A real-time hybrid test control compensation method

Abstract

The present invention discloses a real-time hybrid test control compensation method, comprising: obtaining the displacement response of the actuator through white noise excitation; fitting the white noise with the displacement response to obtain a transfer model; designing a compensation filter based on the transfer model, and optimizing the parameters of the compensation filter, and completing the hybrid test control compensation based on the optimized compensation filter model. The purpose of the present invention is to reduce the loading error in the hybrid test and increase the test accuracy. At the same time, the stability and reliability of the hybrid test are also guaranteed.

Description

Technical Field

[0001] The invention belongs to the technical field of real-time hybrid test vibration table control, and in particular relates to a real-time hybrid test control compensation method. Background Art

[0002] Real-time hybrid testing is applied in the field of civil engineering. The hybrid test consists of a physical substructure with a physical specimen and a numerical substructure. The numerical substructure is numerically simulated in a computer, and the signal from the sensor on the physical specimen is transmitted to the numerical model. Within a simulation step, the numerical model calculates its command and outputs it to the loading system for execution, acting on the physical specimen to complete a closed-loop interactive simulation. Figure 2 It is a typical uniaxial mixed test for civil structures.

[0003] For the loading system of the hybrid test, loading errors are common. In order to minimize the errors, a compensation algorithm needs to be added. A large part of the factors causing the system loading errors come from the uncertainty of the specimen reaction force. Figure 2 As shown in the figure, in the test, the earthquake loading actuator B is used, and the seismic loading is performed at the same time. Therefore, the reaction force of the physical specimen does not only come from the hybrid test actuator A. In addition, in the hybrid test, the model of the physical substructure is unknown. These two reasons cause the uncertainty of the specimen reaction force. Therefore, the control compensation of the hybrid test is very complicated. The commonly used compensation algorithm is the inverse model compensation algorithm. In the civil engineering field, there are also ATS compensation algorithm and MPC compensation algorithm. However, none of the above methods can directly eliminate the influence of the uncertainty of the specimen reaction force on the loading error. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a real-time hybrid test control compensation method, which aims to reduce the loading error in the hybrid test and increase the test accuracy. It not only increases the stability of the hybrid test, but also ensures the accuracy and reliability of the test results.

[0005] A real-time hybrid test control compensation method, characterized by comprising:

[0006] Generate multi-channel white noise;

[0007] The white noise excitation control system is used to obtain the displacement response of the actuator;

[0008] Fitting the white noise with the displacement response to obtain a transfer model;

[0009] Designing a compensation filter based on the transfer model, wherein the compensation filter comprises: parameter adjustment of a parameterized multi-order compensation filter and a numerical substructure;

[0010] Optimizing the parameters of the compensation filter based on the transfer model;

[0011] Based on the optimized compensation filter, the compensation control of the hybrid test is completed.

[0012] Optionally, the multi-channel white noise includes: displacement command white noise, feedforward white noise and force loading white noise;

[0013] The transmission model includes: displacement instruction transmission model H d , feedforward transfer model H v , force transmission model H F ;

[0014] Fitting the white noise with the displacement response to obtain the transfer model includes:

[0015] The displacement command white noise is fitted with the displacement response to obtain the displacement command transfer model H d ;

[0016] The feedforward white noise is fitted with the displacement response to obtain the feedforward transfer model H v ;

[0017] The force loading white noise is fitted with the displacement response to obtain the force transfer model H F .

[0018] Optionally, the parameterized multi-order compensation filter H p Split into feature filters H of different orders p,i The weighted sum of , that is, Formula 17;

[0019]

[0020] Among them, k i is the weighting coefficient of the filter. i is the i-th order characteristic filter H p,i The characteristic time parameter, T i Used to adjust the bandpass filter H p,i The passband range of the filter. n is the order of the multi-order compensation filter.

[0021] Optionally, the compensation filter G v The multi-order compensation filter H p and the parameter setting model G′ of the hybrid test numerical sub-model G, i.e., formula 16;

[0022] G v =H p G′ Formula 16

[0023] Among them, G′ is the result of parameter tuning of the original numerical substructure model G while maintaining the model structure, and H p It is a parameterized multi-order compensation filter.

