Dominant frequency extraction device, dominant frequency extraction method, dominant frequency extraction program

Abstract

To provide a dominant frequency extraction device, a method for extracting a dominant frequency, and a dominant frequency extraction program that can precisely extract a dominant frequency of a signal including a component input regularly.SOLUTION: The dominant frequency extraction device performs: calculation processing of calculating a frequency spectrum by performing Fourier transform on a signal including a component input regularly; and extraction processing of extracting a dominant frequency from the frequency spectrum by a template matching technique.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present invention relates to a dominant frequency extraction device, a dominant frequency extraction method, and a dominant frequency extraction program. [Background technology]

[0002] The International Roughness Index is known as an index for evaluating road surface flatness. Patent Document 1 discloses an example of a road surface flatness measuring device that calculates the International Roughness Index. This road surface flatness measuring device is equipped with a processing means that is attached to, for example, the lower part of the axle or suspension of a test vehicle and calculates the International Roughness Index based on the detection results of a sensor that detects acceleration in the vertical direction perpendicular to the axle direction. The processing means calculates a road surface profile by back-calculating a quarter-car simulation using an equation of motion based on the acceleration detected by the sensor, and then uses the calculated road surface profile to perform a quarter-car simulation on a reference vehicle, thereby calculating the International Roughness Index. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 5226437 Summary of the Invention [Problem to be solved by the invention]

[0004] In calculating the road surface profile, if the center of gravity of the tire-wheel combination of the test vehicle is not located at the center of rotation, vibrations due to centrifugal force occur with each rotation period of the wheel. This vibration is superimposed on the unsprung acceleration, and is therefore an unnecessary component for measuring the road surface profile, so it is preferable to remove it. For this reason, for example, it is conceivable to extract the frequency of the dominant component (hereinafter referred to as the "dominant frequency") from the frequency spectrum obtained by Fourier transforming the unsprung acceleration at regular intervals, and then remove the components existing near the dominant frequency using a band-stop filter or the like.

[0005] In this case, for example, if the frequency of a peak that simply has a large value in the frequency spectrum is extracted, there is a risk of extracting a non-stationary, idiosyncratic component. Furthermore, for example, if the frequency of a peak point of a component that has a large value in a smoothed frequency spectrum is extracted, there is a risk of extracting a frequency that is shifted from the original dominant frequency due to the smoothing. Therefore, a method is needed that can accurately extract the dominant frequency of a signal that includes components input at regular intervals, such as unsprung acceleration. Note that this problem also occurs when extracting the dominant frequency of a signal that includes components input at regular intervals, such as pressure, voltage, and displacement.

[0006] An object of the present invention is to provide a dominant frequency extraction device, a dominant frequency extraction method, and a dominant frequency extraction program that can accurately extract the dominant frequency of a signal that includes components input at regular intervals. [Means for solving the problem]

[0007] A dominant frequency extraction device according to a first aspect of the present invention performs a calculation process of calculating a frequency spectrum by Fourier transforming a signal including components input at regular intervals, and an extraction process of extracting a dominant frequency from the frequency spectrum by template matching.

[0008] According to the above dominant frequency extraction device, the template matching method is used, so that the dominant frequency can be extracted from the frequency spectrum with high accuracy.

[0009] A dominant frequency extraction device according to a second aspect of the present invention is the dominant frequency extraction device according to the first aspect, wherein the extraction process calculates a matching value based on values ​​at multiple points of a template that imitates a peak waveform of the frequency spectrum and spectral intensities at multiple points of the frequency spectrum that correspond to the multiple points of the template, and extracts the dominant frequency based on the matching value.

[0010] According to the above dominant frequency extraction device, the precision of the extraction of dominant frequencies can be improved.

[0011] In a dominant frequency extraction device according to a third aspect of the present invention, the frequency spectrum includes a fundamental wave and harmonics or subharmonics whose frequencies can be predicted based on the frequency of the fundamental wave, and the extraction process extracts the dominant frequency based on the matching value between the template and the fundamental wave and the matching value between the template and the harmonics or subharmonics.

[0012] According to the dominant frequency extraction device, the matching value is calculated taking into account harmonics or subharmonics in the extraction process, so that the precision of the dominant frequency extraction is further improved.

[0013] A method for extracting a dominant frequency according to a fourth aspect of the present invention includes a calculation step of calculating a frequency spectrum by Fourier transforming a signal including components input at regular intervals, and an extraction step of extracting a dominant frequency from the frequency spectrum by template matching.

[0014] According to the method for extracting a dominant frequency, the same effects as those of the device for extracting a dominant frequency according to the first aspect can be obtained.

[0015] A dominant frequency extraction program according to a fifth aspect of the present invention causes a computer to execute a calculation step of calculating a frequency spectrum by Fourier transforming a signal including components input at regular intervals, and an extraction step of extracting a dominant frequency from the frequency spectrum by template matching.

[0016] According to the program for extracting a dominant frequency, the same effects as those of the device for extracting a dominant frequency according to the first aspect can be obtained. [Effects of the Invention]

[0017] According to the dominant frequency extraction device, dominant frequency extraction method, and dominant frequency extraction program of the present invention, it is possible to extract with high precision the dominant frequency of a signal that includes components input at regular intervals. [Brief explanation of the drawings]

[0018] [Figure 1] 1 is a schematic configuration diagram of a road surface flatness measurement system according to an embodiment; [Figure 2] 10 is a flowchart showing an example of a processing procedure for a displacement calculation process. [Figure 3] 10 is a flowchart showing an example of a processing procedure of a resampling process. [Figure 4] 10 is a graph showing an example of the relationship between data on unsprung displacement and data on the cumulative distance traveled by a test vehicle. [Figure 5] 10 is a graph showing the correspondence between unsprung displacement and an equidistant distance axis. [Figure 6] 10 is a flowchart showing an example of a processing procedure for vehicle characteristic estimation processing. [Figure 7] 10 is a graph showing an example of the relationship between data on unsprung acceleration and data on the cumulative distance traveled by a test vehicle. [Figure 8] 10 is a graph showing the correspondence relationship between unsprung acceleration and an equidistant distance axis. [Figure 9] 1 is a schematic diagram of a spatial frequency spectrum of a dominant component of a spatial frequency spectrum and its surroundings. [Figure 10]Schematic diagram of the template. [Figure 11] Schematic diagram of the template matching method used for the fundamental wave and harmonics. [Figure 12] A front view showing the suspension of the test vehicle and its surroundings. [Figure 13] 10 is a flowchart showing an example of a processing procedure for road surface profile calculation processing. [Figure 14] FIG. 1 shows the spatial frequency spectrum of the first test. [Figure 15] This is a diagram in which the trend component of the spatial frequency spectrum has been removed from Figure 14. [Figure 16] A diagram of the dominant components extracted from the spatial frequency spectrum of the first test. [Figure 17] FIG. 10 shows the spatial frequency spectrum of the second test. [Figure 18] This is a diagram in which the trend component of the spatial frequency spectrum has been removed from Figure 17. [Figure 19] A diagram of the dominant components extracted from the spatial frequency spectrum of the second test. [Figure 20] FIG. 11 shows the spatial frequency spectrum of the third test. [Figure 21] This is a graph in which the trend component of the spatial frequency spectrum has been removed from Figure 20. [Figure 22] FIG. 10 is a diagram showing that the dominant component of the spatial frequency spectrum could not be extracted in the third test. [Figure 23] FIG. 4 shows the spatial frequency spectrum of the fourth test. [Figure 24] This is a diagram in which the trend component of the spatial frequency spectrum has been removed from Figure 23. [Figure 25] A diagram of the dominant components extracted from the spatial frequency spectrum of the fourth test. [Figure 26] A diagram showing the test results of the fifth test. [Figure 27] A diagram showing the test results of the sixth test. [Figure 28] A diagram showing the test results of the 7th test. [Figure 29] A diagram showing the test results of the 8th test. [Figure 30]A diagram showing the test results of the 9th test. DETAILED DESCRIPTION OF THE INVENTION

