JP6878187B2 - System and method - Google Patents

Description

Embodiments of the present invention relate to voice processing.

There is a need for equipment and services that collect the operating noise of industrial equipment to detect abnormalities at an early stage and detect deterioration. By recording the operating sound using a microphone, for example, there is a possibility that an abnormality can be detected earlier than when a vibration sensor is used. Further, since it is a microphone, it is not necessary to bring the sensor into contact with the target device, and the operating sound can be sensed without contact.

Japanese Patent No. 5543023

The sound collected by the microphone may also include noise in the surrounding environment. Therefore, it is necessary to suppress directional noise by performing beamforming processing on the collected sound, which has a sharp directivity toward the operating sound (target sound). In general, in order for the beamforming process to have sharp directivity (for example, ± 10 degrees around 0 degrees), it is better to increase the number of microphones.

However, it is required to reduce the cost of the device for collecting the operating sound, and it is desired to reduce the number of microphones. Therefore, it is necessary to realize a new function capable of acquiring sound for each frequency band having sharp directivity with inexpensive hardware.

One aspect of the present invention is to provide a system and a method capable of acquiring sound for each frequency band having sharp directivity with inexpensive hardware.

According to the embodiment, the system is a system for collecting sound emitted from an object, and includes a first microphone, a second microphone separated from the first microphone by a first distance, and the second microphone. A third microphone separated by a second distance and separated from the first microphone by a third distance is provided. The directivity of sound collection in the first frequency band using the first microphone and the second microphone is the first direction. The directivity of sound collection in the second frequency band using the second microphone and the third microphone is the second direction. The directivity of sound collection in the third frequency band using the first microphone and the third microphone is the third direction. The position of the object is in the direction of the first direction from the first microphone and the second microphone, and in the direction of the second direction from the second microphone and the third microphone, and the first direction. The positional relationship between the first microphone, the second microphone, the third microphone, and the object is determined so that the direction is from the microphone and the third microphone in the third direction. The second distance is larger than the first distance. The third distance is larger than the second distance. The second frequency band is lower than the first frequency band. The third frequency band is lower than the second frequency band.

The perspective view which shows the appearance of the system which concerns on 1st Embodiment. The cross-sectional view which shows the structure in the housing of the system of the same embodiment. Block diagram showing the configuration of the system of the same embodiment The block diagram which shows the functional structure of the system of the same embodiment. The figure which shows one configuration example of the mode information which shows the mode for detecting the abnormality of the target device by the system of the same embodiment. The figure for demonstrating the example in which the system of the same embodiment determines one type of abnormality using the voice data of one frequency band. The figure which shows the example of the directivity of the audio signal acquired when the system of the same embodiment is a 1st mode. The figure which shows the example of the directivity of the audio signal acquired when the system of the same embodiment is a 3rd mode. The figure which shows another example of the directivity of the audio signal acquired when the system of the same embodiment is a 3rd mode. The figure for demonstrating the example of machine learning and identification used for abnormality detection by the system of the same embodiment. The figure for demonstrating another example of machine learning and identification used for anomaly detection by the system of the same embodiment. The flowchart which shows the example of the procedure of the abnormality detection processing executed by the system of the same embodiment. The figure for demonstrating an example in which the system of the same embodiment determines one type of abnormality using voice data of a plurality of frequency bands. The flowchart which shows another example of the procedure of the abnormality detection processing executed by the system of the same embodiment. The block diagram which shows the functional structure of the system of 2nd Embodiment. The figure for demonstrating two modes which can be switched by the microphone used in the system of the same embodiment. The figure for demonstrating the mode for detecting the abnormality of the target device by the system of the same embodiment. The flowchart which shows the example of the procedure of the abnormality detection processing executed by the system of the same embodiment.

Hereinafter, embodiments will be described with reference to the drawings.

(First Embodiment)
FIG. 1 is a perspective view showing the appearance of the system according to the first embodiment. This system is equipped with at least three microphones for collecting sounds emitted from an object, and has a function of processing the sounds collected by those microphones. The objects that the system collects sound from are various industrial devices such as industrial robots, machine tools, and uninterruptible power supplies.

The system can be realized, for example, as various electronic devices used as IoT edge devices. The electronic device may be, for example, a mobile information terminal such as a mobile phone, a smartphone, a PDA, a personal computer, or an embedded system built in various electronic devices. Further, the system may be composed of an electronic device provided with a microphone and collecting sound, and a plurality of electronic devices having functions for the system, such as an electronic device for processing the sound. In the following, a case where this system is realized as an electronic device 1 will be illustrated.

