Non-Gaussian vibration control device

Abstract

The present invention provides a vibration control device capable of having at least two different kurtosis values ​​for jerk, acceleration, displacement, and velocity. [Solution] The vibration generator 2 receives a D / A converted drive signal and vibrates the test specimen 4. The acceleration sensor 6 detects the vibration of the test specimen and outputs it as a response signal. The A / D converter 10 converts the response signal to digital. The response PSD calculation means 12 performs a Fourier transform (FFT) on the response signal and calculates the response PSD (power spectral density) as its frequency characteristic. The control PSD calculation means 14 determines whether the response PSD matches the target PSD and modifies the control PSD so that the two match. In other words, feedback control is performed. The phase control means 18 divides the frequency domain into multiple parts and generates a phase controlled so that the frequency distribution of phases from 0 to 2π is different between the divided frequency domains. The drive waveform calculation means 16 applies this phase to the control PSD and calculates the drive waveform.

Description

Technical Field

[0001] This invention relates to a vibration control device that controls a shaker so as to apply a target non-Gaussian vibration to a specimen.

Background Art

[0002] In order to simulate the influence of vibrations applied during transportation or operation, a vibration test is conducted to apply a desired vibration to a specimen. In the vibration test, a vibration control device controls the shaker so as to obtain a desired vibration.

[0003] If the actually applied vibration is recorded and this vibration can be applied to the specimen, an accurate vibration test can be performed. However, since an enormous recording capacity is required to record and reproduce the actual vibration waveform itself, tests are often conducted by extracting the characteristics of the vibration and reproducing those characteristics.

[0004] For example, there is a test in which the characteristics of the vibration are replaced with a single frequency and level, and a vibration by a sine wave is applied. In this case, since only a sine wave is output, the control is easy, but there is a problem that the deviation from the actually applied vibration is too large.

[0005] Therefore, a random vibration test has been widespread in which the characteristics of the vibration are regarded as frequency characteristics (power spectral density), and a vibration having the target power spectral density is applied to the specimen.

[0006] FIG. 12 shows a conventional vibration control device for a random vibration test disclosed in Patent Document 1. Since the shaker 2 itself also has frequency characteristics, even if a vibration having a target spectrum is applied, the vibration as such cannot be applied to the specimen 4. Therefore, feedback control is performed by the vibration control device so that the spectrum (response spectrum) of the vibration waveform of the specimen 4 becomes equal to the target spectrum.

[0007] The test specimen 4, fixed to the vibration exciter 2, is vibrated by the vibration exciter 2. The vibration of the test specimen 4 is detected by the acceleration sensor 6 and converted into a digital response signal by the A / D converter 8. The PSD calculation means 20 performs a Fourier transform on the response signal and calculates the response PSD.

[0008] The control PSD calculation means 22 compares the target PSD and the response PSD and calculates the control PSD so that they are equal. The drive signal generation means 15 generates a drive signal by applying random phases to each frequency component of the drive PSD, which is created based on the control PSD and the transfer function, and performing an inverse Fourier transform. The D / A converter 16 converts the generated drive signal into an analog signal and supplies it to the exciter 2.

[0009] In this way, it is possible to control the vibrations applied to the specimen 4 to have the target spectrum.

[0010] However, while the vibration control device described in Patent Document 1 can apply vibrations with the target spectrum to the test specimen 4, the probability density distribution of those vibrations is a Gaussian distribution (normal distribution), as shown by the solid line in Figure 13A. In other words, the vibration waveform is as shown in Figure 13B.

[0011] However, real-world vibrations, as shown in Figure 13C, often produce accelerations that are larger than those of a Gaussian distribution, or accelerations that feel like they are being pushed upwards from below (in the case of truck vibrations). In other words, they often follow a non-Gaussian distribution, as shown by the dashed line in Figure 13A.

[0012] Here, the degree of steepness in the probability distribution of oscillations is expressed as kurtosis. Kurtosis K is calculated by the following formula.

[0013]

number

[0014] Here, n is the number of samples of the amplitude waveform, and xi is the amplitude of the waveform. x This is the average value of the waveform amplitude.

[0015] In the case of a Gaussian distribution, as shown by the solid line in Figure 13A, the kurtosis K = 3, while in the case of a non-Gaussian distribution, as shown by the dashed line, the kurtosis K is larger.

[0016] The vibration control device described in Patent Document 1 could not apply vibrations with non-Gaussian characteristics to the test specimen.

[0017] Therefore, in order to solve the above problems, the inventors have already made the invention described in Patent Document 2. Figure 14 shows the configuration of the vibration control device disclosed in Patent Document 2.

[0018] The vibration exciter 2 receives the D / A converted drive signal and vibrates the test specimen 4. The acceleration sensor 6 detects the vibration of the test specimen and outputs it as a response signal. The A / D converter 8 converts the response signal to digital. The response PSD calculation means 20 performs a Fourier transform (FFT) on the response signal and calculates the response PSD (power spectral density) as its frequency characteristic.