[0019] Hereinafter, a dominant frequency extraction device according to an embodiment of the present invention and a road surface flatness measurement system including the same will be described with reference to the drawings.

[0020] <1. Road surface flatness measurement system> <1-1. Overall configuration of road surface flatness measurement system> FIG. 1 is a schematic diagram of a road surface flatness measurement system 10 (hereinafter referred to as "system 10") including a road surface flatness measurement device 60, which is an example of a dominant frequency extraction device. The system 10 is configured to be able to calculate an International Roughness Index (hereinafter referred to as IRI) with high reproducibility based on a quarter-car simulation. More specifically, the system 10 is configured to be able to calculate a road surface profile used in the quarter-car simulation with high reproducibility. The road surface profile is a two-dimensional surface shape obtained by cutting a road surface along a virtual line on the road, and is data related to the height of the road surface at every fixed distance. The main elements constituting the system 10 are a vertical direction sensor 20, a horizontal direction sensor 30, a vehicle speed pulse generator 40, an acquisition device 50, and a road surface flatness measurement device 60. Part or all of the system 10 is mounted on a test vehicle 100 that travels on the road surface to be measured. The road surface to be measured includes, for example, the road surface of a section of a paved public road that is under consideration for repair, the road surface of any section of a paved public road, or the paved road surface of privately owned land.

[0021] <1-2. Up / down direction sensor> The vertical direction sensor 20 is attached to the lower part of the axle or suspension 120 (see FIG. 12 ) of the test vehicle 100 and detects at least one of acceleration and speed in the vertical direction perpendicular to the axle direction. In this embodiment, the vertical direction sensor 20 is an acceleration sensor attached to the lower part of the suspension 120 of the left front wheel 110 (see FIG. 12 ) of the test vehicle 100 and detects acceleration in the vertical direction perpendicular to the axle direction. The vertical direction sensor 20 may be attached to the suspension of the right front wheel, the suspension of the left rear wheel, or the suspension of the right rear wheel of the test vehicle 100. The vertical direction sensor 20 may be attached to the axle of the left front wheel, the axle of the right front wheel, the axle of the left rear wheel, or the axle of the right rear wheel of the test vehicle 100. The vertical direction sensor 20 may also be a speed sensor that detects speed in the vertical direction perpendicular to the axle direction. The vertical direction sensor 20 outputs a signal related to the detected acceleration to the acquisition device 50.

[0022] <1-3. Left and right direction sensor> The left-right direction sensor 30 is attached to the lower part of the axle or suspension 120 (see FIG. 12 ) of the test vehicle 100, and detects at least one of acceleration and speed in the left-right direction perpendicular to the up-down direction. In this embodiment, the left-right direction sensor 30 is an acceleration sensor attached to the lower part of the suspension 120 of the left front wheel 110 of the test vehicle 100, and detects acceleration in the left-right direction perpendicular to the up-down direction. The left-right direction sensor 30 is disposed adjacent to the up-down direction sensor 20. The left-right direction sensor 30 may also be a speed sensor that detects speed in the left-right direction perpendicular to the up-down direction. The left-right direction sensor 30 outputs a signal related to the detected acceleration to the acquisition device 50.

[0023] <1-4. Vehicle speed pulse generator> The vehicle speed pulse generator 40 is disposed, for example, on the back of the speedometer (not shown) of the test vehicle 100, and generates a pulse signal with a frequency proportional to the rotation speed of the axle of the test vehicle 100. The vehicle speed pulse generator 40 outputs the pulse signal to the acquisition device 50 and a control device (not shown) that controls the engine of the test vehicle 100.

[0024] <1-5. Acquisition device> The acquisition device 50 acquires a signal relating to the vertical acceleration output from the vertical direction sensor 20, a signal relating to the horizontal acceleration output from the horizontal direction sensor 30, and a pulse signal output from the vehicle speed pulse generator 40, and outputs these signals to the road surface flatness measurement device 60.

[0025] <1-6. Hardware configuration of road surface flatness measurement device> The road surface flatness measuring device 60 is connected to the acquisition device 50 so as to be able to communicate with the acquisition device 50 by wire or wirelessly. The road surface flatness measuring device 60 has a main body 61, an input device 62, and an output device 63.

[0026] The main body 61 has a storage unit 61A and a control unit 61B. The storage unit 61A and the control unit 61B are connected to each other so as to be able to communicate with each other via a bus 64. The storage unit 61A is, for example, a hard disk or a solid state drive. The storage unit 61A stores one or more application software related to the calculation of a road surface profile (hereinafter referred to as "road surface profile calculation software").

[0027] The control unit 61B includes a CPU (Central Processing Unit), RAM (Random Access Memory), ROM (Read Only Memory), etc., and executes the functions described below. The control unit 61B executes road surface profile calculation software stored in the storage unit 61A in response to a request from the input device 62. As the control unit 61B executes the road surface profile calculation software, one or more functional blocks are constructed in the road surface flatness measurement device 60.

[0028] The input device 62 is, for example, a keyboard, and is connected to the main body 61. The output device 63 is, for example, a liquid crystal display, and is connected to the main body 61 so as to receive signals output from the main body 61. Note that the main body 61, the input device 62, and the output device 63 may be integrated into one unit, like a notebook personal computer.

[0029] <1-7. Software configuration of road surface flatness measurement device> The functional blocks configured by the control unit 61B executing the road surface profile calculation software stored in the memory unit 61A include, for example, a road surface profile calculation unit 70. The road surface profile calculation unit 70 executes a road surface profile calculation process based on the acceleration detected by the vertical direction sensor 20 and the horizontal direction sensor 30. The road surface profile calculation process includes low-pass filtering, displacement calculation, resampling, tire rotational component removal, lateral force component removal, and band-pass filtering. As will be described later, the road surface profile calculation unit 70 executes a vehicle characteristics estimation process to extract a dominant frequency from the spatial frequency spectrum prior to the road surface profile calculation process in order to remove the tire rotational component in the tire rotational component removal process. In the vehicle characteristics estimation process, the road surface profile calculation unit 70 extracts the dominant frequency using a matching method (hereinafter referred to as a "template matching method") that evaluates a calculated value with a template. Each process will be described in detail below.

