The present invention relates to a method and a device for calibrating the acoustic characteristic of a device that has an acoustic inspection function, and an FFT circuit.
In the process of final inspection of a vehicle provided with a plurality of (e.g. over a few hundred of) electronic control units (ECUs), a vehicle inspection device (line end tester (LET)) is used to determine or diagnose whether the functions of the ECUs and devices electrically connected to the ECUs are normal or abnormal.
The vehicle inspection device includes a microphone used to inspect for the normality of the sounds emitted from the acoustic devices of a vehicle. Hence, a reference sound having its acoustic characteristic measured beforehand is used to check the acoustic characteristic or the sensitivity characteristic of the microphone. More specifically, the reference sound is input to the microphone to measure the sensitivity of the microphone, and the deviation of the measurement value from a reference value (the acoustic characteristic of a sound source measured by a noise meter) of the microphone sensitivity is measured. Then, the offset or sensitivity of the microphone is corrected (refer to, for example, Japanese Patent Application Laid-Open No. 2000-121684 and Japanese Patent Application Laid-Open No. 9-210865).
However, according to the conventional calibration methods, the calibration is required to be performed with high repeatability by using a small sound source. In addition, the calibration is required to be easy even in a noisy environment of a plant. Further, when a vehicle inspection device is used to perform the acoustic inspection, the inspection has to be focused on a particular sound. It is publicly known that, in FFT processing, the circuit scale is reduced by repeatedly carrying out processing, whereas the number of times of communication between memories and the number of loops increase, thus leading to a lower processing speed.
It is an object of the present invention, therefore, to provide a technology related to acoustic characteristic calibration, which technology is superior to prior arts in at least one of improved repeatability of inspection state, a reduced influence of a noise in calibrating an acoustic characteristic, and a reduced load in FFT arithmetic processing.
An acoustic characteristic calibration method according to the present invention includes: a step of outputting a reference acoustic signal, the frequency characteristic of which is known, from a sound output unit and inputting the reference acoustic signal to a sound input unit of an acoustic inspection device; a step of subjecting the signal, which has been input to the sound input unit, to A/D conversion and then FFT arithmetic processing thereby to detect a frequency response characteristic of the sound input unit; and a step of determining a correction factor at each frequency by comparing the frequency response characteristic of the sound input unit and the frequency characteristic of the reference acoustic signal, and calibrating the frequency response characteristic of the sound input unit on the basis of the correction factor.
In the acoustic characteristic calibration method according to the present invention, preferably, the relative positions and attitudes of the sound output unit and the sound input unit are fixed by using an instrument which is fixed to a chassis of the acoustic inspection device and which is used to fix the sound output unit, and then the reference acoustic signal is output from the sound output unit.
In the acoustic characteristic calibration method according to the present invention, the instrument is preferably fixed to the chassis of the acoustic inspection device, avoiding the locations where acoustic inspection input interface and output interface, which are provided on the chassis of the acoustic inspection device, are disposed.
In the acoustic characteristic calibration method according to the present invention, the relative positions and attitudes of the sound output unit and the sound input unit are preferably fixed by using the instrument in which the sound output unit is disposed in a window formed of a through hole.
In the acoustic characteristic calibration method according to the present invention, the relative positions and attitudes of the sound output unit and the sound input unit are preferably fixed by using the instrument in which at least a portion where the sound output unit is fixed is composed of a transparent material.
In the acoustic characteristic calibration method according to the present invention, preferably, the relative positions and attitudes of the sound output unit and the sound input unit are fixed such that the interval between a vibration element of the sound output unit and a vibration detection element of the sound input unit is equal to or less than a half wavelength of an acoustic signal of a maximum frequency that defines the frequency characteristic of the sound input unit, and that the vibration element of the sound output unit and the vibration detection element of the sound input unit oppose each other.
In the acoustic characteristic calibration method according to the present invention, preferably, a plurality of calibration sonic signals ψsx which have a specified inter-phase relationship are output from the sound output unit, a plurality of acoustic signals ψ0x input to the sound input unit accordingly are detected, and predetermined complex calculation is carried out on the plurality of acoustic signals ox so as to remove noise sonic signals ψn which are included in the plurality of acoustic signals ψ0x and which are present in an environment of the acoustic inspection device, thereby detecting the frequency response characteristic of the sound input unit.
