Test system and control method of test system

Abstract

The invention provides a test system which can eliminate response delay of a load system caused by an actuator for operating a sample and can perform simulation including a model and a real machine in real time with higher precision. Therefore, a test system (1) is configured, and the test system (1) is provided with: a sample (40) which is a part of a real machine to be tested; an actuator (30) that operates the sample (40); a virtual model (11) that operates in association with the sample (40); a control device (20) that controls the actuator (30); a calculation device (10) that calculates a virtual model (11); and an inverse transfer function correction unit (50) that corrects a signal from the virtual model by means of an appropriate inverse transfer function obtained on the basis of a transfer function of a system that includes a communication unit between the sample (40), the actuator (30), and the control device (20), and a communication unit between the calculation device (10) and the control device (20). The control device (20) controls the actuator (30) on the basis of the signal corrected by the inverse transfer function correction unit (50).

Description

Technical Field

[0001] The present invention relates to a test system and a control method of the test system, and more particularly, to a test system using a virtual model and a test specimen which is a part of an actual machine. Background Art

[0002] By using this system to develop new vehicles, the cost and time of development can be reduced. In this test system, for example, sometimes a real object is used as a test sample instead of a part or assembly that is difficult to make a model, or a part of a prototype is used as a test sample. In addition, the verification of the characteristics of the test sample or the verification of the model other than the test sample is implemented by the so-called HILS (Hardware In the Loop Simulation).

[0003] In such a test system, there may be some errors between the behavior of the actual machine and the requirements from the model. Regarding the correction of this error, Patent Document 1 discloses a trajectory data correction device, which is attached to an NC machine tool, etc., and is used to eliminate the contour cutting error caused by the servo characteristics in the numerical control device that performs contour cutting. The correction of the contour error disclosed in Patent Document 1 is to use an inverse transfer function corrector to produce command data that has been pre-corrected for the error for a known command value, and provide it to the control device. In addition, Patent Document 2 discloses the following test system: a model consisting of a tire, a lower mass, and an upper mass and a shock absorber sample are used, and an actuator applies a load to the shock absorber based on a signal from the model to perform a load test. In Patent Document 2, it is recorded that the control waveform is corrected and updated so that the error (response delay) between the control waveform used by the controller and the reference waveform from the model becomes smaller, but the input waveform to the model is constant (for example, a time waveform of tens of seconds). Every time a test is completed, the output of the model (that is, the input to the load device) is compared with the previous output, and the error amount is corrected offline using the inverse transfer function. The test is repeated using the corrected data, and when the error becomes smaller, a formal verification test is performed.

[0004] Prior art literature

[0005] Patent Literature

[0006] Patent Document 1: Japanese Patent Application Laid-Open No. 63-233402

[0007] Patent Document 2: Japanese Patent Application Publication No. 2004-53452 Summary of the invention

[0008] Problems to be solved by the invention

[0009] However, in the HILS test system, the input signal to the model changes in real time, and the input from the model to the load device (control device) and the feedback signal from the load device to the model also change in real time. In this way, the model and the actual load device's feedback to the sample form a closed loop and interact with each other in real time. According to the frequency characteristics of the load device, when the feedback signal is delayed or different in size, the result becomes different from the original characteristics, or diverges and cannot be verified by simulation. In contrast, both Patent Document 1 and Patent Document 2 require pre-correction calculations based on the inverse transfer function, and cannot solve the problem caused by the delay of the load device in real time. In Patent Document 1, since it is model-independent, the command signal is predetermined and does not need to be calculated in real time.

[0010] In addition, in Patent Document 2, during the simulation, the output signal from the model at the same time is not used, but the signal corrected in the previous step is used, so it becomes an open loop, and when one simulation is completed, the signal for the next step is generated offline, so it is a non-real-time HILS system and cannot cope with HILS verification in which the input signal to the model changes in real time. That is, in Patent Document 2, since there is no inverse transfer function that can be calculated in real time in the time domain with high accuracy, there is a disadvantage that it can only be corrected to a predetermined signal.

[0011] An object of the present invention is to provide a test system and a control method for the test system, which can eliminate the response delay of a load system caused by an actuator for moving a specimen, and can perform simulation including a model and an actual machine in real time with higher accuracy.

[0012] Means for solving problems

[0013] In order to solve the above-mentioned problems, a test system according to an embodiment of the present invention comprises: a sample that is a part of an actual machine that serves as a test object; an actuator that causes the sample to move; a virtual model that moves in association with the sample; a control device that controls the actuator; a computing device that computes the virtual model; and an inverse transfer function correction unit that corrects a signal from the virtual model by using a suitable inverse transfer function obtained based on a transfer function of the following system, wherein the system includes the sample, a communication unit between the actuator and the control device, and a communication unit between the computing device and the control device, and the control device controls the actuator based on the signal corrected by the inverse transfer function correction unit.