[0030] <1-8. Low-pass filter processing> The low-pass filter process is executed before the displacement calculation process, the resampling process, the tire rotation component removal process, the lateral force component removal process, and the band-pass filter process. In the low-pass filter process, the road surface profile calculation unit 70 applies a bidirectional low-pass filter to the acceleration detected by the vertical direction sensor 20 and the acceleration detected by the horizontal direction sensor 30, thereby removing frequency components higher than the frequency band related to the vibration components of the test vehicle 100 caused by road surface irregularities. The frequency components higher than the frequency band related to the vibration components of the test vehicle 100 caused by road surface irregularities are, for example, frequency components of 300 Hz or higher.

[0031] <1-9. Displacement calculation process> The displacement calculation process is executed after the low-pass filter process and before the resampling process. In the displacement calculation process, the road surface profile calculation unit 70 calculates the vertical velocity of the suspension 120 (hereinafter referred to as the "unsprung velocity") by integrating the acceleration (hereinafter referred to as the "unsprung acceleration") detected by the vertical direction sensor 20, and calculates the vertical displacement of the suspension 120 (hereinafter referred to as the "unsprung displacement") by integrating the unsprung velocity.

[0032] The unsprung acceleration, unsprung velocity, and unsprung displacement are expressed by the following equations (1) to (3). Unsprung mass acceleration f(t)=X''(t) (1) Unsprung mass velocity ∫f(t)dt=X'(t)+C1 (2) Unsprung displacement ∫∫f(t)dtdt=X(t)+C1·t+C2···(3)

[0033] As is well known, the unsprung displacement obtained by integrating the unsprung acceleration twice contains an integral error. That is, because the integral constant C1 takes time t, the calculated road profile resembles an uphill or downhill slope over time. If a DC bias A is present in the waveform of the unsprung acceleration, the integral error in the unsprung displacement obtained by integrating the unsprung acceleration twice will be a quadratic curve such as A / 2·t2+C1·t+C2. Furthermore, if a time-varying component such as temperature drift is present instead of a DC component, an even larger integral error may occur. For this reason, in this embodiment, the road profile calculation unit 70 applies a bidirectional high-pass filter to the unsprung acceleration, unsprung velocity, and unsprung displacement during the displacement calculation process to remove low-frequency components lower than the frequency band related to the vibration components of the test vehicle 100 caused by road surface irregularities.

[0034] 2 is a flowchart showing an example of the procedure of the displacement calculation process. The displacement calculation process is started at any timing after the low-pass filter process and before the resampling process.

[0035] In step S11, the road surface profile calculation unit 70 applies a bidirectional high-pass filter to the unsprung acceleration. As a result, bias components and temperature drift are removed from the unsprung acceleration. In step S12, the road surface profile calculation unit 70 calculates the unsprung velocity by integrating the unsprung acceleration. In step S13, the road surface profile calculation unit 70 applies a bidirectional high-pass filter to the unsprung velocity. As a result, bias related to the integration constant in step S12 is removed. In step S14, the road surface profile calculation unit 70 integrates the unsprung velocity to calculate the unsprung displacement. In step S15, the road surface profile calculation unit 70 applies a bidirectional high-pass filter to the unsprung displacement. As a result, bias related to the integration constant in step S14 is removed.

[0036] <1-10. Resampling process> The resampling process is executed after the displacement calculation process. In this embodiment, in the resampling process, the road surface profile calculation unit 70 creates equal-interval distance data having equal-interval distance axes based on the cumulative travel distance of the test vehicle 100, and associates the calculated unsprung displacement with the equal-interval distance axes of the equal-interval distance data.

[0037] 3 is a flowchart showing an example of the procedure of the resampling process, which is started at any timing after the displacement calculation process.

[0038] In step S21, the road surface profile calculation unit 70 acquires a vehicle speed pulse signal and a vehicle speed pulse coefficient from the vehicle speed pulse generator 40. The vehicle speed pulse coefficient is the distance per edge of the vehicle speed pulse. The vehicle speed pulse coefficient may be stored in advance in the storage unit 61A.

[0039] In step S22, the road surface profile calculation unit 70 detects a vehicle speed pulse edge based on the vehicle speed pulse signal.

[0040] In step S23, the road surface profile calculation unit 70 calculates the cumulative travel distance of the test vehicle 100 based on the vehicle speed pulse edges and the vehicle speed pulse coefficient. In step S23, the road surface profile calculation unit 70 calculates the cumulative travel distance of the test vehicle 100 by multiplying the cumulative value of the vehicle speed pulse edges by the vehicle speed pulse coefficient. Note that simply multiplying the cumulative value of the vehicle speed pulse edges by the vehicle speed pulse coefficient may result in multiple samples of the same distance on the time axis. Therefore, the road surface profile calculation unit 70 calculates the cumulative travel distance of the test vehicle 100 using data with a single increasing trend obtained by linearly interpolating the data obtained by multiplying the cumulative value of the vehicle speed pulse edges by the vehicle speed pulse coefficient.

[0041] In step S24, the road surface profile calculation unit 70 associates the calculated unsprung displacement with the equidistant distance axis of the equidistant distance data.

[0042] 4 is an example of a graph showing the relationship between data related to unsprung displacement (hereinafter referred to as "unsprung displacement data") and data related to the cumulative travel distance of the test vehicle 100 (hereinafter referred to as "distance data"). In step S24, the road surface profile calculation unit 70 creates equal-interval distance data having equal-interval distance axes shown in FIG. 5 based on the cumulative travel distance of the test vehicle 100 calculated in step S23, and associates the calculated unsprung displacement with the equal-interval distance axes of the equal-interval distance data. In the example shown in FIG. 5, the road surface profile calculation unit 70 associates the unsprung displacement with each distance value "2" on the equal-interval distance axes.

[0043] In step S24, there may be cases where there is no unsprung displacement data corresponding to the distance data of the equally-spaced distance data having equally-spaced distance axes. In such cases, the road surface profile calculation unit 70 generates the non-existent unsprung displacement data (hereinafter referred to as "interpolated data") by linear interpolation based on the values ​​immediately before and after the non-existent unsprung displacement data. In the example shown in FIGS. 4 and 5, there is no unsprung displacement data for distance value "2," so the road surface profile calculation unit 70 generates the interpolated data for distance value "2" by linear interpolation of the unsprung displacement data for distance values ​​"1" and "3." Similarly, there is no unsprung displacement data for distance value "6," so the road surface profile calculation unit 70 generates the interpolated data for distance value "6" by linear interpolation of the unsprung displacement data for distance values ​​"5" and "7." Because no unsprung displacement data exists for the distance value "8," the road surface profile calculation unit 70 generates interpolated data for the distance value "8" by linearly interpolating the unsprung displacement data for the distance values ​​"7" and "9." Because no unsprung displacement data exists for the distance value "12," the road surface profile calculation unit 70 generates interpolated data for the distance value "12" by linearly interpolating the unsprung displacement data for the distance values ​​"11" and "13."

