key factor affecting the precise control of the wind tunnel's Mach number.
[0006] Therefore, establishing a dynamic mathematical model of the finger-static pressure in the wind tunnel system is the basis for designing advanced controllers and improving the accuracy of Mach number control, and determining the model structure is a prerequisite for establishing the finger-static pressure model. From the analysis of the influence mechanism of the finger-static pressure relationship, it can be seen that the finger-static pressure model is suitable for being described by the Wiener model composed of linear dynamic characteristics and nonlinear static characteristics in series. However, how to determine the specific structure of the Wiener model used to describe the finger-static pressure characteristics, that is, the order of the linear part of the model and the number of basis functions of the nonlinear part, is a problem that needs to be studied. At present, there is no public algorithm for the structure identification of the Wiener model. The trial and error method commonly used lacks theoretical basis and has great uncertainty in the application process, and cannot accurately give a model structure suitable for the modeling object. Summary of the invention
[0007] In order to overcome the above-mentioned shortcomings of the prior art, the present invention proposes a structure identification method for a static pressure mathematical model of a temporary wind tunnel grid finger.
[0008] The technical solution adopted by the present invention to solve the technical problem is: a method for identifying the structure of a static pressure mathematical model of a temporary wind tunnel grid finger, comprising the following steps:
[0009] Step 1, establishing a gate finger static pressure model for representing the input-output relationship between the gate finger control electrical signal and the static pressure;
[0010] Step 2: construct an information matrix based on the parameterized equation of the gate finger static pressure model and the measured input and output data;
[0011] Step 3, determining an estimated value of the order of the linear part of the gate finger static pressure model;
[0012] Step 4, selecting candidate values for the number of basis functions of the nonlinear part of the gate finger static pressure model;
[0013] Step 5: Determine the number of basis functions of the nonlinear part of the gate finger static pressure model.
[0014] Compared with the prior art, the present invention has the following positive effects:
[0015] The method of the present invention represents the input-output relationship between the gate finger and the static pressure through a Wiener model composed of linear dynamic characteristics connected in series with nonlinear static characteristics, thereby estimating the influence of the gate finger on the static pressure, and achieving the purpose of improving the control accuracy of the wind tunnel Mach number. Structural identification is to estimate the order of the linear dynamic part represented by the differential equation and the number of basis functions of the nonlinear static part represented by the basis function through a special information matrix composed of measured input and output data. In the process of identifying the model structure, the model order of the linear part of the Wiener model is first estimated; on this basis, the number of basis functions of the nonlinear part model is estimated. The method of the present invention can realize a reasonable estimation of the structure of the transonic wind tunnel gate finger-static pressure model, provide a basis for the analysis of the characteristic relationship between the gate finger and the static pressure and the high-precision control of the static pressure, and finally realize the high-precision control of the Mach number. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] The present invention will be described by way of example with reference to the accompanying drawings, in which:
[0017] Figure 1 It is a structural diagram of the Wiener model;
[0018] Figure 2 is the estimation step of the order of the linear part of the Wiener model;
[0019] Figure 3 It is the identification result of the first test data of the linear part order;
[0020] Figure 4 It is the estimation step of the number of basis functions of the nonlinear part of the Wiener model;
[0021] Figure 5 The identification result of the experimental data of the number of basis functions of the nonlinear part is shown in the figure;
[0022] Figure 6 is the estimated RMSE of different order pairs;
[0023] Figure 7 These are some simulation results of the step response simulation of the linear part of the 9 models obtained. DETAILED DESCRIPTION
[0024] A structural identification method for a dynamic mathematical model of a finger-static pressure of a transient wind tunnel system, wherein the model is used to represent the input-output relationship between finger control electrical signals and static pressure. The identification process of the model structure includes the following steps:
[0025] Step 1: Use the Wiener model consisting of a linear dynamic part and a nonlinear static part in series to represent the finger-static pressure model: the linear dynamic characteristics of the model are used to represent the characteristics of the finger mechanism in the wind tunnel system and the dynamic characteristics in the air flow field, and are represented by differential equations; the nonlinear static characteristics of the model are used to represent the nonlinear characteristics of the air flow field in the wind tunnel system, and basis functions are used to represent its inverse characteristics, thereby obtaining a parameterized representation of the Wiener model.
[0026] Step 2: Construct an information matrix based on the model parameterized equations obtained from the linear dynamic characteristics and nonlinear static characteristics and the input and output data obtained by measurement: the matrix is composed of the input and output data obtained by measurement, where the input data is the control electrical signal of the gate finger and the output data is the static pressure measurement value. The structure of the matrix is determined by the parameterized expression of the Wiener model, and the number of columns of the matrix is related to the model structure parameters to be judged.