[0019] The control PSD calculation means 22 determines whether the response PSD matches the target PSD and modifies the control PSD so that the two match. In other words, it performs feedback control.

[0020] The Gaussian waveform generation means 26 generates a Gaussian waveform by applying a random phase to the control PSD and performing an inverse Fourier transform (inverse FFT). The conversion means 28 converts the Gaussian waveform into a non-Gaussian waveform with desired characteristics (for example, a waveform with high kurtosis) using a ZMNL function or the like.

[0021] The phase calculation means 36 performs a Fourier transform on the generated non-Gaussian waveform and extracts the phase of each frequency. The second non-Gaussian waveform generation means 38 provides the extracted phase to the control PSD, performs an inverse Fourier transform, and generates a second non-Gaussian waveform.

[0022] The waveform control means 32 calculates and outputs a drive signal so that the specimen 4 vibrates with a second non-Gaussian waveform in accordance with the control characteristics considering the transfer function of the system including the vibrator 2 and the specimen 4.

[0023] The D / A converter 16 converts the digital data drive signal into an analog signal. This drive signal is applied to the vibrator 2 via an amplifier (not shown).

[0024] As described above, vibrations having desired non-Gaussian characteristics (for example, characteristics with a large kurtosis) can be applied to the specimen 4.

Prior Art Documents

Patent Documents

[0025]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0026] In a vibration control device as described in Patent Document 2, tests can be performed with vibrations having desired non-Gaussian characteristics. However, when attempting to conduct a vibration test with vibrations having a large kurtosis, for example, the following problems may occur.

[0027] In vibration generators, upper limits (ratings) are defined for each of the vibration acceleration, displacement, and velocity. Therefore, when conducting vibration tests with high kurtosis, the test cannot be performed unless the ratings for acceleration, displacement, velocity, and jerk are all satisfied. Generally, in conductive vibration generators, the constraints on displacement are stricter than those on acceleration, so a situation may arise where the acceleration rating of the vibration generator is satisfied, but the displacement rating is not. Therefore, in order to conduct vibration tests in such cases, the kurtosis of the displacement must be smaller than the displacement rating of the acceleration. For this reason, a vibration control device is needed that can make the kurtosis of acceleration, displacement, velocity, and jerk different from each other.

[0028] Furthermore, in real-world vibrations, the kurtosis of acceleration, displacement, velocity, and jerk may differ. Therefore, in order to accurately simulate real-world vibrations in the intended test, it may be necessary to use different kurtosis values ​​for acceleration, displacement, velocity, and jerk.

[0029] The purpose of this invention is to solve the above-mentioned problems and to provide a vibration control device that can make the kurtosis of at least two of the acceleration, displacement, velocity, and jerk different. [Means for solving the problem]

[0030] The following lists some independent features of this invention. These features are not necessarily required to be combined, but can be combined as desired.

[0031] (1)(2) The vibration control device according to the present invention includes: a response PSD calculation means that calculates a response PSD by performing a Fourier transform on a response waveform from a vibration sensor that measures the vibration of a test specimen being vibrated by a vibration generator that operates based on a drive waveform; a control PSD calculation means that compares the response PSD with a target PSD and calculates a control PSD such that the response PSD becomes equal to the target PSD; a control waveform calculation means that performs an inverse Fourier transform on each frequency component based on the control PSD by assigning a phase to it and outputs a control waveform; a drive waveform calculation means that calculates a drive waveform to be applied to the vibration generator so that the test specimen vibrates with the control waveform; and a phase control means that divides the frequency domain into at least two and controls the kurtosis of at least two temporal changes in the jerk, acceleration, velocity, and displacement of the control waveform by making the frequency distribution of the phases assigned to each frequency component in the control waveform calculation means in the range of 0 to 2π different in the divided frequency domains.

[0032] Therefore, at least two kurtosis values ​​of the temporal changes in jerk, acceleration, velocity, and displacement can be controlled differently.

[0033] (3) The vibration control device according to this invention is characterized in that the phase control means increases the bias of the phase frequency distribution in the low-frequency region compared to the high-frequency region, thereby reducing the kurtosis of jerk or acceleration compared to either velocity or displacement.

[0034] Therefore, the kurtosis of jerk and acceleration can be made relatively smaller.

[0035] (4) The vibration control device according to this invention is characterized by feedback control of the bias in the frequency distribution of the phase such that any two of the acceleration, velocity, displacement, and jerk each have different desired kurtosis.

[0036] Therefore, the desired kurtosis can be achieved through feedback control.