As shown in FIG. 1, the electronic device 1 has a thin box-shaped housing 10. The housing 10 contains a board (board) 11. Various components for acquiring and processing audio and wiring between the components are provided on the substrate 11. This component includes at least three microphones 21, 22, 23.

At least three openings 12, 13, and 14, which are the same number as the microphones 21, 22, and 23, are provided on the front surface of the housing 10. Microphones 21, 22, and 23 are arranged at positions on the substrate 11 that correspond to the rear of the openings 12, 13, and 14, respectively. Although FIG. 1 shows an example in which microphones 21, 22, 23 are arranged in a row, these microphones 21, 22, 23 may be arranged so as to form a triangle. Further, the number of microphones and openings may be four or more. If there are four microphones, they may be arranged in a row or in a "U" shape.

FIG. 2 is a cross-sectional view showing the structure inside the housing 10 when the electronic device 1 is viewed from the side. As described above, microphones 21, 22, and 23 are arranged at positions on the substrate 11 that correspond to the rear of the openings 12, 13, and 14, respectively. These microphones 21, 22, and 23 are arranged, for example, on the front surface of the substrate 11. The microphones 21, 22, and 23 may be incorporated in the substrate 11 (built-in), or may be connected to the substrate 11 via a conducting wire. Further, the positions of the microphones 21, 22, and 23 may be movable by the user. The electronic device 1 is arranged in front of the target device 5 which is a sound source, for example, and the sound emitted by the target device 5 passes through the openings 12, 13 and 14, and is collected by the respective microphones 21, 22 and 23. ..

Further, the second microphone 22 is separated from the first microphone 21 by a first distance. The third microphone 23 is separated from the second microphone 22 by a second distance, and is separated from the first microphone by a third distance. The first distance, the second distance, and the third distance are different from each other. For example, the second distance is greater than the first distance and the third distance is greater than the second distance. When the microphones 21, 22, and 23 are arranged in a row as in the example shown in FIG. 2, the sum of the first distance (= X, for example, 1 cm) and the second distance (= Y, for example, 2 cm). Is the third distance (= X + Y, for example, 3 cm).

There is a close relationship between the distance between microphones and the directivity of sound collection for each frequency band. The directivity of sound collection in the first frequency band using the first microphone 21 and the second microphone 22 is the first direction. The directivity of sound collection in the second frequency band using the second microphone 22 and the third microphone 23 is the second direction. Further, the directivity of sound collection in the third frequency band using the first microphone 21 and the third microphone 23 is the third direction. The positional relationship between the first microphone 21, the second microphone 22, the third microphone 23, and the target device 5 is that the position of the target device 5 is in the first direction from the first microphone 21 and the second microphone 22. , The direction from the second microphone 22 and the third microphone 23 to the second direction, and the direction from the first microphone 21 and the third microphone 23 to the third direction. The first frequency band, the second frequency band, and the third frequency band are different from each other. For example, the second frequency band is lower than the first frequency band, and the third frequency band is lower than the second frequency band.

The first direction, the second direction, and the third direction may be substantially the same. For example, when the target device 5 and all the microphones 21, 22, and 23 are arranged in a line on one straight line, the first direction, the second direction, and the third direction are substantially the same. Therefore, the directivity of sound collection in the first frequency band using the first microphone 21 and the second microphone 22 and the directivity of sound collection in the second frequency band using the second microphone 22 and the third microphone 23. And the directivity of the sound collection of the third frequency band using the first microphone 21 and the third microphone 23 may be substantially the same first direction. That is, the positional relationship between the first microphone 21, the second microphone 22, the third microphone 23, and the target device 5 is such that the position of the target device 5 is substantially different from that of the first microphone 21, the second microphone 22, and the third microphone 23. It has the same first direction.

On the back surface of the board 11, a processor unit 15 in which a processor such as a CPU is arranged and a memory unit 16 in which various memories are arranged are provided. Not limited to these, various components (for example, components for communication and display) and wiring may be arranged on the substrate 11.