[0024] Compared with the prior art, the present invention has the following advantages and technical effects:

[0025] (1) As a new form of time-delay compensation in the field of hybrid testing, compared with the traditional inverse model compensation method, it eliminates the influence of specimen reaction force uncertainty on the compensation effect;

[0026] (2) Compared with the adaptive algorithms such as ATS commonly used in the civil engineering field, the compensation filter designed by this method only needs offline fitting, avoiding the uncertainty of online identification and adjustment. Therefore, the stability of the hybrid test is guaranteed and it has better practicality;

[0027] (3) Small computing power requirement. The designed compensation filter has much less computing power than ATS, MPC and other algorithms. Whether it is running in the simulator or in the controller, there is no computing power burden.

[0028] (4) The model's own parameter tuning is incorporated into the optimization of control compensation, which broadens the optimization space. This is a feature that none of the existing methods have.

[0029] (5) Easy to use. The designed compensation filter has a clear structure and a bandpass characteristic. It does not require an additional low-pass filter required by ATS to filter high-frequency noise, providing a complete solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0031] Figure 1 It is the signal flow diagram of hydraulic servo control system;

[0032] Figure 2 It is a schematic diagram of the principle of a typical uniaxial mixing test;

[0033] Figure 3 This is a control flow chart of a conventional hybrid test;

[0034] Figure 4 is a control flow chart of an embodiment of the present invention;

[0035] Figure 5 Schematic diagram of implementation steps of an embodiment of the present invention. DETAILED DESCRIPTION

[0036] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0037] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0038] The embodiments of the present invention are based on Figure 1 Hydraulic servo control system. Figure 1 The signal flow chart of is given in detail the mechanism parameters, which are used to establish the corresponding mathematical model. The theoretical derivation and design implementation of the following embodiments are closely related to the model of the hydraulic servo control system. It should be noted that the application of the present invention is not limited to the hybrid test of the hydraulic servo control system, nor is it limited to the implementation of, for example Figure 2 Uniaxial mixing test.

[0039] right Figure 1 The mathematical model of the control system is established, then:

[0040] x=H d u d +H v u v -H F F Formula 1

[0041] Where x is the displacement of the actuator, u d is the target displacement command input by the controller, u v is the feedforward command of the controller, and F is the reaction force of the specimen on the actuator.

[0042] According to H d , H v , H F The transfer model characteristics are further transformed as follows:

[0043]

[0044] Then there is

[0045]

[0046] Among them, k D is the differential parameter in the PID controller, k P is the parameter of the proportional link in the PID controller, s = jω is the Laplace transform, k q is the gain of the hydraulic servo valve, k a is the gain of pressure feedback, AP is the cross-sectional area of ​​the actuator cavity, β is the effective bulk modulus of the system (including oil, trapped air and mechanical compliance), C c is the leakage parameter of the actuator piston, K c is the leakage parameter of the hydraulic servo valve, and V is the total volume of the oil in the actuator chamber.

[0047] Transfer Model and There is no specific physical meaning, but this transformation can facilitate the implementation of subsequent steps. For example, in the above derivation, the order of the transfer model can be clearly distinguished. and is of order 1, It is third order, which is convenient for the design of compensation filter.

[0048] Figure 2 This is a schematic diagram of a typical uniaxial mixing test principle in accordance with an embodiment of the present invention. The control compensation method designed in the present invention is aimed at Figure 2 Actuator A in a hybrid test. In this structure, actuator A is used to execute the displacement command calculation results of the numerical substructure simulation model running in the computer. The physical substructure is also often called the "test piece".

[0049] Figure 3 yes Figure 2 Control flow diagram of the uniaxial hybrid test, where G is the numerical substructure model, G Phys It is a physical substructure model, which satisfies the following relationship by definition:

[0050]

[0051] According to formula 1 and formula 2, Figure 3 The following mathematical model is established:

[0052]

[0053] According to Formula 4 and Formula 5, the following relationship is derived:

[0054]

[0055] exist Figure 3 In the conventional control compensation strategy shown in the figure, the compensation link H c The optimization fitting objective is

[0056]

[0057] Compensation link H c To achieve no compensation, that is, H c =1, input u dThe inverse model of the output x transfer model is therefore called the inverse model method. The ATS and MPC methods consider the online adjustment of the model, that is, real-time identification and adjustment, but the modeling ideas are the same, and the optimization target is still Formula 7.