[0044] <1-11. Tire rotation component removal processing> The tire rotation component removal process is started at any timing after the low-pass filter process. In this embodiment, the tire rotation component removal process is started after the resampling process. In calculating the road surface profile, if the center of gravity of the tire and wheel combination of the test vehicle 100 is not located at the center of rotation, vibrations due to centrifugal force occur with each wheel rotation period. Since this vibration is superimposed on the unsprung acceleration, it is an unnecessary component for measuring the road surface profile and is preferably removed. For this reason, in the tire rotation component removal process, the road surface profile calculation unit 70 removes the dominant frequency component corresponding to the tire rotation component based on the dominant frequency extracted in the vehicle characteristic estimation process.

[0045] One possible method for removing vibration components due to centrifugal force is to use the static load radius of the tire, which is the distance from the center of the wheel to the tire's contact patch. The distance from the center of the wheel to the tire's contact patch is measured, for example, by an operator using a tape measure or the like. The tire circumference is calculated based on the static load radius of the tire, and the tire rotation frequency can be estimated by dividing the vehicle speed of the test vehicle 100 by the tire circumference. Vibration components due to centrifugal force can be removed by applying a band-stop filter with this estimated tire rotation frequency as the center frequency. However, if an operator makes a measurement error of even a few millimeters when measuring the distance from the center of the wheel to the tire's contact patch, the effectiveness of the band-stop filter may be reduced.

[0046] Another method for removing the vibration component due to centrifugal force is, for example, to extract a dominant frequency from a time frequency spectrum obtained by Fourier transforming the unsprung mass acceleration at regular intervals and then remove components near the dominant frequency using a band-stop filter. However, if the vehicle speed of the test vehicle 100 is not constant, the time frequency of the dominant component in the time frequency spectrum is also not constant, resulting in dispersion of the peak of the dominant component, making it difficult to accurately extract the dominant frequency. On the other hand, if the spatial frequency spectrum is used, the peak of the dominant component is less likely to be dispersed regardless of the vehicle speed of the test vehicle 100. In other words, the peak of the dominant component is maintained in its mountain shape, allowing for accurate extraction of the dominant frequency. For this reason, in this embodiment, the road surface profile calculation unit 70 performs a vehicle characteristic estimation process to extract a dominant frequency from the spatial frequency spectrum before the road surface profile calculation process, and then, in the tire rotation component removal process, removes the tire rotation component using the dominant frequency extracted from the spatial frequency spectrum.

[0047] For example, when extracting the frequency of a component that simply has a large value in a time frequency spectrum or a spatial frequency spectrum, there is a risk of extracting a non-stationary, idiosyncratic component. Furthermore, when extracting the frequency of a peak point of a component that has a large value in a smoothed time frequency spectrum or a spatial frequency spectrum, there is a risk of extracting a frequency that is different from the original dominant frequency due to the smoothing. To avoid this risk, in this embodiment, the road surface profile calculation unit 70 extracts the dominant frequency using a template matching method in the vehicle characteristic estimation process. The dominant frequency, which is one of the vehicle characteristics of the test vehicle 100, hardly changes unless the characteristics of the test vehicle 100 change. For this reason, the vehicle characteristic estimation process is typically performed once before the road surface profile calculation process, and data related to the extracted dominant frequency is used in the tire rotation component removal process. It is preferable to perform the vehicle characteristic estimation process each time the characteristics of the test vehicle 100 are predicted to change, for example, when the tires of the test vehicle 100 are replaced. The vehicle characteristic estimation process will be described in detail below.

[0048] <1-12. Vehicle characteristics estimation processing> FIG. 6 is a flowchart showing an example of a processing procedure for the vehicle characteristic estimation processing. The road surface profile calculation unit 70 starts the vehicle characteristic estimation process, for example, when the test vehicle 100 starts traveling. In another example, the road surface profile calculation unit 70 starts the vehicle characteristic estimation process when the test vehicle 100 finishes traveling and when a request is input from the input device 62. Note that the road surface on which the test vehicle 100 travels in the vehicle characteristic estimation process is a flat road or a reference road. Note that if the road surface to be measured is a flat road, the test vehicle 100 may travel on the road surface to be measured in the vehicle characteristic estimation process.

[0049] In step S31, the road surface profile calculation unit 70 acquires measurement data from the acquisition device 50. The measurement data includes the acceleration output from the vertical direction sensor 20, the acceleration output from the horizontal direction sensor 30, and the vehicle speed pulse output from the vehicle speed pulse generator 40. If the vehicle speed pulse coefficient is not stored in the memory unit 61A, in step S41, the road surface profile calculation unit 70 may acquire the vehicle speed pulse coefficient from the control device of the test vehicle 100.

[0050] In step S32, the road surface profile calculation unit 70 executes low-pass filtering, which is the same as the low-pass filtering described in the road surface profile calculation process.

[0051] In step S33, similar to the resampling process, the road surface profile calculation unit 70 creates equal-interval distance data having equal-interval distance axes based on the cumulative travel distance of the test vehicle 100. The road surface profile calculation unit 70 associates the unsprung acceleration with the equal-interval distance axes of the equal-interval distance data.

[0052] 7 is an example of a graph showing the relationship between data related to unsprung acceleration (hereinafter referred to as "unsprung acceleration data") and data related to the cumulative travel distance of the test vehicle 100 (hereinafter referred to as "distance data"). In step S33, the road surface profile calculation unit 70 creates equal-interval distance data having equal-interval distance axes shown in FIG. 8 based on the cumulative travel distance of the test vehicle 100, and associates the unsprung acceleration with the equal-interval distance axes of the equal-interval distance data. In the example shown in FIG. 8, the road surface profile calculation unit 70 associates the unsprung acceleration with every distance value "2" on the equal-interval distance axes.

[0053] In step S33, there may be cases where there is no unsprung acceleration data corresponding to the distance data of the equally-spaced distance data having equally-spaced distance axes. In such cases, the road surface profile calculation unit 70 generates the non-existent unsprung acceleration data (hereinafter referred to as "interpolated data") by linear interpolation based on the values ​​immediately before and after the non-existent unsprung acceleration data. In the example shown in FIGS. 7 and 8, there is no unsprung acceleration data for the distance value "2," so the road surface profile calculation unit 70 generates the interpolated data for the distance value "2" by linear interpolation of the unsprung acceleration data for the distance values ​​"1" and "3." Similarly, there is no unsprung acceleration data for the distance value "6," so the road surface profile calculation unit 70 generates the interpolated data for the distance value "6" by linear interpolation of the unsprung acceleration data for the distance values ​​"5" and "7." Because no unsprung acceleration data exists for distance value "8," the road surface profile calculation unit 70 generates interpolated data for distance value "8" by linearly interpolating the unsprung acceleration data for distance value "7" and distance value "9." Because no unsprung acceleration data exists for distance value "12," the road surface profile calculation unit 70 generates interpolated data for distance value "12" by linearly interpolating the unsprung acceleration data for distance value "11" and distance value "13." As described above, in step S33, the road surface profile calculation unit 70 creates unsprung acceleration data at equal intervals on the spatial (distance) axis rather than the time axis.