[0027] Step 3: Determine the order of the linear part of the model through the properties of the information matrix rank: After determining the search range of the linear part order and the number of basis functions of the nonlinear part, place the number of basis functions of the nonlinear part in the information matrix at the upper limit of the search range, and obtain different information matrices by successively increasing the linear part order values. Estimate the model order of the linear part difference equation by judging the change in the information matrix rank.
[0028] Step 4: On this basis, select alternative values for the number of basis functions of the nonlinear part through the information provided by the information matrix: construct an information matrix based on the estimated value of the order of the linear part, obtain different information matrices by successively increasing the number of nonlinear basis functions, and obtain alternative values for the number of basis functions of the nonlinear part by judging the change in the rank of the information matrix.
[0029] Step 5: Select the number of basis functions of the nonlinear part based on the modeling error: Use the estimated value of the linear part order and the candidate value of the number of basis functions of the nonlinear part to establish the gate finger-static pressure model, and select the number of basis functions of the nonlinear part comprehensively based on the modeling error and model complexity.
[0030] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments:
[0031] For a transient transonic wind tunnel with a test section cross-sectional area of 2.4m×2.4m, the air flow field of the wind tunnel is adjusted by the gate finger and the main exhaust valve, where the gate finger mainly adjusts the static pressure of the flow field, and the main exhaust valve mainly adjusts the total pressure of the flow field. The control of the air flow field is mainly achieved by stabilizing the total pressure and adjusting the static pressure. In order to achieve high-precision static pressure control, it is necessary to establish a gate finger-static pressure model, that is, the input-output relationship between the gate finger control electrical signal and the static pressure. To this end, the structure of the model must be determined first. The steps for estimating the gate finger-static pressure model structure are as follows:
[0032] Step 1: Use the Wiener model to represent the gate finger-static pressure model. The Wiener model consists of a linear dynamic part connected in series with a nonlinear static part. The input of the model is the gate finger position measurement value, and the output of the model is the static pressure measurement value, such as Figure 1 The linear dynamic characteristics in the model are used to represent the characteristics of the grid finger mechanism in the wind tunnel system and the dynamic characteristics of the air flow field, and are expressed by the following differential equations:
[0033]
[0034] Where n is the order of the linear part of the model; a i and b i is the parameter of the linear part of the model; u is the model input, that is, the gate finger control electrical signal; v is the output of the linear part model, that is, the input of the nonlinear part. v cannot be measured and has no practical meaning. It is a virtual variable in the model; t is the sampling time.
[0035] The nonlinear static characteristics are used to represent the nonlinear characteristics of the air flow field in the wind tunnel system. The following polynomial basis functions are used to represent its inverse characteristics:
[0036]
[0037] Where r is the order of the polynomial, that is, the number of basis functions; c i are the parameters of the polynomial; y is the model output, i.e., the static pressure measurement value.
[0038] Step 2: The following Wiener model parameterization equation is derived from equations (1) and (2):
[0039]
[0040] The following information matrix is obtained from formula (3):
[0041]
[0042] in and is the estimated value of the linear and nonlinear order of the model;
[0043]
[0044] y t =[y t ,y t-1 ,…,y t-L+1 ] T ;
[0045] u t =[u t ,u t-1 ,…,u t-L+1 ] T
[0046] Where L is the number of data samples.
[0047] Step 3: Determine the order of the linear part of the model by the properties of the information matrix rank, which is divided into the following sub-steps:
[0048] Substep 1: Determine an upper limit n on the estimate of the model order max and r max ;
[0049] Sub-step 2: Let middle
[0050] Sub-step 3: Analyze the information matrix the nature of the rank;
[0051] Sub-step 4: Let if Then jump to sub-step 3, otherwise jump to sub-step 5;
[0052] Sub-step 5: Observe the differences Information matrix generated by value The change of rank is as The increase, when When the under-rank state changes to the full-rank state, is the estimated value of the order of the linear part n e .
[0053] In this embodiment, the condition number is used To analyze the information matrix The rank property of . When the matrix When the condition number of is a large value, it is considered is out of order; when When the condition number is a relatively small value, it is considered is of full rank. The increase, when When the condition number of The nature of the rank has changed, so it is believed that The condition number of The value is an estimate of the order of the linear part of the model n e .