[0037] (5) The vibration control device according to this invention is characterized in that the phase control means increases the bias of the phase frequency distribution in the high frequency region compared to the low frequency region, thereby increasing the kurtosis of jerk or acceleration compared to either velocity or displacement.

[0038] Therefore, the kurtosis of acceleration can be made relatively larger.

[0039] (6) The vibration control device according to this invention is characterized by feedback control of the bias in the frequency distribution of the phase such that any two of the acceleration, velocity, displacement, and jerk each have different desired kurtosis.

[0040] Therefore, the desired kurtosis can be achieved through feedback control.

[0041] (7) The vibration control device according to this invention is characterized by controlling the kurtosis by changing the frequency of the division.

[0042] Therefore, kurtosis can be controlled by changing the frequency used for classification.

[0043] (8) The vibration control device according to this invention is characterized by feedback control of the frequency of the divisions so that any two of the acceleration, velocity, displacement, and jerk each have different desired kurtosis.

[0044] Therefore, the desired kurtosis can be achieved through feedback control.

[0045] (9) The vibration control method according to this invention is characterized by: performing a Fourier transform on the response waveform from a vibration sensor that measures the vibration of a test specimen being vibrated by a vibration generator that operates based on a drive waveform to calculate a response PSD; comparing the response PSD with a target PSD to calculate a control PSD so that the response PSD is equal to the target PSD; performing an inverse Fourier transform on each frequency component based on the control PSD to calculate a control waveform; outputting a drive waveform to the vibration generator so that the test specimen vibrates with the control waveform; dividing the frequency domain into at least two; and not making the frequency distribution of the phases given to each frequency component in the control waveform calculation means in the range of 0 to 2π the same in the divided frequency domains, thereby controlling at least two kurtosis values ​​of the temporal changes in the jerk, acceleration, velocity, and displacement of the control waveform different.

[0046] Therefore, at least two kurtosis values ​​of the temporal changes in jerk, acceleration, velocity, and displacement can be controlled differently.

[0047] In this invention, the "response PSD calculation means" corresponds to step S2 in the embodiment.

[0048] In this embodiment, step S3 corresponds to the "control PSD calculation means".

[0049] In this embodiment, step S12 corresponds to the "control waveform calculation means".

[0050] In this embodiment, step S17 corresponds to the "drive waveform calculation means".

[0051] In this embodiment, steps S6 and S10 correspond to the "phase control means".

[0052] The term "device" is a concept that includes not only devices composed of a single computer, but also devices composed of multiple computers connected via a network or the like. Therefore, if the means (or even a part of the means) of the present invention are distributed across multiple computers, these multiple computers constitute the device.

[0053] The term "program" is a concept that includes not only programs that can be directly executed by the CPU, but also source code programs, compressed programs, encrypted programs, and programs that work in conjunction with the operating system to perform their functions. [Brief explanation of the drawing]

[0054] [Figure 1] This is a functional configuration diagram of a vibration control device according to one embodiment of the present invention. [Figure 2] This figure shows the relationship between acceleration PSD, velocity PSD, and displacement PSD. [Figure 3] This is the hardware configuration of the vibration control device. [Figure 4] This is a flowchart of the vibration control program 98. [Figure 5] This is a flowchart of the vibration control program 98. [Figure 6] This diagram shows the relationship between the target PSD, response PSD, and control PSD. [Figure 7] This diagram shows the boundary frequency fd, and the low-frequency and high-frequency regions. [Figure 8] This figure shows the process for generating drive waveforms. [Figure 9] This is an example of the control system. [Figure 10] This is an example of the control system. [Figure 11] This figure schematically shows the probability density distribution of the phase. [Figure 12] This diagram shows the configuration of a conventional vibration control device. [Figure 13] This figure shows the probability density of the amplitude of vibrations. [Figure 14] This diagram shows the configuration of a conventional vibration control device. [Modes for carrying out the invention]

[0055] 1.Functional configuration Figure 1 shows the functional configuration of a vibration testing apparatus according to one embodiment of the present invention. The vibration generator 2 receives a D / A converted drive signal and vibrates the test specimen 4. The acceleration sensor 6, which is a vibration sensor, detects the vibration of the test specimen and outputs it as a response signal. The A / D converter 10 converts the response signal to digital. The response PSD calculation means 12 performs a Fourier transform (FFT) on the response signal and calculates the response PSD (power spectral density) as its frequency characteristic.

[0056] The control PSD calculation means 14 determines whether the response PSD matches the target PSD and modifies the control PSD so that the two match. In other words, it performs feedback control.

[0057] The phase control means 18 divides the frequency domain into multiple parts and generates a phase controlled so that the frequency distribution of phases from 0 to 2π is different between the divided frequency domains. The control waveform calculation means 16 provides this phase to the control PSD and calculates the control waveform.

[0058] The following relationships exist between jerk PSD, acceleration PSD, velocity PSD, and displacement PSD.