In the acoustic characteristic calibration method according to the present invention, preferably, each of a first calibration sonic signal ψs1 and a second calibration sonic signal ψs2, which have a specified relationship in which the phases are different, is output from the sound output unit, a first acoustic signal ψ01 and a second acoustic signal ψ02 input to the sound input unit accordingly are detected, and subtraction as the predetermined complex calculation is carried out on the first acoustic signal ψ01 and the second acoustic signal ψ02 thereby to detect a frequency response characteristic |ψ01−ψ02| of the sound input unit.
In the acoustic characteristic calibration method according to the present invention, preferably, each of the first calibration sonic signal ψs1 and the second calibration sonic signal ψs2, which have a specified relationship in which the phases are the same, while the amplitudes are different, is output from the sound output unit, the first acoustic signal ψ01 and the second acoustic signal ψ02 input to the sound input unit accordingly are detected, and subtraction as the predetermined complex calculation is carried out on the first acoustic signal ψ01 and the second acoustic signal ψ02 thereby to detect a frequency response characteristic |ψ01−ψ02| of the sound input unit.
In the acoustic characteristic calibration method according to the present invention, preferably, a butterfly arithmetic circuit for performing Fourier transform FFT (DFT) compares the configurations of circuits leading up to the bins of combinations that provide the same result after power value conversion out of the results obtained by reversing, at a Nyquist axis, an output result of a real image region and an output result of a mirror image region, and cuts a combination in which calculation closer to an initial stage in a butterfly arithmetic circuit composed of a plurality of stages can be deleted, thereby reducing the circuit scale of the butterfly arithmetic circuit.
According to the acoustic characteristic calibration method in accordance with the present invention, the repeatability of the inspection state is improved by using the instrument. The influence of a noise in acoustic characteristic calibration is reduced by carrying out the predetermined complex calculation on a plurality of acoustic signals detected in the case where a plurality of calibration sonic signals having a specified phase relationship are output. The processing load on an FFT circuit is reduced by using the improved butterfly calculation.
(Configuration Related to Acoustic Characteristic Calibration)
An acoustic controller 1 illustrated in
The acoustic controller 1 includes a sound output unit 10, a signal separation element 11, a first signal memory 121, a second signal memory 122, a signal arithmetic element 14, and a correction value arithmetic element 18. Each of the elements 11, 14 and 18 is composed of a processor or a processor core (hardware), which executes software and arithmetic processing based on the software. In the acoustic controller 1, some or all of the elements 10, 11, 121, 122, 14, and 18 may be the constituent elements of the vehicle inspection device 2.
The sound output unit 10 is composed of a speaker, and outputs an acoustic signal by vibrating a vibration plate (vibrating element). The signal separation element 11 separates an output signal from the vehicle inspection device 2 into a first acoustic signal and a second acoustic signal. The first signal memory 121 stores and retains the first acoustic signal separated by the signal separation element 11. The second signal memory 122 stores and retains the second acoustic signal separated by the signal separation element 11. The signal arithmetic element 14 carries out predetermined complex calculation on the first acoustic signal stored in the first signal memory 121 and the second acoustic signal stored in the second signal memory 122. The correction value arithmetic element 18 calculates the correction factor for calibrating the frequency response characteristic of the vehicle inspection device 2 or the sound input unit 20 on the basis of the result of comparison between the complex calculation result obtained by the signal arithmetic element 14 and the complex calculation result of a reference acoustic signal.
The vehicle inspection device 2 is connected to a plurality of ECUs (electronic control units) mounted in a vehicle, and used to determine or diagnose whether the plurality of ECUs are operating properly or not on the basis of the results of communication with the plurality of ECUs. Devices electrically connected to the ECUs mounted in the vehicle include acoustic devices, such as a meter alarm, a security horn, an audio speaker, a parking sensor buzzer, a smart buzzer, and an approach alarm. The vehicle inspection device 2 constitutes an acoustic inspection device for determining whether the sounds emitted from the acoustic devices are normal or not.
The vehicle inspection device 2 has the sound input unit 20, an amplifier 21, an A/D converter 22, and an FFT arithmetic element 24. The FFT arithmetic element 24 is composed of a dedicated circuit (hardware) in addition to a processor or a processor core that executes software and performs arithmetic processing according to software.