[0014] In addition, in a control method of a test system of an embodiment of the present invention, the test system includes: a sample that is a part of an actual machine that serves as a test object; an actuator that causes the sample to move; a virtual model that moves in association with the sample; a control device that controls the actuator; and a computing device that computes the virtual model, and the control method includes: a process of correcting a signal from the virtual model by using a suitable inverse transfer function obtained based on a transfer function of the following system, wherein the system includes the sample, the actuator, the control device, and a communication unit between the computing device and the control device; and a process of controlling the actuator based on the signal corrected in the inverse transfer function correction process.

[0015] Effects of the Invention

[0016] According to the above aspect, in the test system, it is possible to eliminate the response delay of the load system caused by the actuator for moving the specimen, and to perform a simulation including a model and an actual machine in real time with higher accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 (a) is a functional block diagram showing a known test system that does not perform correction using a transfer function. Figure 1 (b) is a functional block diagram for explaining a test system that performs correction using an inverse transfer function according to the present embodiment.

[0018] Figure 2 It is used to illustrate Figure 1 Figure 2. Virtual model shown.

[0019] Figure 3 It is used to illustrate Figure 1 A flowchart of a method for preparing an inverse transfer function used in an inverse transfer function correction unit.

[0020] Figure 4 Indicates that it contains Figure 1 An example of a measured transfer function from the output section of the computing device shown in (a) to the control device, the actuator, and the load device of the sample.

[0021] Figure 5 The transfer characteristic approximated as a cubic polynomial is compared with the measured transfer characteristic ( Figure 4 ) is compared with the curve diagram, Figure 5 (a) represents the gain, Figure 5 (b) represents the phase.

[0022] Figure 6 is a diagram for comparing a theoretical inverse transfer function with an inverse transfer function generated by approximation in this embodiment. Figure 6 (a) represents the gain, Figure 6 (b) represents the phase.

[0023] Figure 7 is a diagram showing a comparison between an input signal to the actuator control device that has not been corrected by the inverse transfer function correction unit and an input signal to the actuator control device that has passed through the inverse transfer function correction unit. Figure 7 (b) is to Figure 7 A diagram showing an enlarged portion of (a).

[0024] Figure 8 An example in which the test system of the present embodiment is applied to a HILS test of a vehicle will be described. Figure 8 (a) indicates that Figure 2 The road shown is Figure 1 (a) Figure 1 (b) Displacement of the virtual model input. Figure 8 (b) Figure 8 (f) indicates the direction Figure 1 (a) Figure 1 (b) The input signal of the load device (through Figure 2 The calculation result of the displacement difference between Y2 and Y1 calculated by the model is Figure 8 (c) Figure 8 (g) means from Figure 1 (a) Figure 1 (b) The load of the specimen fed back to the virtual model. Figure 8 (d), (e), (h), and (i) are Figure 8 Enlarged views of a portion of (b), (c), (f), and (g). DETAILED DESCRIPTION

[0025] Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings. The purpose of the drawings used in the description of this embodiment is to illustrate the structure of this embodiment, the positional relationship of each part included in the structure, the function, effect and technical concept, and the present invention is not limited to the specific structure of this embodiment.

[0026] Figure 1 (a) Figure 1 (b) is a functional block diagram for explaining the test system of this embodiment. Figure 1 (a) shows a test system 100 as a comparative example in which correction is not performed using a transfer function. Figure 1(b) shows a test system 1 of the present embodiment that performs correction by an inverse transfer function. The test system 100 of the comparative example and the test system 1 of the present embodiment both include: a sample 40 that is a part of an actual machine to be tested; an actuator 30 that moves the sample 40; a virtual model 11 that moves in association with the sample; a control device 20 that controls the actuator 30; and a computing device 10 that computes the virtual model 11. The control device 20, the actuator 30, and the sample 40 constitute a load device (system) 60.

[0027] Furthermore, the difference between the test system 1 and the test system 100 of the comparative example is that an inverse transfer function correction unit 50 for correcting the signal from the virtual model 11 is provided between the virtual model 11 and the control device 20. The inverse transfer function that the inverse transfer function correction unit 50 has as a function is an appropriate inverse transfer function obtained based on the transfer function of the following system, wherein the system includes the communication unit between the specimen 40, the actuator 30 and the control device 20, and the communication unit between the calculation device 10 and the control device 20. In addition, although the communication unit between the specimen 40, the actuator 30 and the control device 20, and the communication unit between the calculation device 10 and the control device 20 are omitted from illustration, it is clear that the communication unit exists between the specimen 40, the actuator 30 and the control device 20 by the signals S1 and S2 being transmitted from the control device 20 to the actuator 30 and the specimen 40 in sequence. In addition, it is clear that the communication unit exists between the calculation device 10 and the control device 20 by the signals S0, S1, and S2 being transmitted from the calculation device 10 to the control device 20 in sequence.