[0054] In step S34 (calculation process), the road surface profile calculation unit 70 calculates a spatial frequency spectrum by Fourier transforming the unsprung acceleration data at equal intervals on the distance axis, which was created in step S33.

[0055] In step S35 (extraction process), the road surface profile calculation unit 70 extracts a dominant frequency from the spatial frequency spectrum using a template matching method. In the extraction process, the road surface profile calculation unit 70 calculates a matching value based on the sum of values ​​obtained by multiplying values ​​at multiple points on a template 80 that mimics the peak waveform of the spatial frequency spectrum by the spectral intensities at multiple points on the spatial frequency spectrum 90 that correspond to the multiple points on the template 80, and extracts the dominant frequency based on the matching value.

[0056] FIG. 9 is a schematic diagram of a spatial frequency spectrum 90 of a dominant component of the spatial frequency spectrum and its surroundings. FIG. 10 is a schematic diagram of a template 80. The template size is determined based on the data length when unsprung acceleration data at equal intervals on the distance axis is Fourier transformed and preprocessing such as smoothing of the spectrum. In the example shown in FIG. 10, the template 80 is a size 9 template. The template 80 may be a size other than size 9. The longer the data length during the Fourier transform, the higher the frequency resolution of the spatial frequency spectrum. In other words, because the dominant component is finer, it becomes a pinpoint feature point. If a template with a pinpoint, steep waveform were used, there is a high possibility that it will match noise that is not a feature point. Therefore, a wider template, in other words, a template with a large amount of data, is required. Smoothing reduces noise components, making incorrect matching less likely, and therefore the template size can be reduced.

[0057] In the extraction process, the road surface profile calculation unit 70 calculates a matching value Mν at each position by moving the center frequency ν of the template 80 within a predetermined search range. The road surface profile calculation unit 70 extracts the frequency at which the matching value Mν at each position is the optimum value as the dominant frequency. In this embodiment, the optimum value of the matching value Mν at each position is the maximum value. The predetermined search range is a range that is set in advance as a frequency band in which tire rotation components may exist based on the size of commercially available tires. The matching value Mν is calculated by dividing the values ​​A at multiple points T1 to T9 of the template 80 by the T1 ~A T9 and the spectral intensities B at the points S1 to S9 of the spatial frequency spectrum 90 corresponding to the points T1 to T9 of the template 80. S1 ~B S9 and is calculated using the following formula (4):

[0058] Mν=A T1 ·B S1 +A T2 ·B S2 +···+A T8 ·B S8 +A T9 ·B S9 ···(4)

[0059] By the way, the dominant component of the spatial frequency spectrum is the frequency ν RThe spatial frequency spectrum may include a fundamental wave, which is a frequency that is a multiple of the fundamental wave, and harmonics, whose frequencies can be predicted based on the frequency of the fundamental wave. Harmonics, for example, have frequencies that are integer multiples of the fundamental wave. If the dominant frequency calculated from the spatial frequency spectrum deviates from the actual frequency of the fundamental wave, even if a band-stop filter is applied to the harmonics, the position (frequency) of the filter band deviates more significantly as the harmonic becomes higher, which may prevent the dominant components corresponding to the harmonics from being removed. Furthermore, there is a risk that necessary spatial frequency spectrum components will be removed. For this reason, in this embodiment, the road surface profile calculation unit 70 extracts the dominant frequency of the spatial frequency spectrum in the extraction process based on the matching value between the template 80 and the fundamental wave and the matching value between the template 80 and the harmonics.

[0060] 11 is a schematic diagram of a case where the template matching method is used for a fundamental wave and a harmonic wave. In the example shown in FIG. 11, a spatial frequency spectrum 200 is formed by a fundamental wave 200X, a harmonic wave 200X, and a harmonic wave 200X. 2 Harmonic 210, 3 Harmonic 220, 4 Harmonic 230 and 5 11, for the sake of simplicity, only the fundamental wave 200X and the harmonics 240 are shown. 2 Harmonic 210, 3 Harmonic 220, 4 Harmonic 230 and 5 The harmonic 240 is shown as a bar. 2 Harmonic 210, 3 Harmonic 220, 4 Harmonic 230 and 5 The actual shape of the harmonic 240 is the spectral waveform shown in FIG.

[0061] 11, the template 80 includes a fundamental wave template 80X, a first template 81, a second template 82, a third template 83, and a fourth template 84. Note that in FIG. 11, for the sake of simplicity, the fundamental wave template 80X, the first template 81, the second template 82, the third template 83, and the fourth template 84 are illustrated as bar shapes. The actual shapes of the fundamental wave template 80X, the first template 81, the second template 82, the third template 83, and the fourth template 84 are mountain shapes as shown in FIG.

[0062] In the extraction process, the road surface profile calculation unit 70 calculates the total matching value MT by moving the center frequency v of the fundamental wave template 80X within a predetermined search range. The total matching value MT is calculated by the following equation (5).

[0063] MT=M Rx +M R1 +M R2 +M R3 +M R4 ···(5) In addition, the value M Rx is the matching value between the fundamental wave template 80X and the fundamental wave 200X. R1 The first template 81 and the 2 This is the matching value with the harmonic 210. The value M R2 The second template 82 and the 3 This is the matching value with the harmonic 220. The value M R3 The third template 83 and 4 This is the matching value with the harmonic 230. The value M R4 The fourth template 84 and 5 This is the matching value with the harmonic 240.

[0064] In the extraction process, the road surface profile calculation unit 70 extracts the frequency at the position where the total matching value MT is the optimum value as the predominant frequency. In this embodiment, the optimum value of the total matching value MT is the maximum value.

[0065] As shown in FIG. 6, in step S36, the road surface profile calculation unit 70 records the dominant frequency extracted in step S35.

[0066] <1-13. Lateral force component removal processing> The lateral force component removal process is executed after the displacement calculation process and before or after the resampling process. In this embodiment, the road surface profile calculation unit 70 performs the lateral force component removal process to correct the acceleration detected by the up-down direction sensor 20 based on the lateral acceleration detected by the left-right direction sensor 30.

[0067] As shown in FIG. 12 , when the angle θA formed by the sensitivity axis XA of the vertical direction sensor 20 and the sensitivity axis XB of the left-right direction sensor 30 is 90° and the kingpin angle is θB, the vertical acceleration Az due to road surface input and the left-right acceleration Ay due to turning, etc. are calculated by the following equations (6) and (7) based on the acceleration az detected by the vertical direction sensor 20 and the acceleration ay detected by the left-right direction sensor 30.

[0068] Az=az·cosθB-ay·sinθB···(6) Ay=az·sinθB+ay·cosθB···(7)

[0069] Furthermore, when turning on a flat road surface, the vertical acceleration Az can be considered to be substantially 0, so the kingpin angle θB can be calculated as follows based on the acceleration az detected by the vertical direction sensor 20 and the acceleration ay detected by the horizontal direction sensor 30.