The sound input unit 20 is composed of a microphone, and an acoustic signal is input or detected by the vibration of a diaphragm (vibration detection element) caused by a sound wave. An acoustic signal input to the sound input unit 20 is amplified by the amplifier 21, and the amplified signal is converted from an analog signal into a digital signal by the A/D converter 22. The FFT arithmetic element 24 carries out FFT arithmetic processing, more specifically, the improved butterfly arithmetic processing in accordance with the present invention (refer to
As illustrated in
An instrument 4 illustrated in
The first base 41 is attached or joined to the second base 42 such that the first base 41 is placed over the central portion of the horizontal beam part in the longitudinal direction (in the direction of thickness). At least the first base 41 and the second base 42 are composed of a transparent resin or material, such as PET. An approximately circular window 410, which passes through in the direction of thickness, is provided in the first base 41. A window 420, which passes through in the direction of thickness and which has a smaller diameter than the window 410, is provided at the center of the horizontal beam part of the second base 42.
The instrument 4 is fixed to the chassis 200 by bringing the hooks 44 into contact with the left side surface, the right side surface, and the bottom surface, respectively, of the chassis 200. Further, a buffer material, such as sponge or urethane (not illustrated), is used for the contact surfaces of the instrument 4 and the chassis 200, thus improving adhesiveness and reducing the influences of external sound entering through a gap.
At this time, as illustrated in
Further, as illustrated in
(Acoustic Characteristic Calibration Method)
First, the sound output unit 10 is placed and fixed to the window 410 of the first base 41 of the instrument 4, and then the instrument 4 is fixed to the chassis 200 of the vehicle inspection device 2, as illustrated in
In the vehicle inspection device 2, it is determined whether the sound input unit 20 is ON (STEP202 of
In response to the command, the sound output unit 10 of the acoustic controller 1 continuously outputs a reference acoustic signal, the frequency characteristic of which is known (STEP102 of
Alternatively, as the reference acoustic signals, a first calibration signal ψs1 and a second calibration signal ψs2, which have a specified relationship in which the phases thereof are the same and which have different amplitudes, may be output in the first period and the second period, which is different from the first period, respectively (refer to
In the vehicle inspection device 2, it is determined whether a predetermined time has elapsed (STEP206 of
Meanwhile, if the determination result is affirmative (YES in STEP206 of
Thereafter, in the vehicle inspection device 2, the FFT calculation is carried out on the acoustic signals stored in the storage device (not illustrated), thereby to detect the frequency response characteristic (e.g. the intensity of an acoustic signal at each frequency) of the sound input unit 20 (STEP210 of
The frequency characteristic of the sound input unit 20 is transmitted from the vehicle inspection device 2 to the acoustic controller 1, and the first acoustic signal ψ01 and the second acoustic signal ψ02, which represent the frequency characteristics in the first period and the second period, respectively, are separated and then stored in the memories 121 and 122, respectively (STEP106 of
The correction factor is transmitted from the acoustic controller 1 to the vehicle inspection device 2 and reflected on the amplification factor of the amplifier 21, thereby calibrating the frequency response characteristic of the vehicle inspection device 2 or the sound input unit 20 (STEP212 of
In place of the flowchart illustrated in
Thus, the acoustic signals continuously input to the sound input unit 20 are subjected to the amplification and the A/D conversion, and the FFT processing is concurrently carried out in succession. Therefore, when the sound output OFF command (STEP208 of
(Improved Butterfly Calculation)
In the FFT arithmetic processing (refer to STEP210 of
The difference between the signals in the real image region section and the mirror image region section is denoted by a reversed sign of an imaginary signal (Q output), so that conversion to power values provides results that are laterally symmetrical with respect to the Nyquist frequency fs/2. Therefore, the bin of the calculation result of the mirror image region section can be folded back and replaced by the bin of the result of the real image region section. In other words, the calculation volume of the FFT arithmetic processing (DFT) can be reduced by cutting one of the equivalent bins.
The following describes a specific example of the cutting. When equivalent output bins are compared, one bin includes rotation calculation in the calculation base in a block constituting the butterfly calculation, whereas the other bin does not include the rotation calculation in the calculation base in a block constituting the butterfly calculation. Therefore, the calculation volume of the FFT arithmetic processing (DFT) can be cut down by reducing the arithmetic load by cutting the butterfly calculation part that includes the rotation calculation involving a large calculation volume (the result obtained after scrambling (bit reversal processing) is the same as the result of FFT arithmetic processing (DFT) of normal algorithm).
Further,
Number | Date | Country | Kind |
---|---|---|---|
2016-190453 | Sep 2016 | JP | national |