[0059] Acceleration PSD = Jerk PSD / (2πf) 2 Speed ​​PSD=Acceleration PSD / (2πf) 2 Displacement PSD = Acceleration PSD / (2πf) 4 Therefore, as shown in Figure 2, the frequency response of a velocity PSD corresponding to an acceleration PSD with a flat frequency response is not flat. As the frequency increases, the amplitude decreases. This trend is even more pronounced in displacement PSDs. Similarly, the frequency response of an acceleration PSD corresponding to a jerk PSD with a flat frequency response is not flat. As the frequency increases, the amplitude decreases. This trend is even more pronounced in velocity PSDs, and even more pronounced in displacement PSDs.

[0060] On the other hand, when generating a waveform by assigning phase to each component of a PSD (i.e., when performing an inverse Fourier transform), if the phase of each component is assigned using uniformly distributed random numbers between 0 and 2π, the kurtosis of the generated waveform will be 3 (i.e., normally distributed). If the phase of each component is assigned in such a way that the frequency distribution between 0 and 2π is biased, the kurtosis of the generated waveform will increase.

[0061] Changing the phase frequency distribution in the high-frequency range has a significant effect on the kurtosis of acceleration waveforms, but less effect on the kurtosis of velocity waveforms and displacement waveforms. Similarly, changing the phase frequency distribution in the low-frequency range has a significant effect on the kurtosis of velocity waveforms and displacement waveforms, but less effect on the kurtosis of acceleration waveforms.

[0062] Therefore, by dividing the frequency domain into multiple segments and using phases controlled to have different frequency distributions between 0 and 2π in each of these segmented frequency domains, it is possible to create different kurtosis values ​​for waveforms in the jerk, acceleration, velocity, and displacement domains.

[0063] The control waveform generated by the control waveform calculation means 16 is provided to the drive waveform calculation means 17. The drive waveform calculation means 17 calculates a drive waveform that excites the vibration generator 2 based on the control waveform, taking into account the transfer function of the system, so that the test specimen 4 vibrates according to the control waveform.

[0064] The D / A converter 26 converts the drive waveform of the digital data into an analog signal. This drive signal is supplied to the vibration generator 2 via the amplifier 28.

[0065] In this embodiment, phases with different distributions across frequency domains are generated and provided to the control PSD to calculate the control waveform. Therefore, the kurtosis of at least two of the jerk, acceleration, displacement, and velocity can be made different.

[0066] 2. Hardware Configuration Figure 3 shows the hardware configuration of the vibration control device. The vibration generator 2 has a vibration table (not shown) on which to place and fix the test specimen 4. The vibration generator 2 vibrates this vibration table. An acceleration sensor 6 is also provided on the test specimen 4 to detect this vibration.

[0067] The CPU 90 (which may also use a DSP) is connected to memory 92, a touchscreen display 94, non-volatile memory 96, a D / A converter 26, and an A / D converter 10. The output to the vibration generator 2 is provided to the vibration generator 2 as an analog signal via the D / A converter 26 and amplifier 28. The input from the acceleration sensor 6 is captured as digital data via the A / D converter 10.

[0068] The non-volatile memory 96 stores the operating system 97 and the control program 98. The control program 98 works in cooperation with the operating system 97 to perform its functions.

[0069] 3. Vibration control processing Figures 4 and 5 show the flowchart of the control program 98. Below, we will explain as an example the case in which the test specimen 4 is vibrated to have a target PSD, while controlling it so that the kurtosis of the acceleration waveform is 6 and the kurtosis of the displacement waveform is 3.

[0070] As mentioned above, even if vibrations with a desired PSD (target PSD) are applied to the vibration generator 2, the vibrations of the test specimen 4 will not necessarily have the target PSD. Therefore, vibrations with a control PSD are applied to the vibration generator 2 so that the test specimen 4 vibrates with the target PSD, and the control PSD is feedback-controlled so that the PSD of the actual vibration (response PSD) becomes equal to the target PSD.

[0071] The CPU 90 acquires the response acceleration waveform from the acceleration sensor 6 attached to the vibration table on which the test specimen 4 is fixed (step S1). The CPU 90 performs a Fourier transform (FFT) on this response acceleration waveform to calculate the response acceleration PSD (step S2). An example of the calculated response acceleration PSD is shown in Figure 6B. In this embodiment, the response acceleration PSD was calculated for the response acceleration waveform of one frame, but the response acceleration PSD may be calculated for the response acceleration waveform of a predetermined number of past frames.

[0072] Next, the CPU 90 compares the response acceleration PSD and the target acceleration PSD and modifies the control PSD so that they match (step S3). For example, suppose the control acceleration PSD when the above response acceleration PSD was obtained was as shown in Figure 6C. That is, suppose that when the vibration generator 2 is operated with vibrations generated based on this control acceleration PSD, the response acceleration PSD shown in Figure 6B is obtained.