[0054] In this embodiment, the following ratio of condition numbers is used to determine Whether the condition number of jumps:
[0055]
[0056] If hour, We get a maximum value, and then we set n e as an estimate of the order of the linear part of the model.
[0057] The estimation steps of the linear part order of the Wiener model are as follows: Figure 2 As shown in the figure, the identification result of a test data is as follows Figure 3 shown.
[0058] Step 4: Obtain the linear part model order n e On the basis of , the candidate values of the number of basis functions of the nonlinear part are selected again by analyzing the information obtained from the information matrix. The steps are as follows:
[0059] Sub-step 1: Let middle
[0060] Sub-step 2: Analyze the information matrix the nature of the rank;
[0061] Sub-step 3: Let if Then jump to sub-step 2, otherwise jump to sub-step 4;
[0062] Sub-step 4: Observe the differences Information matrix generated by value The change in rank is The value of r e Increase to r e +1, If the condition number increases rapidly, then it is considered that r eis an alternative value for the number of basis functions in the nonlinear part.
[0063] In this embodiment, the following ratio of condition numbers is used to determine The increase in the condition number of :
[0064]
[0065] If hour, A maximum value is obtained. At this time, r e As an alternative to the number of basis functions in the nonlinear part, the estimation steps of the number of basis functions in the nonlinear part of the Wiener model are as follows: Figure 4 As shown in Figure 2, the identification results of an experimental data are as follows: Figure 5 shown.
[0066] Step 5: Select the order of the nonlinear part based on the modeling error. Combining formula (3), using each alternative The model is identified using the following least squares method:
[0067]
[0068] Where θ is the model parameter vector. The model output estimate is calculated using the parameter vector obtained through identification and the following formula:
[0069]
[0070] And the root mean square error of the modeling output estimates:
[0071]
[0072] in
[0073]
[0074] The number of basis functions is determined by combining the modeling errors corresponding to the alternative values of the number of basis functions of the nonlinear part and comprehensively considering the complexity of the model. Taking the model error and the complexity of the model into consideration, the gate finger-static pressure model structure is selected with a linear part order of 3 and a nonlinear part basis function number of 4.
[0075] Verification results:
[0076] (1) Verification based on model complexity and modeling error
[0077] In order to verify the effectiveness of the method of the present invention, the linear part order is 2, 3, 4, and the nonlinear part order is 2, 3, 4. A total of 9 order combinations are used to build the model and calculate the model output estimation RMSE. The model parameters are still estimated using formula (8), and the model estimation RMSE is still calculated using formula (9).
[0078] Table 1 Identification results of linear denominator parameters for Ma condition
[0079]
[0080] Table 2 Identification results of linear molecular parameters of Ma condition
[0081]
[0082] Table 3 Parameter identification results of nonlinear part of Ma condition
[0083]
[0084]
[0085] The linear part parameter a obtained by modeling is shown in Table 1, the linear part parameter b is shown in Table 2, the nonlinear part parameter c is shown in Table 3, and the estimated RMSE of different order pairs is shown in Figure 6 As shown in the figure, the RMSE of the model estimated by each modeling pair with different orders shows that the most appropriate selection is 3 for the linear part order and 4 for the number of basis functions for the nonlinear part. The complexity of the model is simplified on the basis of ensuring that the RMSE of the model estimation is as small as possible, which verifies the effectiveness of the method of the present invention.
[0086] (2) Model-based and rational verification
[0087] In order to further verify the effectiveness of model order identification, the linear part of the nine models is simulated for step response. Some simulation results are shown in the figure. Figure 7 For an actual wind tunnel system, when a unit step control signal is input to the gate finger, the linear part output should present a monotonically increasing trend that is close to a straight line. Figure 7 Only the case where the step response curve is monotonically increasing is given. For the step responses of the model orders (n=2, r=2), (n=2, r=3), (n=2, r=4), (n=3, r=2), and (n=3, r=3), the obtained model unit step response shows divergent oscillation, which does not conform to the input-output characteristics of the actual gate finger-static pressure. Therefore, it is not included in the paper. Figure 7 Draw its step response curve. Figure 7It can be seen from the step response curves of (n=3, r=3), (n=3, r=4), (n=4, r=3), and (n=4, r=4) that only the curve with a model order of (n=3, r=4) presents a monotonically increasing form that is close to a straight line, further demonstrating the effectiveness of the method of the present invention.