[0028] The inverse transfer function of the inverse transfer function correction unit 50 is basically Figure 1 In the test system 100 shown in (a), an inverse transfer function of a transfer function established between a signal S0 output from the virtual model 11 and a signal Sf1 indicating a displacement of a specimen 40 driven by an actuator 30. The inverse transfer function is obtained by approximating an inverse transfer function obtained based on the above transfer function of the test system 100 to a suitable inverse transfer function.

[0029] The test system 1, 100 is related to the test of the suspension including the tires of the vehicle. Figure 2), a shock absorber of a real machine of the sample 40 is used instead of the virtual model 11, and the shock absorber A is operated by the actuator 30 controlled by the control device 20. The elements other than the shock absorber A, namely the spring, the tire under the spring, and the body on the spring are used as the virtual model 11. Moreover, the virtual model 11 is installed in a real-time computing device 10 capable of high-speed computing to execute its operations. The signal S3 input to the virtual model 11 represents the displacement given to the tire due to the unevenness of the road surface Z when the vehicle is traveling. The displacement is generated by other road surface models not shown in the figure in the computing device 10. In addition, the description Figure 1 The contents of the virtual model 11 shown in (a) and (b) schematically show the formulas expressing the virtual model 11 and do not represent the actual formulas.

[0030] Figure 2 express Figure 1 (a) Figure 1 The physical system of the virtual model 11 shown in (b) is shown. In the virtual model 11, the sprung mass M1 corresponds to the mass of the vehicle body above the spring Sp, and the unsprung mass M2 corresponds to the mass of the axle, tire T, etc. below the spring Sp. A shock absorber A is arranged between the sprung mass M1 and the unsprung mass M2. The lower surface of the tire T is in contact with the road surface Z, and the position Y0 of the lower end of the tire T is the displacement caused by the unevenness of the road surface Z when the above-mentioned vehicle is running. When the position of the unsprung mass M2 is set to Y2 and the position of the sprung mass M1 is set to Y1, the input to the shock absorber A is the displacement caused by the difference between the position Y2 and the position Y1. The output from the shock absorber A is a load, and the load acts on the unsprung mass M2 and the sprung mass M1.

[0031] When the HILS of the actual machine is used instead of the theoretical model, the shock absorber A gives the difference between the position Y2 and the position Y1 to the load device 60, and the load generated by the shock absorber A driven by the load device 60 is fed back to the virtual model 11, playing the same role as the shock absorber A of the virtual model 11, and the calculation of the virtual model 11 is implemented.

[0032] Refer again Figure 1 (a) and (b), the computing device 10 performs sampling at regular intervals (for example, 0.5 ms) to calculate a signal S0 (Y2-Y1), which represents the displacement (Y1, Y2) of each mass in the virtual model 11 and the displacement of the shock absorber A based on the input Y0 from the road surface Z.

[0033] The control device 20 is configured to include a comparison unit 21 and a control unit 22. The comparison unit 21 calculates the value of the model 11 (in Figure 1The difference between the signal S0 (signal S1: the shock absorber displacement Y2-Y1) and the signal Sf1 fed back from the sample 40 in (b) via the inverse transfer function correction unit 50 is output to the control unit 22. The control unit 22 performs PID control based on the difference signal from the comparison unit 21, thereby controlling the movement amount of the actuator 30. In addition, the control device 20 can be set to, for example, an ECU (Electronic Control Unit) in the HILS test.

[0034] The actuator 30 operates according to the signal S2 from the control unit 22, so that the sample 40 (shock absorber A) performs an action (displacement Y2-Y1) related to its test evaluation item. The actuator 30 can be a hydraulic actuator, an actuator driven by a motor, etc. The sample 40 is driven by the actuator 30. In addition, the action is detected by a predetermined sensor (not shown), and the detected signal Sf1 is output to the comparison unit 21 for the control of the actuator. The signal Sf2 (load) is output to the virtual model 11 for the next calculation.

[0035] When the shock absorber A is simulated only by the model instead of the real machine, the displacement S0 of the shock absorber calculated on the model is input to the shock absorber of the model at the same time, and the load calculated by the shock absorber of the model is used to perform the calculation in the next step. However, in the HILS system using the shock absorber A of the real machine, a delay and amplitude change caused by the control device 20 and the actuator 30 occur between the signal Sf1 representing the displacement of the shock absorber A and the signal S0. Therefore, the load signal generated from the shock absorber A of the real machine is also delayed compared to the original load signal, and accurate simulation cannot be performed. Depending on the conditions, simulation divergence may occur.

[0036] In the test system 1 of the present embodiment, the inverse transfer function correction unit 50 that functions as the inverse transfer function corrects the signal S0 from the virtual model 11, thereby making it possible to advance the signal S1 input to the comparison unit 21 by the same amount as the amount delayed by the load device 60 including the control device 20 and the actuator 30 relative to the signal S0. As a result, the test system 1 can eliminate the delay in time of the signal Sf1 indicating the displacement of the shock absorber A caused by the actuator 30 and the like relative to the signal S0 from the virtual model 11, and can perform simulation with higher accuracy.