[0058] Rearranging formula 7 gives

[0059]

[0060] In formula 7 and formula 8, "G a →G b "The mathematical meaning of the optimization fitting problem is

[0061]

[0062] G a Approximation to G through parameter fitting b , where θ is G a The adjustable parameter variable, ω is the frequency variable, which takes values ​​within the reference frequency range, that is, ω min <ω<ω max , and the upper and lower frequency limits ω min and ω max Determined by the actual problem. The meaning of formula 9 is to optimize the parameter θ so that G a and G b At the frequency point with the largest norm deviation, the minimum value is obtained. Formula 9 is only an optional implementation method of the method of the present invention, and technicians can also easily adjust it to other forms. Therefore, the following will continue to use Formula 7 and Formula 8 to solve the optimization problem, "G a →G b ” to simplify the description and ensure generality.

[0063] In conventional compensation, H c The transfer function G of the physical substructure is included Phys , the model needs to be obtained through system identification. c The fitting is affected by the specimen identification results. However, in the field of civil engineering, physical substructure specimens often show strong nonlinearity, so the fitted transfer function often cannot accurately express the actual model, resulting in poor compensation effect and even system instability.

[0064] This embodiment aims to establish a new compensation idea, which eliminates the influence of the uncertain specimen reaction force by establishing an optimization compensation strategy including parameter optimization of the numerical model. Figure 4 As shown, the compensation control strategy G designed by the present invention v Acting on the feedforward channel. Figure 4 The signal flow chart shown in the figure and the mathematical model established according to formula 1 and formula 2 are as follows:

[0065]

[0066] Right now,

[0067]

[0068] For Equation 11, use the optimized fit objective

[0069]

[0070] Arranged

[0071]

[0072] Comparing the two optimization fitting problems of Formula 13 and Formula 8, G in Formula 13 is the model of the numerical substructure, which has an accurate model in the mixed test. And Formula 13 does not contain the model G of the physical substructure Phys , that is, the uncertain interference factor of the test piece reaction force is eliminated. This is one of the key innovations of the present invention, and is a feature that the existing methods in the field of mixed tests do not have.

[0073] In formula 13, since the numerical sub-model is known, the conventional idea is to control the compensation filter G v Design the following structure for optimal fitting:

[0074] G v =G1G+G2 Formula 14

[0075] Among them, G1 and G2 are fitting variables, and G is a known numerical sub-model.

[0076] The limitation of the filter in the form of Equation 14 is that according to Equation 13 Figure 3 The result of the derivation of the mechanism mathematical model is that the fitting target of the variable G1 is the advance link, and the actual computer controller cannot truly achieve the advance link. In actual operation, the transmission model H of the hydraulic servo system d , H v , H F The model identified by white noise may be more complex than the simple mechanism model and more difficult to optimize.

[0077] Therefore, the following compensation filter structure is designed:

[0078] G v =H p G′ Formula 16

[0079] Where G′ is the result of parameter tuning of the original numerical substructure model G while maintaining the model structure. p It is a parameterized multi-order compensation filter, which is divided into characteristic filters of different orders H p,iThe weighted sum of is expressed as:

[0080]

[0081] Among them, k i is the weighting coefficient of the filter. i is the i-th order characteristic filter H p,i The characteristic time parameter, T i Used to adjust the bandpass filter H p,i The passband range of the filter. n is the order of the multi-order compensation filter.

[0082] Then the specific expansion of formula 16 is

[0083]

[0084] According to the optimized compensation filter structure designed by formula 18, the optimization problem defined by formula 13 is solved, where the optimization adjustment parameters are: the typical parameters M, C, K of the numerical substructure model G, and the multi-order filter H p k i and T i Sequence parameters.

[0085] The multi-order compensation filter H used in the present invention p , can be obtained by the characteristic specimen constant T i Sequence and weight k i The sequence conveniently adjusts the effect of the filter on each frequency range, and compared with the method of directly fitting the numerator and denominator of the transfer function, the stability and interpretability of the model parameters are guaranteed. This is the innovation of the present invention.