Claims
1. A method for identifying the structure of a static pressure mathematical model of a temporary wind tunnel grid finger, characterized by: The steps include: Step 1, establishing a gate finger static pressure model for representing the input-output relationship between the gate finger control electrical signal and the static pressure; Step 2: construct an information matrix based on the parameterized equation of the gate finger static pressure model and the measured input and output data; Step 3, determining an estimated value of the order of the linear part of the gate finger static pressure model; Step 4, selecting candidate values for the number of basis functions of the nonlinear part of the gate finger static pressure model; Step 5: Determine the number of basis functions of the nonlinear part of the gate finger static pressure model.
2. The method for identifying the static pressure mathematical model structure of a temporary wind tunnel grid finger according to claim 1, characterized in that: The gate finger static pressure model in step 1 is a Wiener model consisting of a linear dynamic part and a nonlinear static part connected in series, wherein: (1) The linear dynamic part is expressed by the following difference equation: Where n is the order of the linear part of the model; a i and b i is the linear part parameter of the model; u is the model input; v is the linear part model output; t is the sampling time; (2) The nonlinear static part uses the following basis functions to represent its inverse characteristics: Where r is the number of basis functions; c i are the parameters of the basis function; y is the model output.
3. The method for identifying the static pressure mathematical model structure of a temporary wind tunnel grid finger according to claim 2, characterized in that: The parameterized equation of the Wiener model is:
4. The method for identifying the static pressure mathematical model structure of a temporary wind tunnel grid finger according to claim 3, characterized in that: The information matrix is: and t =[and t ,and t-1 ,…,and t-L+1 ] T ; in t =[in t ,in t-1 ,…,in t-L+1 ] T ; in, is the information matrix, and is the estimated value of the linear and nonlinear order of the model, and L is the number of data samples.
5. The method for identifying the static pressure mathematical model structure of a temporary wind tunnel grid finger according to claim 4, characterized in that: The method for determining the estimated value of the order of the linear part of the gate finger static pressure model in step 3 is: The first step is to determine the upper limit n of the model order estimate max and r max ; Step 2: Let the information matrix In Step 3: Judge the information matrix Does the condition number of mutate? If so, let the estimated value of the linear part order be If not, proceed to step 4; Step 4: Order Then determine whether it satisfies If yes, go to step 3; if no, end.
6. A method for identifying the structure of a static pressure mathematical model of a temporary wind tunnel grid finger according to claim 5, characterized in that: Judgment Information Matrix The method to determine whether the condition number of mutates is as follows: when When is a maximum value, the information matrix is determined The condition number has changed suddenly.
7. The method for identifying the static pressure mathematical model structure of a temporary wind tunnel grid finger according to claim 5, characterized in that: In step 4, the method for selecting the candidate value of the number of basis functions of the nonlinear part of the gate finger static pressure model is: The first step is to make the information matrix In Step 2: Judge the information matrix Whether the condition number of has increased rapidly: If so, an alternative value r of the number of basis functions in the nonlinear part is obtained e ; If not, go to step 3; Step 3: Order Then determine whether it satisfies If yes, go to step 2; if no, end.
8. The method for identifying the static pressure mathematical model structure of a temporary wind tunnel grid finger according to claim 7, characterized in that: Judgment Information Matrix The method to determine whether the condition number of has increased rapidly is: when When is a maximum value, the information matrix is determined The condition number increased rapidly.
9. The method for identifying the static pressure mathematical model structure of a temporary wind tunnel grid finger according to claim 7, characterized in that: In step five, the method for determining the number of basis functions of the nonlinear part of the gate finger static pressure model is: using the estimated value of the linear part order and the alternative value of the number of basis functions of the nonlinear part to establish the gate finger static pressure model, and comprehensively determining the number of basis functions of the nonlinear part through modeling error and model complexity.
10. The method for identifying the static pressure mathematical model structure of a temporary wind tunnel grid finger according to claim 9, characterized in that: The method for comprehensively determining the number of basis functions of the nonlinear part by modeling error and model complexity is: Combined with the parameterized equations of the Wiener model, various alternative Values for model identification: Where θ is the model parameter vector, and the model output estimate is calculated by the parameter vector obtained through identification and the following formula And the root mean square error of the modeling output estimates: in, Finally, the number of basis functions is determined by combining the modeling errors corresponding to the alternative values of the number of basis functions in different nonlinear parts and comprehensively considering the complexity of the model.
Citation Information
Cited By
Time-lag identification method for large-scale transient high-speed wind tunnel
CN122505524A
A time delay identification method for a large-scale transient high-speed wind tunnel
CN122505524B