[0037] That is, in Figure 1 In the system (a), S0=S1, and the signal Sf1 is delayed by δt seconds compared to the signal S1. Therefore, the signal Sf2 is also delayed by δt seconds compared to the simulation by the model alone. Figure 1In (b), the signal S1 is advanced by δt seconds compared to the signal S0 by the correction of the inverse transfer function correction unit 50. Since the relationship between the signal Sf1 and the signal S1 does not change, the signal Sf1 is delayed by δt seconds relative to the signal S1 which is advanced by δt seconds relative to the signal S0, thereby eliminating the delay relative to the signal S0 and also eliminating the delay of the signal Sf2 representing the load input to the virtual model 11.

[0038] In addition, Figure 1 In (b), for the sake of explanation, the inverse transfer function correction unit 50 is shown between the computing device 10 and the control device 20 of the load device 60, but in the implementation state, the inverse transfer function correction unit 50 is installed on either the computing device 10 or the control device 20.

[0039] (Generation of inverse transfer function)

[0040] Figure 3 Yes means Figure 1 Flowchart of the process of generating the inverse transfer function in the inverse transfer function correction unit 50. Figure 1 The test system 1 is carried out before operation.

[0041] First, in step S101, the control device 20 inputs a signal S2 representing input data of each of a plurality of frequencies to the actuator 30 without being connected to the virtual model 11, thereby applying a vibration input to the sample 40 (shock absorber), and obtaining the displacement Sf1 of the shock absorber A at this time. Vibration can be applied using sine wave sweep excitation or random wave excitation within a predetermined frequency range. The predetermined frequency range is determined by the characteristics of the simulated sample 40. In the case of a shock absorber of a vehicle, for example, it can be set to be higher than 0 Hz and up to 100 Hz.

[0042] Table 1 below shows an actual measurement example of this embodiment. The gain and phase (deg) of signal Sf1 relative to signal S0 are shown for each frequency of vibration. In addition, the gain is represented by both dB display and real number display. The gain is the amplitude ratio of signal Sf1, and a negative phase means a time delay.

[0043] [Table 1]

[0044]

[0045] Figure 4 The transfer function represented by the measured data in Table 1 is shown in a Bode plot. Figure 4 In the figure, G represents gain and P represents the frequency characteristic of phase.

[0046] When the measurement data in Table 1 is expressed as a complex number, the following Table 2 is obtained. That is, the relationship between the gain and the phase in Table 1 can be expressed as a complex number as in the following formula.

[0047] G(jω k )=Ae iφ =A(cosφ+jsinφ)=Acosφ+jAsinφ

[0048] =R m +j I m

[0049] Here, A is the gain (actual amount), is the phase (deg), so the values ​​in Table 1 become the contents of Table 2 according to the above formula.

[0050] [Table 2]

[0051]

[0052] Next, in step S102 , an approximate expression of the transfer function is obtained using the transfer function data expressed in discrete numerical values ​​acquired in step S101 .

[0053] (General form of transfer function)

[0054] In order to obtain a transfer function that matches the measurement data, first, as a comparison, a transfer function represented by a general formula is obtained. The general formula of the transfer function can be expressed as the following formula (1).

[0055]

[0056] Then, in a system using the actuator of the present embodiment, based on experience, the above-mentioned equation (1) can be approximated by the following equation whose numerator is 1.

[0057]

[0058] When formula (2) is expressed in complex form,

[0059]

[0060] in,

[0061] In formula (3), when n is an even number,

[0062] R e =a n (jω) n +a n-2 (jω) n-2 +…-a2ω 2 +1

[0063] =[(jω) n ,(jω) n-2 ...,-ω * ][a n , a n-2 ,…,a2]′+1=A RE a re +1···Formula (4)

[0064] I e =j n-2 a n-1 ω n-1 +j n-4 a n-3 ω n-3 +…+a1ω

[0065] =[j n-2 ω n-1 ,j n-4 ω n-3 ,…,ω][a n-1 / a n-3 ,…,a1]′=A IM a im ···Formula (5)

[0066] In addition, in formula (3), when n is an odd number,

[0067] R e =a n-1 (jω) n-1 +a n-3 (jω) n-3 +…-a2ω 2 +1

[0068] =[(jω) n-1 ,(jω) n-3 ,...,-ω 2 ][a n-1 , a n-3 ,…,α2]′+1=A RE a re +1···Formula (6)

[0069] I e =j n-1 a n ω n +j n-3 a n-2 ω n-2 +…+a1ω

[0070] =[j n-1 ω n ,j n-3 ω n-2,...,ω][α n ,a n-2 ,…,a1]′=A IM a im Formula (7)

[0071] In the above formulas (4) to (7), the exponents of the imaginary symbol j are all even numbers, so the calculation result is [-1] or [1]. RE a re and A IM a im is the product of matrices. RE , A IM It can be calculated as long as the angular frequency ω is known.

[0072] (Transfer function based on measurement data)

[0073] Next, the measurement data obtained above are compared with the general formula of the transfer function to obtain a transfer function based on the measurement data.