[0070] az·cosθB-ay·sinθB=0···(8) az / ay=tanθB (9) θB=tan-1(az / ay) (10)

[0071] Furthermore, when the test vehicle 100 is traveling on a paved road in an urban area, etc., the movement of the suspension 120 is mainly due to the linear motion of the spring and damper, so the kingpin angle θB hardly changes and can be regarded as a constant value. Therefore, the trigonometric functions of the above equations (6) and (7) can also be regarded as constants, and the road surface displacement D can be calculated using the following equations (11) and (12) based on the measurement value d.

[0072] Dz=dz·cosθB-dy·sinθB··(11) θB=tan−1(dz / dy) (12)

[0073] Therefore, the kingpin angle θB and the unsprung displacement with the lateral force component removed can be calculated based on the unsprung displacement calculated in the displacement calculation process and the displacement calculated by integrating twice the acceleration detected by the left-right direction sensor 30.

[0074] <1-14. Bandpass filter processing> The bandpass filter process is a process executed after the resampling process. In the bandpass filter process, the road surface profile calculation unit 70 applies a bidirectional bandpass filter to the calculated unsprung displacement to remove spatial frequency components higher and lower than the spatial frequency band related to the vibration components of the test vehicle 100 caused by road surface irregularities. The spatial frequency components higher than the spatial frequency band related to the vibration components of the test vehicle 100 caused by road surface irregularities are, for example, spatial frequency components of 10 cycles / m or higher. The spatial frequency components lower than the spatial frequency band related to the vibration components of the test vehicle 100 caused by road surface irregularities are, for example, spatial frequency components of 0.01 cycles / m or lower.

[0075] <1-15. Road surface profile calculation process> An example of a processing procedure for road surface profile calculation processing will be described with reference to FIG. The road surface profile calculation unit 70 starts the road surface profile calculation process, for example, when the test vehicle 100 starts traveling. In another example, the road surface profile calculation unit 70 starts the road surface profile calculation process when the test vehicle 100 finishes traveling and when a request is input from the input device 62.

[0076] In step S41, the road surface profile calculation unit 70 acquires measurement data from the acquisition device 50. The measurement data includes the acceleration output from the vertical direction sensor 20, the acceleration output from the horizontal direction sensor 30, and the vehicle speed pulse output from the vehicle speed pulse generator 40. If the vehicle speed pulse coefficient is not stored in the memory unit 61A, in step S41, the road surface profile calculation unit 70 may acquire the vehicle speed pulse coefficient from the control device of the test vehicle 100.

[0077] In step S42, the road surface profile calculation unit 70 performs low-pass filtering. In step S43, the road surface profile calculation unit 70 performs displacement calculation. In step S44, the road surface profile calculation unit 70 performs resampling. In step S45, the road surface profile calculation unit 70 performs tire rotation component removal. In step S45, the road surface profile calculation unit 70 removes dominant frequency components corresponding to tire rotation components by applying a band-stop filter whose center frequencies are the fundamental and harmonic frequencies obtained from the dominant frequency extracted in step S35 of the vehicle characteristic estimation process. In step S46, the road surface profile calculation unit 70 performs lateral force component removal. In step S47, the road surface profile calculation unit 70 performs band-pass filtering. When step S47 is completed, the road surface profile is calculated. The road surface profile calculation unit 70 calculates the IRI by performing a known quarter-car simulation based on the calculated road surface profile.

[0078] <2. Effects of this embodiment> The road surface flatness measuring device 60, which is an example of a dominant frequency extracting device configured as described above, can provide the following effects.

[0079] <2-1> The road surface flatness measuring device 60 uses a template matching method in the vehicle characteristic estimation process, and therefore can extract the dominant frequency from the spatial frequency spectrum with high accuracy.

[0080] <2-2> In this embodiment, the signal including the component input at regular intervals is the unsprung acceleration of the test vehicle 100 input at regular intervals, and the spectral intensity of the spatial frequency spectrum calculated by the calculation process varies depending on the vehicle speed and the measurement section. Therefore, when the extraction process is performed using a template matching method such as SSD (Sum of Squared Difference) or SAD (Sum of Absolute Difference), the difference value from the prepared template also increases or decreases depending on the vehicle speed and the measurement section, so there is a risk that the frequency of a feature point other than the target dominant component will be extracted.

[0081] According to the road surface flatness measuring device 60, the values ​​A at the plurality of points T1 to T9 of the template 80 are T1 ~A T9 and the spectral intensities B at the points S1 to S9 of the spatial frequency spectrum 90 corresponding to the points T1 to T9 of the template 80. S1 ~B S9 Since the matching value Mν is calculated based on the sum of the values ​​obtained by multiplying and , the accuracy of the calculation of the matching value Mν is improved.

[0082] <2-3> According to the road surface flatness measuring device 60, in the extraction process of the vehicle characteristic estimation process, the templates 81 to 84 and 2 Harmonic 210~ 5 Matching value M with harmonic 240 R1 ~M R4 Therefore, even if the spatial frequency spectrum of the fundamental wave 200X is not dominant, the dominant frequency is extracted.2 Harmonic 210~ 5 By taking the harmonics 240 into consideration, it is possible to suitably extract the dominant frequency. Note that the number of harmonics is an example. The number of harmonics may be one to three, or five or more.

[0083] <3. Test Results of the Road Surface Flatness Measuring Device of the Present Embodiment> 14 to 29, the results of extracting dominant frequencies from spatial frequency spectra in the first to eighth tests will be described. Note that the first to fourth tests are tests related to the vehicle characteristics estimation process, and the fifth to eighth tests are tests related to the tire rotation component removal process.

[0084] <3-1. Results of the first test> 14 to 16 are diagrams relating to the first test. FIG. 14 shows a spatial frequency spectrum calculated by Fourier transforming unsprung acceleration data spaced at equal intervals on the distance axis. FIG. 15 shows data obtained by removing the trend component of the spatial frequency spectrum relating to unsprung acceleration from FIG. 14. Note that a trend component is a component that corresponds to a wide-ranging distribution and tendency, unlike a local feature point such as a prominent component. FIG. 16 shows the extraction result of the prominent component of the spatial frequency spectrum. In the first test, the vehicle speed of the test vehicle 100 was a relatively high speed of approximately 80 km / h. In the first test, the template matching method was applied to only the fundamental wave in the extraction process.

[0085] 16, in the first test, it was confirmed that the dominant frequency of the spatial frequency spectrum could be suitably extracted in the extraction process even when the template matching method was applied to only the fundamental wave. This is thought to be because when the vehicle speed of the test vehicle 100 is high, the centrifugal force generated when the tires rotate is also large, which further emphasizes the dominant component of the fundamental wave.

[0086] <3-2. Results of the second test> 17 to 19 are diagrams relating to the second test. FIG. 17 shows a spatial frequency spectrum calculated by Fourier transforming unsprung acceleration data spaced at equal intervals on the distance axis. FIG. 18 shows data obtained by removing the trend component of the spatial frequency spectrum relating to unsprung acceleration from FIG. 17. FIG. 19 shows the extraction results of the dominant component of the spatial frequency spectrum. In the second test, the vehicle speed of the test vehicle 100 was a relatively high speed of approximately 80 km / h. In the first test, a template matching method was applied to the fundamental wave and harmonics in the extraction process.