[0073] The response acceleration PSD shown in Figure 6B does not match the target acceleration PSD in some parts. The CPU 90 compares the magnitude of each frequency component (called a line). For each frequency component, if the response acceleration PSD is lower than the target acceleration PSD, the control acceleration PSD is increased; if the response acceleration PSD is higher than the target acceleration PSD, the control acceleration PSD is decreased. The CPU 90 makes these corrections and calculates a new control acceleration PSD as shown in Figure 6D.

[0074] In this implementation, the amplitude components of the acceleration spectrum are obtained from the calculated control acceleration PSD, and an inverse Fourier transform (inverse FFT) is performed by assigning an appropriate random phase to each component to obtain the control waveform.

[0075] In this case, the random phase used is preferably such that not only are the phases φ1 to φq assigned to frequency components A1 to Aq when generating the control waveform for one frame random, but when focusing on a specific frequency component Ak, the phases φk(t), φk(t+1), etc., when viewed in time series are also random.

[0076] In this embodiment, the frequency distribution of the time series phases φk(t), φk(t+1)... assigned to each frequency component is controlled not only by using uniform random numbers, but also by using random numbers that follow a normal distribution with a predetermined standard deviation σ. In the case of uniform random numbers, the frequency distribution of the phases is unbiased, while in the case of a normal distribution, the frequency distribution of the phases becomes biased. In the case of a normal distribution, the smaller the standard deviation σ, the more biased the frequency distribution of the phases becomes.

[0077] The CPU90 controls the bias of the above-mentioned phase frequency distribution so that the kurtosis of the response acceleration waveform matches the target kurtosis (as shown above, in this case "6").

[0078] The CPU90 first calculates the kurtosis of the response acceleration waveform (step S4). It then determines whether the calculated kurtosis matches the kurtosis of the target acceleration waveform (as mentioned above, "6" in this case) (step S5). If they do not match, it corrects the bias in the phase frequency distribution in the high-frequency region (standard deviation of the normal distribution) (step S6).

[0079] In this embodiment, as shown in Figure 7, the boundary frequency fd is set to 10 Hz, dividing the system into a low-frequency region and a high-frequency region. Therefore, if the kurtosis of the response acceleration waveform does not match the target kurtosis, the phase frequency distribution in the region above 10 Hz (high-frequency region) is adjusted. If a control acceleration waveform is generated from the control acceleration PSD based on a phase with such a frequency distribution, the kurtosis of the control acceleration waveform changes, and the kurtosis of the drive waveform also changes, which results in a change in the kurtosis of the response acceleration.

[0080] In this mechanism, the terms "high frequency" and "low frequency" are relative and do not refer to high frequency and low frequency in the absolute sense as they are commonly used.

[0081] For example, if the kurtosis of the calculated response acceleration waveform is "4", it is smaller than the target kurtosis of "6". Therefore, by increasing the bias in the phase frequency distribution (i.e., by taking the frequency of phases between 0 and 2π as a normal distribution and decreasing the standard deviation of that normal distribution), the kurtosis of the control acceleration waveform (and drive waveform) generated based on that phase is increased. Since the kurtosis of the response acceleration waveform depends on the kurtosis of the control acceleration waveform (and drive waveform), the kurtosis of the response acceleration waveform can be increased by decreasing the above standard deviation (increasing the bias in the phase frequency distribution).

[0082] As mentioned above, the kurtosis of the response acceleration waveform is susceptible to influences in the high-frequency range, so the standard deviation in the high-frequency range is reduced by a predetermined value (increasing the bias in the phase frequency distribution).

[0083] Next, the CPU 90 integrates the response acceleration waveform to calculate the response velocity waveform, and then integrates this again to calculate the response displacement waveform (step S7). The CPU 90 calculates the kurtosis of the calculated response displacement waveform (step S8). It determines whether the calculated kurtosis matches the kurtosis of the target displacement waveform (as mentioned above, "3" in this case) (whether the difference is within an acceptable range) (step S9). If they do not match, it corrects the standard deviation of the normal distribution of the phase in the low-frequency region (step S10).

[0084] For example, if the kurtosis of the calculated response displacement waveform is "4", it is greater than the target kurtosis of "3". Therefore, by reducing the bias in the phase frequency distribution (i.e., by assuming a normal distribution for the occurrence frequency of phases between 0 and 2π, and increasing the standard deviation of that normal distribution), the kurtosis of the response displacement waveform can be reduced.

[0085] As mentioned above, the kurtosis of the response displacement waveform is susceptible to influence in the low-frequency region, so the standard deviation in the low-frequency region is increased by a predetermined value.