[0074] The measurement data shown in Table 2 are expressed in plural form.

[0075] G(jω k )=Ae jφ =A(cosφ+jsinφ)=Acosφ+jAsinφ=R m +jI m Formula (8)

[0076] In addition, the complex expression of formula (2) is

[0077]

[0078] When comparing equation (8) with equation (9), we obtain equation (10) and equation (11).

[0079]

[0080] Here, by approximating the numerical gain Ae with the gain A of the measurement data, the equations (10) and (11) are expressed as the equations (12) and (13), respectively.

[0081] R e =R m / A 2 Formula (12)

[0082] I e =-I m / A 2 Formula (13)

[0083] In equations (12) and (13), R for each frequency m,I m , because A is obtained from the measured data, the right side of equation (12) and equation (13) is also known. In addition, in equation (12) and equation (13), the left side is replaced from equation (4) to equation (7), and the following equations (14) and (15) are obtained.

[0084] A RE a re =B RE ···Formula (14)

[0085] A IM a im =B IM ···Formula (15)

[0086] In equations (14) and (15), the right side B RE , B IM It is represented by a matrix because it is a set of measured data. When plotting it, the A on the left side multiplied by the plotted curve is obtained by the least squares method. RE a re and A IM a im In formula (2), the coefficient a can be obtained from the measured data. re and a im .

[0087] When equation (2) is approximated by a cubic equation with n=3, the equation expressed by the following equation (16) is:

[0088]

[0089] That is, it becomes:

[0090] A RE =[-ω 2 ] re =[α2]

[0091] A IM =[-ω 3 ,ω] a re =[a3,a1]

[0092] In the case of the actual measurement example in Table 1, a1, a2, and a3 are as follows:

[0093] a3=4.634×10 -9

[0094] a2=6.609×10 -6

[0095] a1=4.058×10 -3

[0096] When a1, a2, and a3 are substituted into equation (16), equation (17) is derived as follows.

[0097]

[0098] Figure 5 (a) and (b) are Bode plots of the transfer function approximated by the cubic polynomial shown in the above equation (17). Figure 4 The gain and the phase are shown by comparing the gain G and the phase P obtained based on the measurement data. Figure 5 Rg and Rp shown in (a) and (b) respectively represent the gain and phase obtained by measurement, and Ng and Np respectively represent the gain and phase obtained by calculation. Figure 5 As shown in (a) and (b), the gain error at 100 Hz is about 1 dB, and the phase error is within 5 degrees, which is sufficient accuracy for shock absorber simulation. Depending on the system, if the accuracy is not sufficient, n can be increased and approximation can be performed through the same calculation.

[0099] (Factorization of transfer function)

[0100] Reference again Figure 3 In step S103, the transfer function obtained in step S102 is decomposed into a secondary vibration element and a primary delay element. This is to facilitate understanding of the influence of each element when obtaining the inverse transfer function (step S104). The primary delay element is represented by the following equation (18):

[0101]

[0102] In addition, in the following equation (19), the solution of the quadratic equation of the denominator of the secondary vibration element is a complex number when ζ<1, and is the product of two first-order delay systems when ζ≥1.

[0103]

[0104] Based on the above, the transfer function represented by equation (2) can be transformed as shown in equation (20).

[0105]

[0106] Then, by solving the following equation (21), the factorized coefficients can be found.

[0107] α n s n +a n-1 s n-1 +…+a1s 1 α+1=0 ··Formula (21)

[0108] In the solution, the two solutions that become complex numbers are in a conjugate relationship, so they become secondary vibration elements. The specific solution is the following formula:

[0109]

[0110] According to equation (22) and equation (23), the following relationship is obtained.

[0111]

[0112]

[0113] In the example of formula (17), the denominator

[0114] 4.634×10 -9 s 3 +6.609×10 -6 s 2 +4.058×10 -3 s 1 +1=0

[0115] When solving, the following solution is obtained.

[0116] s1=439.95+j448.84

[0117] s2=439.95-j448.84

[0118] s3=546.3

[0119] s1 and s2 are in a conjugate relationship, so they become secondary vibration elements. That is,

[0120]

[0121] ω2=s3=546.3rad / s···(87Hz)

[0122] Based on the above results, equation (17) is rewritten as follows.

[0123]

[0124] Reference again Figure 3In step S104, a suitable inverse transfer function is obtained based on the transfer function decomposed as described above. Generally speaking, taking the inverse of the obtained transfer function formula will result in an inverse transfer function, but this is not a suitable transfer function. In this case, actual time calculation cannot be performed, delay cannot be corrected by the inverse transfer function, and correct HILS simulation cannot be implemented. In order to obtain a suitable transfer function, it is necessary to set a term of degree s that is the same as or greater than the degree s of the numerator in the denominator. That is, the so-called suitable transfer function is a transfer function that satisfies "the degree s of the numerator ≤ the degree s of the denominator". In this embodiment, considering the amount of calculation and computing power, a term of degree s that is the same as the degree s of the term in the numerator is set in the denominator.