[0087] As shown in Fig. 19, in the second test, it was confirmed that the extraction process was able to extract the dominant frequency of the spatial frequency spectrum appropriately. This is thought to be because the template matching method was applied to the fundamental wave and harmonics, and because the vehicle speed of the test vehicle 100 was high, which further emphasized the dominant component of the fundamental wave.

[0088] <3-3. Results of the third test> 20 to 22 are diagrams relating to the third test. FIG. 20 shows a spatial frequency spectrum calculated by Fourier transforming unsprung acceleration data spaced at equal intervals on the distance axis. FIG. 21 shows data obtained by removing the trend component of the spatial frequency spectrum relating to unsprung acceleration from FIG. 20. FIG. 22 shows the extraction results of the dominant component of the spatial frequency spectrum. In the third test, the vehicle speed of the test vehicle 100 was relatively low, ranging from 0 to 50 km / h. In the third test, template matching was applied to only the fundamental wave in the extraction process.

[0089] As shown in Fig. 22, in the third test, the extraction process was unable to extract the dominant frequency of the spatial frequency spectrum. This is thought to be because when the vehicle speed of the test vehicle 100 is low, the centrifugal force when the tires rotate is also small, and the dominant component of the fundamental wave is extremely small.

[0090] <3-4. Results of the 4th test> 23 to 25 are diagrams relating to the fourth test. FIG. 23 shows a spatial frequency spectrum calculated by Fourier transforming unsprung acceleration data spaced at equal intervals on the distance axis. FIG. 24 shows data obtained by removing the trend component of the spatial frequency spectrum relating to unsprung acceleration from FIG. 23. FIG. 25 shows the extraction results of the dominant component of the spatial frequency spectrum. In the fourth test, the vehicle speed of the test vehicle 100 was relatively low, ranging from 0 to 50 km / h. In the fourth test, a template matching method was applied to the fundamental wave and harmonics in the extraction process.

[0091] As shown in Fig. 25, in the fourth test, it was confirmed that the extraction process was able to suitably extract the dominant frequency of the spatial frequency spectrum, even though the vehicle speed of the test vehicle 100 was low. This is thought to be because the template matching method was applied to the fundamental wave and harmonics.

[0092] <3-5. Results of Tests 5 to 8> 26 to 29 are diagrams relating to the fifth to eighth tests. FIGS. 26 to 29 show spatial frequency spectra obtained by filtering unsprung acceleration data using the predominant frequency extracted in the extraction process (step S35) of the vehicle characteristic estimation process as the characteristic value of the test vehicle 100. In the fifth to eighth tests, the spatial frequency spectra of unsprung acceleration for different vehicle speeds of the test vehicle 100 were separated into predominant frequency components and other frequency components based on the predominant frequency extracted by the template matching method. FIG. 26 shows the results of the fifth test. In the fifth test, the vehicle speed of the test vehicle 100 was 20 km / h. FIG. 27 shows the results of the sixth test. In the sixth test, the vehicle speed of the test vehicle 100 was 40 km / h. FIG. 28 shows the results of the seventh test. In the seventh test, the vehicle speed of the test vehicle 100 was 60 km / h. FIG. 29 shows the results of the eighth test. In the eighth test, the vehicle speed of the test vehicle 100 is 80 km / h.

[0093] As shown in FIGS. 26 to 29, it was confirmed that the dominant frequency components corresponding to the tire rotation components of the spatial frequency spectrum could be suitably extracted regardless of the vehicle speed of the test vehicle 100.

[0094] <3-6. Results of the 9th test> With reference to FIG. 30, the results of extracting the dominant frequency components of the spatial frequency spectrum of the unsprung displacement in the ninth test will be described.

[0095] 30 shows a spatial frequency spectrum obtained by filtering the unsprung displacement data using the dominant frequency extracted in the extraction process (step S35) of the vehicle characteristic estimation process as the characteristic value of the test vehicle 100. In the ninth test, the spatial frequency spectrum of the unsprung displacement was separated into dominant frequency components and other frequency components based on the dominant frequency extracted by the template matching method.

[0096] As shown in FIG. 30, it was confirmed that the dominant frequency component corresponding to the tire rotation component of the spatial frequency spectrum was suitably extracted even for the unsprung displacement, which has the same dimensions as the road surface profile.

[0097] <4. Modifications> The above-described embodiments are merely examples of possible forms of the dominant frequency extraction device, dominant frequency extraction method, and dominant frequency extraction program of the present invention, and are not intended to limit the forms. The dominant frequency extraction device, dominant frequency extraction method, and dominant frequency extraction program of the present invention may take forms different from those exemplified in the embodiments. Examples of such forms include forms in which part of the configuration of the embodiments is replaced, modified, or omitted, or forms in which new configuration is added to the embodiments. Some examples of modified embodiments are shown below.

[0098] <4-1> In the above embodiment, the dominant frequency extraction device extracts dominant components from the spatial frequency spectrum of the unsprung acceleration of the test vehicle 100 using template matching. However, the frequency spectrum from which the dominant components are extracted is not limited to this, as long as it is a frequency spectrum of components input at regular intervals. For example, the dominant frequency extraction device may extract dominant components from the spatial frequency spectrum relating to displacement or acceleration other than the unsprung acceleration of the test vehicle 100. Examples of displacement include displacement of the road surface shape, displacement of pavement joints, displacement of railway track joints, and vibration of the rotation axis. By removing dominant components from these spatial frequency spectra, flat sections can be calculated. Examples of acceleration other than the unsprung acceleration of the test vehicle 100 include vibrations caused by the track shoes of a caterpillar or the steps of a stepped or flat escalator. By removing dominant components from these spatial frequency spectra, steady-state components can be calculated. Additionally, the technique of extracting dominant components from the spatial frequency spectrum using template matching can also be applied to, for example, torsional resonance, loads associated with walking such as ascending and descending stairs, and visual stimuli.

[0099] In another example, the dominant frequency extraction device may use a template matching method to extract dominant components of a time-frequency spectrum related to pressure, acceleration other than the unsprung acceleration of the test vehicle 100, or voltage. The pressure may be, for example, pulsation. By removing dominant components from the time-frequency spectrum related to pressure, static pressure can be calculated. The acceleration other than the unsprung acceleration of the test vehicle 100 may be, for example, natural vibration. By removing dominant components from the time-frequency spectrum related to natural vibration, phenomenal acceleration can be calculated. The voltage may be, for example, steady noise and oscillation. By removing dominant components from the time-frequency spectrum related to steady noise and oscillation, phenomenal signals can be calculated. Alternatively, the dominant frequency extraction device may extract dominant components of a time-frequency spectrum related to voltage or radio waves from a spectrum analyzer and sound waves from a tuner.

[0100] <4-2> In the above embodiment, the dominant frequency extraction device extracts the dominant frequencies of the fundamental and harmonics using template matching. However, if the relationship between the multiple frequencies to be extracted can be predicted in advance, the dominant frequency can also be extracted using template matching. If the frequency band in which the fundamental wave may exist can be predicted in advance, as in this embodiment, it is possible to improve the accuracy of matching and reduce the number of calculations (speed up processing). Furthermore, since any component whose frequency can be predicted based on the frequency of the fundamental wave can be used as a matching target, this method is effective not only for harmonics but also for extracting subharmonics. Another example of a predictable component to which template matching can be applied is aliasing noise present at frequencies folded back at the Nyquist frequency.