[0086] Next, the CPU 90 applies a random phase to the amplitude components A1 to Aq of each frequency component (each line) of the control acceleration PSD in Figure 6D and performs an inverse Fourier transform (inverse FFT) to obtain a control waveform for one frame (step S12). The random phase used here is generated to follow a normal distribution with the standard deviation determined above for the low-frequency and high-frequency regions. This random phase following the standard deviation is such that (a) the phases φ1 to φq applied to the amplitude components A1 to Aq of each frequency when generating the control acceleration waveform for one frame have randomness following the standard deviation, or (b) when focusing on the amplitude component Ak of ​​a specific frequency, the phases φk(t), φk(t+1), etc., when viewed in time series, also have randomness following the standard deviation. It is necessary to satisfy either (a) or (b), but it is preferable to satisfy both if possible.

[0087] In the same manner as described above, control acceleration waveforms for multiple frames can be calculated. However, these generated control acceleration waveforms for multiple frames are discontinuous at their ends (the last amplitude of the waveform in one frame does not match the first amplitude of the waveform in the next frame), and if used as is, unnecessary frequency components will be generated. Therefore, as shown below, the control acceleration waveform of each frame is multiplied by a window function that has zero at both ends, so that the waveform smoothly converges to zero at the beginning and end of the frame, and no extraneous frequency components are introduced at the connection points.

[0088] Once the CPU 90 obtains the control acceleration waveform as described above, it multiplies the control waveform for one frame by a window function (step S13). For example, as shown in Figure 8A, a function is used that is "0" at the start and end of one frame and takes its maximum value at the midpoint. Preferably, this function is used such that when these functions are shifted by a certain amount and superimposed, the sum of the values ​​becomes "1" at all points in time.

[0089] The properties that the window function used in this process should possess are described in Japanese Patent Publication No. 6-5192. Furthermore, the process involves overlapping the wave packet-like waveform data generated by multiplying by the window function, shifting it by 1 / M of the frame width. The value of M must satisfy certain conditions determined by the characteristics of the window function used. Thus, there is a certain degree of freedom in the selection of the window function and the numerical value M. The Hanning window is usually used, and in that case, the minimum possible value of M is 4. This specification also provides an example of the case where M=4.

[0090] The problem of discontinuity described above can be solved by continuously shifting and superimposing control acceleration waveforms multiplied by a window function. The control acceleration waveform generated for one frame is a pseudo-random waveform with a discrete spectrum. By multiplying it by a window function and superimposing it, a truly irregular waveform (true random waveform) without periodicity can be obtained.

[0091] The CPU 90 then superimposes the control acceleration waveforms multiplied by the window function, shifting them by 1 / 4 frames (step S14). Therefore, by repeating the processes in steps S1 to S14, waveforms shifted by 1 / 4 frames are superimposed, as shown in Figures 8B to 8E, to obtain a continuous control acceleration waveform as shown in Figure 8F.

[0092] Next, the CPU 90 performs control to vibrate the test specimen 4 according to the continuous control acceleration waveform. However, if the subsequent drive signal generation process is carried out frame by frame, discontinuities may occur at the frame transitions.

[0093] Therefore, the waveform data extracted by shifting the extraction start point by 1 / 2 frame each time is multiplied by a window function, and the impulse response (inverse function of the transfer function) as a control characteristic is convolved into the resulting waveform to generate an acceleration drive waveform. These waveforms are then superimposed and connected (overlap processing) while shifting them again by 1 / 2 frame each time (steps S15-S18). This process will be explained below.

[0094] The CPU 90 extracts a control acceleration waveform for one frame from the continuous control acceleration waveform generated in step S14 (step S15). Next, it multiplies the extracted control acceleration waveform by a window function (step S16).

[0095] Next, the drive waveform due to acceleration is generated by convolving the impulse response, which is the control characteristic, onto the control acceleration waveform for one frame multiplied by the window function (step S17). In this embodiment, the inverse characteristic of the transfer function of the system including the vibration generator 2 and the test specimen 4 is used as the control characteristic. In other words, in order to vibrate the test specimen 4 with the control acceleration waveform, this can be achieved by providing a waveform obtained by convolving the inverse characteristic of the transfer function onto the control acceleration waveform as the drive waveform. Alternatively, the impulse response corresponding to the inverse characteristic of the transfer function may be used as the control characteristic.

[0096] The CPU 90 performs an overlap process, where the signals are superimposed with a 1 / 2 frame shift by the aforementioned window function, and then concatenates the drive signals obtained in this way (step S18). A continuous drive waveform is obtained in this manner and output to the amplifier 28 via the D / A converter 26 (step S19).

[0097] Therefore, the drive signal amplified by the amplifier 28 is supplied to the vibration generator 2, causing the test specimen 4 to vibrate.