[0125] In the case of this embodiment, G(s)=1 / (((s / ω1) 2 +2ζ(s / ω1)+1)˙(s / ω2)) to replace equation (24), the denominator is set to the same s as the numerator. 3 The appropriate transfer function of the term is expressed by the following equation (25).

[0126]

[0127] Formula (25) contains the parameter ω a1 ,ω a2 ,ω b1 ,ω b2 , a , b . This embodiment adjusts such parameters, taking into account the influence of the first-order delay element (low-pass filter) and the second-order vibration element, and obtains a suitable inverse transfer function of equation (25) in a manner close to the inverse transfer characteristics of the transfer function of the test system. The determination of each parameter can be performed manually or automatically. The following is an example of the conditions for determining the parameters.

[0128] (1) Regarding the denominator of equation (25), in order to reduce the influence of phase delay, the frequency ω a1 ,ω a2 It is set to be 4 times or more of ω1, ω2 or the maximum frequency that the calculation device can simulate.

[0129] (2) With the same denominator, in order to reduce the influence of phase delay, let ζ a Is less than 0.2.

[0130] (3) Regarding the numerator of equation (25), the phase is slightly delayed and the gain is reduced due to the quadratic and primal terms in the denominator. Therefore, ω b1 ,ω b2 It is set to a smaller value than ω1 and ω2.

[0131] (4) Adjust the parameters as above and obtain the Bode plot (gain and phase frequency characteristics) of the inverse transfer function based on the parameters. Then, repeat the above adjustment until the difference between the obtained gain and the required gain is within 3 dB, and the phase difference is within 10 degrees. Here, the "required gain or phase" is the gain and phase frequency characteristics represented by the inverse transfer function (hereinafter referred to as the "theoretical inverse transfer function") of the transfer function obtained by equation (24).

[0132] The above setting is one of the standards, and the present embodiment is not limited to adjusting the parameters by such an order or setting.

[0133] Figure 6 (a) and (b) are Bode diagrams showing the inverse transfer function obtained in this embodiment by comparison with the theoretical inverse transfer function. Figure 6 (a) represents the frequency characteristic of gain (dB), Figure 6 (b) shows the frequency characteristics of phase (deg). Figure 6 The curves Ig and Ip in (a) and (b) respectively represent the gain and phase of the theoretical inverse transfer function, and the curves Mg and Mp respectively represent the gain and phase of the inverse transfer function obtained in this embodiment. As described above, the difference between the obtained gain and phase frequency characteristics and the theoretical inverse transfer function is made within 3 dB for gain and within 10 degrees for phase, and the response delay (phase delay) in the test system of this embodiment is fully suppressed.

[0134] The gain and phase tolerances described here are for the damper of the embodiment and are not applicable to all systems. They vary depending on the characteristic frequency range of the system being evaluated.

[0135] In this embodiment, by maximizing the parameter ω a1 ,ω a2 , which enables the parameter ω b1 , b ,ω b2 By approaching the parameters ω1, ζ1, and ω2, it is possible to approach the theoretical inverse transfer function at higher frequencies. In this case, calculations need to be performed with a shorter sampling time, that is, a higher-performance computing device or control device is required. When the frequency range of the system evaluated by HILS is low, the agreement at lower frequencies is sufficient, so the parameter ω can be reduced. a1 ,ω a2 , reduce the computational load and build a cheap evaluation system.

[0136] Table 3 shows the parameters ω1, ω2, ζ1 of the theoretical inverse transfer function and the parameters ω of the inverse transfer function generated by approximation. a1 ,ω a2 ,ωb1 ,ω b2 , a , b .

[0137] [Table 3]

[0138]

[0139] (Evaluation based on inverse transfer function)

[0140] Refer again Figure 3 , when a suitable inverse transfer function is found, in step S105, as Figure 1 As shown in (b), the inverse transfer function is installed in the test system 1. In addition, the inverse transfer function can be installed on the same computing device side as the model calculation, or installed in the unillustrated computing unit of the control device 20 of the sample 40 as the actual machine. In addition, the inverse transfer function is usually installed on the side with higher computing power.

[0141] Figure 7 (a) and (b), Figure 8 (a) to (i) are for explaining the effect of the inverse transfer function correction unit 50 which functions as an inverse transfer function according to the present embodiment. Figure 7 (a) is to have no Figure 1 A graph comparing a signal (command) c2 input to the actuator 30 corrected by the inverse transfer function correction unit 50 of the test system 100 and a signal (command) c1 input to the actuator 30 after passing through the inverse transfer function correction unit 50 of the test system 1 is shown. Figure 7 (b) is to Figure 7 This is a diagram showing an enlarged view of the signals c1 and c2 in the range of 30 s to 30.1 s on the horizontal axis in (a). Figure 7 (a) Figure 7 In (b), the horizontal axis represents time (s), and the vertical axis represents displacement (mm).