[0101] <4-3> In the above embodiment, the vehicle characteristic estimation process is executed by the road surface profile calculation unit 70, but the entity that executes the vehicle characteristic estimation process is not limited to the road surface profile calculation unit 70. For example, the vehicle characteristic estimation process may be executed by another functional block constructed in the control unit 61B, or a functional block constructed in a control unit mounted on another device other than the road surface flatness measurement device 60. In this modification, the dominant frequency extracted by the other functional block constructed in the control unit 61B, or a functional block constructed in a control unit mounted on another device other than the road surface flatness measurement device 60, is recorded in the road surface profile calculation unit 70.

[0102] <4-4> In the above embodiment, the tire rotation component removal process is performed using the predominant frequency extracted by the vehicle characteristics estimation process executed before the road surface profile calculation process. However, in the road surface profile calculation process, the vehicle characteristics estimation process may be executed before the tire rotation component removal process, and the extracted predominant frequency may be updated each time.

[0103] <4-5> In the above embodiment, in the template matching method, values ​​A T1 ~A T9and the spectral intensities B at the points S1 to S9 of the spatial frequency spectrum 90 corresponding to the points T1 to T9 of the template 80. S1 ~B S9 The matching value Mv was calculated based on the sum of the values ​​obtained by multiplying and . However, the specific method of template matching can be changed as desired. For example, the matching value Mv may be calculated based on the sum of differences or a correlation coefficient.

[0104] <4-6> The equation (5) for calculating the total matching value MT in the above embodiment can be changed to, for example, the following equation (13). MT=Wx·M Rx +W1·M R1 +W2·M R2 +W3·M R3 +W4·M R4 ···(13)

[0105] The value Wx is a weighting coefficient for the matching value between the fundamental wave template 80X and the fundamental wave 200X. 2 The value W3 is a weighting factor for the matching value between the second template 82 and the first harmonic 210. 3 The value W3 is a weighting factor for the matching value between the third template 83 and the third harmonic 220. 4 The value W4 is a weighting factor for the matching value between the fourth template 84 and the third harmonic 230. 5 These are weighting coefficients for the matching value with the harmonic 240. For example, by setting the relationship between these weighting coefficients Wx, W1, W2, W3, and W4 as Wx>W1>W2>W3>W4>0, it is possible to perform a matching evaluation in which importance is placed on the closer to the fundamental wave 200X, in other words, the lower the order.

[0106] <4-7> The contents of the road surface profile calculation process can be changed as desired. For example, in the road surface profile calculation process shown in Fig. 13, at least one of the low-pass filter process, the lateral force component removal process, and the band-pass filter process can be omitted. Furthermore, the lateral force component removal process may be performed after the displacement calculation process and before the resampling process.

[0107] <4-8> In the above embodiment, the road surface profile calculation unit 70 calculated the cumulative distance traveled by the test vehicle 100 based on the vehicle speed pulse in the resampling process, but the cumulative distance traveled by the test vehicle 100 may also be calculated based on the speed of the test vehicle 100. However, the accuracy of calculating the cumulative distance tends to be lower, particularly when the speed of the test vehicle 100 changes, than when the cumulative distance traveled by the test vehicle 100 is calculated based on the vehicle speed pulse as in this embodiment.

[0108] <4-9> In the above embodiment, the vehicle speed pulse coefficient is stored in advance in the storage unit 61A. However, for example, if the length of the road surface to be measured is known in advance, the road surface profile calculation unit 70 may calculate the vehicle speed pulse coefficient by dividing the length of the road surface to be measured by the cumulative value of vehicle speed pulse edges obtained while the test vehicle 100 is traveling on the road surface to be measured. Alternatively, the road surface profile calculation unit 70 may calculate the vehicle speed pulse coefficient by dividing the vehicle speed of the test vehicle 100, measured using a global positioning system (GPS) or the like, by the vehicle speed pulse frequency. In calculating the vehicle speed pulse frequency, the road surface profile calculation unit 70 may, for example, determine the interval time TX between an arbitrary vehicle speed pulse edge and the next vehicle speed pulse edge, and calculate the reciprocal of the interval time TX, 1 / TX, as the average frequency of that interval. In a preferred example, the road surface profile calculation unit 70 calculates the vehicle speed pulse frequency by smoothing the calculated average frequencies using a low-pass filter.

[0109] <4-10> In the above embodiment, the road surface flatness measuring device 60 is connected to the acquisition device 50 so as to be able to communicate with it via wire or wirelessly, but the road surface flatness measuring device 60 and the acquisition device 50 do not have to be connected so as to be able to communicate with each other. In this case, information such as measurement data acquired by the acquisition device 50 is stored in an auxiliary storage device such as a USB flash memory or an SD memory card. The road surface flatness measuring device 60 calculates a road surface profile by acquiring information such as measurement data from the auxiliary storage device. [Explanation of symbols]

[0110] 60: Road surface flatness measurement device (extraction device for dominant frequencies)

Claims

1. A calculation process of calculating a frequency spectrum by Fourier transforming a signal including components input at regular intervals; extracting a dominant frequency from the frequency spectrum; In the extraction process, a center frequency of a template that mimics a peak waveform of a frequency spectrum is moved within a predetermined search range, and a matching value is calculated based on values ​​at multiple points of the template and spectral intensities at multiple points of the frequency spectrum that correspond to the multiple points of the template, and the frequency at the position where the matching value is a predetermined optimum value is extracted as the dominant frequency. Dominant frequency extractor.

2. the frequency spectrum includes a fundamental wave and harmonics or subharmonics whose frequencies can be predicted based on the frequency of the fundamental wave; In the extraction process, the dominant frequency is extracted based on the matching value between the template and the fundamental wave and the matching value between the template and the harmonic or the subharmonic.

2. The dominant frequency extraction device according to claim 1.

3. a calculation step of calculating a frequency spectrum by Fourier transforming a signal including components input at regular intervals; an extraction step of extracting dominant frequencies from the frequency spectrum; In the extraction step, a matching value is calculated based on values ​​at multiple points on the template and spectral intensities at multiple points on the frequency spectrum corresponding to the multiple points on the template by moving the center frequency of the template that mimics the peak waveform of the frequency spectrum within a predetermined search range, and the frequency at the position where the matching value is a predetermined optimum value is extracted as the dominant frequency. Method for extracting dominant frequencies.

4. a calculation step of calculating a frequency spectrum by Fourier transforming a signal including components input at regular intervals; extracting dominant frequencies from the frequency spectrum; In the extraction step, a matching value is calculated based on values ​​at multiple points on the template and spectral intensities at multiple points on the frequency spectrum corresponding to the multiple points on the template by moving the center frequency of the template that mimics the peak waveform of the frequency spectrum within a predetermined search range, and the frequency at the position where the matching value is a predetermined optimum value is extracted as the dominant frequency. Dominant frequency extraction program.

Images