[0086] The above description process uses Figure 3 The design method is described by the mechanism model, which can be replaced by the identified model in actual implementation. In addition, for the parameters of the numerical sub-model, typical examples are the mass matrix M, the damping matrix C and the stiffness matrix K, which are adjusted according to the specific numerical model in actual implementation.

[0087] In the compensation filter structure designed by formula 18, G ′ The model structure of the numerical substructure G is kept unchanged, and only the parameters are adjusted, so the optimization space is expanded on the basis of ensuring the overall stability and model certainty. This is another key innovation of the present invention.

[0088] like Figure 5 As shown, the specific implementation steps of a hybrid test control compensation method proposed in this embodiment include:

[0089] S1. Use a random algorithm to generate three white noise time histories. The bandwidth of the white noise should cover the frequency band of interest in the hybrid test. The duration of the white noise should be several times longer than the simulation time of the hybrid test to ensure the accuracy of the subsequent identification transfer function.

[0090] S2. The generated white noise time history is saved and converted into a data format recognized by the controller, and the white noise of the displacement command, the white noise of the feedforward, and the white noise of the force feedback are loaded onto the stand on the vibration table, and the displacement feedback of the actuator is recorded;

[0091] The data format converted into data recognized by the controller is obtained by converting white noise into a format recognizable by the controller (for controllers from different manufacturers, the data formats recognized by their host computer software are also different. For example, the Pulsar controller of SERVOTEST recognizes the .sef format, and its own software can be used to convert different formats into the required .sef format).

[0092] S3. After the vibration table is loaded with white noise, the three actuator displacement feedback data are recorded and the transfer function is fitted. The transfer functions H of displacement command white noise input and actuator displacement output, feedforward white noise input and actuator displacement output, and force loading white noise input and actuator displacement output are fitted in the same way. d , H v , and H F .

[0093] S4. According to the identified transfer model, select the appropriate multi-order compensation filter H according to formula 18. p Order n, select the adjustment parameters of the numerical model G to define the optimization fitting problem of formula 13. The typical sequence of optimization parameters is: k in multi-order filter i and T i , and M, C, K in the numerical model.

[0094] S5. Use numerical optimization to perform numerical optimization on the problem defined above, obtain the optimal solution of the above optimization parameters, and complete the feedforward channel transfer model G defined in formula 18 v design.

[0095] S6. Compile the obtained feedforward filter into the controller and conduct formal hybrid experiments.

[0096] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A real-time hybrid test control compensation method, characterized in that: include: Generate multi-channel white noise; The white noise excitation control system is used to obtain the displacement response of the actuator; Fitting the white noise with the displacement response to obtain a transfer model; Designing a compensation filter based on the transfer model, wherein the compensation filter comprises: parameter adjustment of a parameterized multi-order compensation filter and a numerical substructure; Optimizing the parameters of the compensation filter based on the transfer model; Based on the optimized compensation filter, the compensation control of the hybrid test is completed; The multi-channel white noise includes: displacement command white noise, feedforward white noise and force loading white noise; The transmission model includes: displacement instruction transmission model H d , feedforward transfer model H v , force transmission model H F ; Fitting the white noise with the displacement response to obtain the transfer model includes: The displacement command white noise is fitted with the displacement response to obtain the displacement command transfer model H d ; The feedforward white noise is fitted with the displacement response to obtain the feedforward transfer model H v ; The force loading white noise is fitted with the displacement response to obtain the force transfer model H F ; The parameterized multi-order compensation filter H p Split into feature filters H of different orders p,i The weighted sum of , that is, Formula 17; Among them, k i is the weighting coefficient of the filter, T i is the i-th order characteristic filter H p,i The characteristic time parameter, T i Used to adjust the bandpass filter H p,i The passband range; n is the order of the multi-order compensation filter; The compensation filter G v The multi-order compensation filter H p and the parameter setting model G′ of the hybrid test numerical sub-model G, i.e., formula 16; G v =H p G ′ Formula 16 Among them, G ′ is the result of parameter tuning for the original numerical substructure model G while maintaining the model structure. p It is a parameterized multi-order compensation filter.

Images