[0142] Regarding the relationship between the delay time and phase of a signal, if the frequency is set to F and the delay is set to δT, the phase delay Φ is expressed by the following equation (26).

[0143] φ=360°×F×δT···Formula (26)

[0144] exist Figure 7 In the example shown in (a), when δT = 0.0043s and F = 60Hz, a delay of Φ = 92.8° (approximately 1 / 4 cycle) occurs (the time until the signal Sf2 output from the sample 40 is input to the computing device 10), and an accurate simulation cannot be performed. Figure 7As shown in (b), the displacement of signal c1 changes at a timing earlier than signal c2 (signal c1 is advanced compared to signal c2). In this embodiment, due to the advancement of signal c1, signals Sf1 and Sf2 output by sample 40 are also advanced, and the signal delay in the computing device 10 disappears, so accurate simulation can be performed.

[0145] Figure 8 (a) indicates that Figure 2 The road surface shown in Z direction Figure 1 (a) Figure 1 (b) The virtual model 11 inputs a signal S0 representing displacement. Figure 8 (b) represents the signal S1 input to the load device 60 of the test system 100, Figure 8 (c) represents the signal Sf2 fed back from the sample 40 of the test system 100 to the virtual model 11, Figure 8 (d) Enlarged representation Figure 8 (b) of region B, Figure 8 (e) Enlarged representation Figure 8 (c) area C. Figure 8 The horizontal axes of (a) to (e) are all time. Figure 8 The vertical axis of (a), (b), and (d) is displacement. Figure 8 The vertical axis of (c) and (e) is the load.

[0146] Figure 8 (f) represents the signal S1 input to the load device 60 of the test system 1, Figure 8 (g) represents the signal Sf2 fed back from the sample 40 of the test system 1 to the virtual model 11, Figure 8 (h) Enlarged representation Figure 8 The region F of (f), Figure 8 (i) Enlarged representation Figure 8 The area G of (g). Figure 8 The horizontal axes of (f) to (i) are all time. Figure 8 The vertical axis of (f) and (h) is displacement. Figure 8 The vertical axis of (g) and (i) is the load. That is, Figure 8 (b) to Figure 8 (e) indicates that the correction without the inverse transfer function correction unit 50 is used. Figure 1 (a) Results of the test system 100. Figure 8 (f) to Figure 8 (i) shows the result of using the test system 1 which performed the calibration by the inverse transfer function calibration unit 50 .

[0147] In this embodiment, the road surface input to the virtual model 11 is a simulated road surface with a concavity of ±50 mm. Figure 8 As shown in (a), it changes sinusoidally along the vehicle travel direction. It changes at a cycle of 0.5 Hz in the simulated vehicle speed.

[0148] In this test, a real shock absorber was used for sample 40. Figure 8 As shown in (b) and (f) of FIG. 1 , for a sinusoidal road input, the displacement input to the shock absorber of the actual machine also varies smoothly close to a sinusoidal wave, but the change gradient of the shock absorber load is discontinuous and contains strong nonlinear and high-frequency components. This load signal is fed back to the model (the part shown in the area B and F in the figure).

[0149] like Figure 8 As shown in (b) to (e), without correction based on the inverse transfer function, Figure 8 (b) Figure 8 In (c), the parts shown in regions B and C are Figure 8 (d) Figure 8 In (e), the influence of high-frequency components less than about 40 Hz gradually increases from time 2.4s to time 3.0s, superimposing the vibration of high-frequency components that should not have been in the instructions to the actual machine calculated from the model. Due to this vibration instruction, shock absorber A also vibrates at a high frequency, and the load signal to the virtual model 11 also vibrates. This is due to the phase delay of the high-frequency components in the above-mentioned load signal, which is caused by an erroneous calculation result. On the other hand, Figure 8 (f) to Figure 8 As shown in (i), when the correction based on the inverse transfer function is performed, the above-mentioned vibration disappears, and a correct HILS evaluation can be performed. In the case of this embodiment, when the actual vehicle is running, it is known that there is no such vibration, so it can be assumed that it is a simulation error, but in the case of a new product with unknown characteristics, it becomes an erroneous result without correction based on the inverse transfer function.

[0150] In addition, the timing, frequency, and magnitude of the above vibrations vary depending on the test conditions, test body, and characteristics of the load device. In other words, oscillations are generated due to changes in the phase and amplitude of the frequency contained in the load component. Figure 8 In the examples shown in (b) and (c), the frequency and time of oscillation are within the areas B and C shown in the figure, but the present embodiment is not limited to such examples, and the frequency and time (timing) of oscillation vary according to the test conditions.