[0098] Next, the CPU 90 acquires the response acceleration waveform from the acceleration sensor 6 (step S20). Based on the given drive waveform and the corresponding response acceleration waveform, the transfer function of the system is calculated (step S21). That is, the response acceleration waveform is subjected to an FFT to calculate the response spectrum (including phase information), and the drive waveform is subjected to an FFT to calculate the drive spectrum (including phase information). From both, the transfer function is calculated as the ratio of the response spectrum to the drive spectrum.

[0099] Next, the reciprocal of the calculated transfer function is updated as the control characteristic (step S22). This control characteristic will be used when generating the next drive signal.

[0100] Thereafter, the CPU 90 repeatedly executes steps S2 and below. In steps S6 and S10, the standard deviation of the phase frequency distribution is modified to approach the target kurtosis. This feedback control allows the kurtosis of the vibration acceleration waveform and displacement waveform applied to the test specimen 4 to be different target kurtosis values.

[0101] 4. Variations and Others (1) In the above embodiment, no trial excitation is performed, and the desired kurtosis is achieved by feedback control during the main excitation. However, the standard deviation of the frequency distribution of the phases that achieve the desired kurtosis may be calculated during the trial excitation, and the main excitation may be performed based on the calculated standard deviation.

[0102] (2) In the above embodiment, vibrations that match the desired PSD are applied to the specimen 4, while the kurtosis of the acceleration waveform and the kurtosis of the displacement waveform are controlled to be different desired values. However, the same control may be performed on any two of the jerk waveform, acceleration waveform, velocity waveform, and displacement waveform.

[0103] Furthermore, similar control may be applied to three or more of the waveforms described in item 4 above.

[0104] (3) In the above embodiment, the boundary frequency fd shown in Figure 7 is fixed when performing feedback control. However, by raising (lowering) the boundary frequency fd, the kurtosis of the displacement waveform can be increased (decreased) even if the standard deviation of the phase frequency distribution in the high-frequency region remains the same. Therefore, this boundary frequency fd may also be used in feedback control.

[0105] Alternatively, instead of changing the boundary frequency fd from the beginning, the standard deviation of the phase frequency distribution may be controlled initially, and the boundary frequency fd may be controlled once the resulting change in kurtosis becomes smaller than a predetermined value (when the control limit is reached).

[0106] (4) In the above embodiment, the acceleration sensor 6 is used to calculate the displacement by double integrating the acceleration. However, the displacement may also be obtained using a displacement sensor. That is, jerk, acceleration, velocity, and displacement may each be obtained by providing sensors, or they may be obtained by calculating them from other values ​​detected by sensors.

[0107] (5) In the above embodiment, the standard deviation of the phase frequency distribution is corrected by feedback control to obtain the desired kurtosis. However, the standard deviation of the phase frequency distribution that has been calculated in advance may be used to obtain the desired kurtosis.

[0108] (6) In the above embodiment, the frequency domain is divided into two, but it may be divided into three or more to control the phase frequency distribution. In this case, the phase frequency distribution in the lowest frequency domain greatly affects the displacement kurtosis, the phase frequency distribution in the intermediate frequency domain greatly affects the velocity, and the phase frequency distribution in the highest frequency domain greatly affects the acceleration and jerk kurtosis.

[0109] (7) In the above embodiment, in order to control the kurtosis, random phases are generated using a normal distribution, and the standard deviation of that normal distribution is changed. That is, by controlling the standard deviation of the normal distribution, the degree of bias in the frequency distribution of phases between 0 and 2π is controlled. However, instead of a normal distribution, a Cauchy distribution or other characteristics may be used for control.

[0110] Furthermore, control may be performed based on the shape of the distribution (which does not have to be a normal distribution), rather than just the standard deviation. For example, as shown in Figure 11, an arbitrary probability density distribution (a triangle in the figure, but any shape is acceptable) may be set, and the degree of bias in the phase frequency distribution may be controlled by its height H, width W, or both. In this case, a larger height H results in a larger bias, and a smaller height H results in a smaller bias. Similarly, a larger width W results in a smaller bias, and a smaller width W results in a larger bias.

[0111] Alternatively, control may be performed based on a width W1 at which a predetermined proportion (for example, 80%) is included. The larger the width W1, the smaller the bias, and the smaller the width W1, the larger the bias.

[0112] (8) The above embodiments and their modifications can be implemented in combination with each other. [Examples]

[0113] Figure 9 shows data when the standard deviation σ of the normal distribution of phase in the high-frequency and low-frequency regions, and the boundary frequency fd, are controlled by feedback. In this example, the kurtosis of the acceleration waveform was controlled to be "6" and the kurtosis of the displacement to be "3". The initial boundary frequency fd was set to 70 Hz.

[0114] Figure 9A shows the changes in kurtosis of the acceleration and displacement waveforms associated with the control. Figure 9B shows the changes in the standard deviation σdisp of the normal phase distribution in the high-frequency region, the standard deviation σacc of the normal phase distribution in the low-frequency region, and the boundary frequency fd. As shown in Figure 9A, it can be seen that the target kurtosis has been reached in general.