[0151] Here, a suitable transfer function is a transfer function that can be strictly realized in the real world, specifically, a transfer function that satisfies the degree of s in the numerator ≤ the degree of s in the denominator. That is, an inappropriate transfer function is divided into suitable terms and inappropriate terms. If it is expanded, the output of the test system is obtained. In the case where the transfer function is inappropriate, the time differential of the input under the action of the inappropriate term is included in the output. The time differential becomes the main reason that hinders the realization in the real world. As described above, the present embodiment corrects the output signal of the virtual model 11 by the inverse transfer function of the suitable transfer function, that is, the suitable inverse transfer function. As a result, the present embodiment can eliminate the response delay in the test system caused by the actuator 30, etc., and can perform the simulation performed in the test system 1 with higher accuracy.

[0152] The first invention described above is a test system comprising: a specimen that is a part of an actual machine that serves as a test object; an actuator that moves the specimen; a virtual model that moves in association with the specimen; a control device that controls the actuator; a computing device that computes the virtual model; and an inverse transfer function correction unit that corrects a signal from the virtual model by using a suitable inverse transfer function obtained based on a transfer function of the following system, wherein the system includes the specimen, a communication unit between the actuator and the control device, and a communication unit between the computing device and the control device, and the control device controls the actuator based on the signal corrected by the inverse transfer function correction unit.

[0153] A second aspect of the present invention is the test system of the above aspect, wherein the inverse transfer function correction unit is mounted on the calculation device.

[0154] A third aspect of the present invention is the test system of the above aspect, wherein the inverse transfer function correction unit is mounted on a real machine control calculation unit that operates the sample.

[0155] A fourth aspect of the present invention is the test system of the above aspect, wherein the transfer function is an approximate transfer function whose numerator is a zero-order Laplace variable.

[0156] A fifth aspect of the present invention is the test system of the above aspect, wherein the transfer function is an approximate transfer function whose denominator is a cubic Laplace variable.

[0157] The sixth invention is a control method for a test system, wherein the test system includes: a sample that is a part of an actual machine that serves as a test object; an actuator that causes the sample to move; a virtual model that moves in association with the sample; a control device that controls the actuator; and a computing device that computes the virtual model, wherein the control method includes the following steps: an inverse transfer function correction step of correcting a signal from the virtual model by using a suitable inverse transfer function obtained based on a transfer function of the following system, wherein the system includes the sample, the actuator, the control device, and a communication unit between the computing device and the control device; and a step of controlling the actuator based on the signal corrected in the inverse transfer function correction step.

[0158] A seventh aspect of the invention is the control method of the test system, wherein a transfer function of a system including the actuator and the sample is divided into a real part and an imaginary part, and the transfer function is obtained by an approximation based on a least squares method.

[0159] An eighth aspect of the present invention is the control method of the test system, wherein the transfer function is divided into a primary delay system and a secondary vibration system to obtain parameters of an approximate inverse transfer function.

[0160] Description of Reference Numerals

[0161] 1. 100 test system

[0162] 10Computing device

[0163] 11 Virtual Model

[0164] 20 Control device

[0165] 21 Comparison Department

[0166] 22. Control Unit

[0167] 30 Actuator

[0168] 40 samples

[0169] 50 Inverse transfer function correction unit

[0170] 60 Load Device

[0171] A shock absorber.

Claims

1. A test system, characterized in that: have: A test specimen that is part of the actual machine that is the subject of the test; an actuator that actuates the specimen; a virtual model that acts in association with the specimen; a control device that controls the actuator; A computing device for computing the virtual model; and an inverse transfer function correction unit that corrects the signal from the virtual model by using an appropriate inverse transfer function obtained based on a transfer function of the following system, wherein the system includes the sample, a communication unit between the actuator and the control device, and a communication unit between the computing device and the control device, The control device controls the actuator based on the signal corrected by the inverse transfer function correction unit.

2. The test system according to claim 1, characterized in that: The inverse transfer function correction unit is mounted on the computing device.

3. The test system according to claim 1, characterized in that: The inverse transfer function correction unit is mounted on a real machine control calculation unit that operates the sample.

4. The test system according to claim 1, characterized in that: The transfer function is an approximate transfer function whose numerator is a Laplace variable of order 0.

5. The test system according to claim 4, characterized in that: The transfer function is an approximate transfer function whose denominator is a cubic Laplace variable.

6. A control method for a test system, The test system comprises: A test specimen that is part of the actual machine that is the subject of the test; an actuator that actuates the specimen; a virtual model that acts in association with the specimen; a control device that controls the actuator; and a computing device for computing the virtual model, It is characterized in that The control method comprises the following steps: an inverse transfer function correction step of correcting the signal from the virtual model by using an appropriate inverse transfer function obtained based on a transfer function of the following system, wherein the system includes the sample, a communication unit between the actuator and the control device, and a communication unit between the computing device and the control device; and A step of controlling the actuator based on the signal corrected in the inverse transfer function correction step.

7. The control method of the test system according to claim 6, characterized in that: A transfer function of a system including the actuator and the sample is divided into a real part and an imaginary part, and the transfer function is obtained by an approximation formula based on a least square method.

8. The control method of the test system according to claim 6, characterized in that: The transfer function is divided into a primary delay system and a secondary vibration system to obtain parameters of an approximate inverse transfer function.

Images