[0115] Figure 10 shows the result of controlling the displacement kurtosis to 3 or less by controlling the acceleration waveform kurtosis to "6" and the velocity kurtosis to "3". This control utilizes the property that displacement kurtosis is smaller than velocity kurtosis, by controlling the velocity kurtosis to 3 to reduce the displacement kurtosis to 3 or less. The initial boundary frequency fd is set to 10 Hz.

Claims

1. A response PSD calculation means calculates the response PSD by performing a Fourier transform on the response waveform from a vibration sensor that measures the vibration of a test specimen being vibrated by a vibration generator that operates based on a drive waveform, A control PSD calculation means that compares the response PSD with the target PSD and calculates a control PSD such that the response PSD becomes equal to the target PSD, A control waveform calculation means that, based on a control PSD, assigns phase to each frequency component, performs an inverse Fourier transform, and outputs a control waveform, A drive waveform calculation means calculates a drive waveform to be applied to a vibration generator so that the test specimen vibrates with the aforementioned control waveform, A phase control means that divides the frequency domain into at least two, and controls at least two kurtosis values ​​of the temporal changes in the jerk, acceleration, velocity, and displacement of the control waveform by making the frequency distribution of the phase applied to each frequency component in the control waveform calculation means from 0 to 2π different in the divided frequency domains, A vibration control device equipped with [a specific feature].

2. A vibration control program for implementing a vibration control device using a computer, wherein the computer A response PSD calculation means calculates the response PSD by performing a Fourier transform on the response waveform from a vibration sensor that measures the vibration of a test specimen being vibrated by a vibration generator that operates based on a drive waveform, A control PSD calculation means that compares the response PSD with the target PSD and calculates a control PSD such that the response PSD becomes equal to the target PSD, A control waveform calculation means calculates a control waveform by assigning phase to each frequency component and performing an inverse Fourier transform based on a control PSD, A drive waveform calculation means that outputs a drive waveform to be supplied to a vibration generator so that the test specimen vibrates with the aforementioned control waveform, A vibration control program that functions as a phase control means for controlling at least two different kurtosis values ​​of the temporal changes in the jerk, acceleration, velocity, and displacement of the control waveform, by dividing the frequency domain into at least two and making the frequency distribution of the phase given to each frequency component in the control waveform calculation means from 0 to 2π different in the divided frequency domains.

3. In the vibration control device according to claim 1 or the vibration control program according to claim 2, The vibration control device or vibration control program is characterized in that the phase control means increases the bias in the phase frequency distribution in the low-frequency region compared to the high-frequency region, thereby reducing the kurtosis of jerk or acceleration compared to either velocity or displacement.

4. In the vibration control device or vibration control program of claim 3, A vibration control device or vibration control program characterized by feedback control of the bias in the frequency distribution of the phase such that any two of the acceleration, velocity, displacement, and jerk each have different desired kurtosis.

5. In the vibration control device according to claim 1 or the vibration control program according to claim 2, The vibration control device or vibration control program is characterized in that the phase control means increases the bias of the phase frequency distribution in the high-frequency region compared to the low-frequency region, thereby increasing the kurtosis of jerk or acceleration compared to either velocity or displacement.

6. In the vibration control device or vibration control program of claim 5, A vibration control device or vibration control program characterized by feedback control of the bias in the frequency distribution of the phase such that any two of the acceleration, velocity, displacement, and jerk each have different desired kurtosis.

7. In the vibration control device according to claim 1 or the vibration control program according to claim 2, A vibration control device or vibration control program characterized by controlling the kurtosis by changing the frequency of the classification.

8. In the vibration control device or vibration control program of claim 7, A vibration control device or vibration control program characterized by feedback control of the frequency of the segmentation so that any two of the acceleration, velocity, displacement, and jerk each have different desired kurtosis.

9. The response waveform from a vibration sensor, which measures the vibration of a test specimen being vibrated by a vibration generator operating based on the drive waveform, is Fourier transformed to calculate the response PSD. The response PSD is compared with the target PSD, and the control PSD is calculated so that the response PSD becomes equal to the target PSD. Based on the control PSD, the control waveform is calculated by assigning phase to each frequency component, performing an inverse Fourier transform, and then... The drive waveform is output to the vibration generator so that the test specimen vibrates with the aforementioned control waveform. A vibration control method characterized by dividing the frequency domain into at least two, and by not making the frequency distribution of the phase applied to each frequency component in the control waveform calculation means in the range of 0 to 2π the same in the divided frequency domain, thereby controlling at least two kurtosis values ​​of the temporal changes in the jerk, acceleration, velocity, and displacement of the control waveform different.

Images