In order to record the operating sounds of various target devices 5 such as industrial devices and detect abnormalities such as deterioration, it is desired that sounds in various frequency bands corresponding to the target devices 5 be acquired. For example, in a device having a rotating mechanism, when the bearing is worn, signs of deterioration appear in the non-audible high frequency band, and signs of deterioration gradually appear in the low frequency band, and then vibration of the device occurs. Therefore, by acquiring the operating sound using a microphone, for example, there is a possibility that an abnormality of the device can be detected earlier than when a vibration sensor is used.

Therefore, the electronic device 1 acquires sounds in a plurality of frequency bands such as 20 to 40 kHz, 40 to 60 kHz, and 80 to 100 kHz in order to detect an abnormality in one target device 5. In order to produce such an electronic device 1 at low cost, it is desirable that the electronic device 1 has general-purpose hardware and software for various target devices 5. For example, the beamforming process using the sound collected by the two microphones included in the software to be used is defined as the beamforming process for obtaining the sound having sharp directivity in each of the above three frequency bands. There is a problem that it is necessary to prepare three sets of microphone pairs having different microphone intervals, and therefore six microphones are required, which is expensive.

In the present embodiment, as described above, three microphones 21, 22, and 23 can be used to form three patterns having different distances between microphones and frequency characteristics. Therefore, it is possible to acquire the sound for each frequency band having sharp directivity with inexpensive hardware. When the electronic device 1 is provided with three microphones, the variation of the microphone set for acquiring the sound used for the beamforming process is limited to three patterns using the sound acquired by the two microphone pairs. However, three patterns using the sound acquired by one microphone and one pattern using a set of three microphones may be included. Further, when four microphones are provided in the electronic device 1, there are three patterns in which the sound acquired by one microphone is used in the variation of the set of microphones for acquiring the sound used for the beamforming process. , Six patterns using sounds acquired by two microphone pairs, four patterns using a set of three microphones, and one pattern using a set of four microphones can be included.

Therefore, by performing beamforming processing so as to select a target frequency using sounds acquired by using three or more N microphones provided in the electronic device 1, N or more types of directivity patterns can be obtained. Sounds (N or more target frequencies) can be acquired. The target frequencies are, for example, two types, a low frequency band and a high frequency band, or three types, a low frequency band, a medium frequency band, and a high frequency band. The electronic device 1 can detect an abnormality in the target device 5 by using the acquired sound.

FIG. 3 shows an example of the system configuration of the electronic device 1. The electronic device 1 includes a CPU 101, a main memory 102, a non-volatile memory 103, an audio controller 104, a display controller 105, a communication interface 106, and the like.

The CPU 101 controls the operation of various components in the electronic device 1. The CPU 101 executes various programs loaded from the non-volatile memory 103, which is a storage device, into the main memory 102. These programs include an operating system (OS) 102A and various application programs. The application program includes an abnormality detection program 102B. The abnormality detection program 102B includes a group of commands for collecting the sound emitted by the target device 5, a group of commands for detecting an abnormality of the target device 5 based on the sound, a group of commands for recording, and the collected sounds. Includes instructions for outputting information about detected anomalies, etc.

The audio controller 104 controls each microphone 21, 22, 23. The audio controller 104 controls, for example, whether to turn on or off each microphone 21, 22, 23, or to acquire an audio signal from each microphone 21, 22, 23. Further, when each microphone 21, 22, 23 has a plurality of modes having different sound collecting characteristics, the mode to be used can be switched. The plurality of modes are, for example, a low frequency mode suitable for collecting sound in a low frequency band centered on a low frequency (pole) and a high frequency mode suitable for collecting sound in a high frequency band centered on a high frequency. including.

The display controller 105 controls the display 105A. The display signal generated by the display controller 105 is sent to the display 105A. The display 105A displays a screen image based on the display signal. The display 105A may be built in the electronic device 1 or may be an external display connected to the electronic device 1 via various interfaces.

Communication interface 106 is a device configured to perform wired or wireless communication. The communication interface 106 includes a transmitting unit that transmits a signal and a receiving unit that receives the signal. By data transmission via the communication interface 106, it is possible to notify the terminal or the like used by the administrator of information on the sound collected from the target device 5 or an abnormality of the target device 5.

The functional configuration of the abnormality detection program 102B will be described with reference to FIG. The abnormality detection program 102B includes a beamforming (BF) control unit 31, a beamforming processing unit 32, a recording unit 33, a calculation unit 34, a display control unit 35, and an abnormality detection unit 36.

The beamforming control unit 31 controls the beamforming processing unit 32 so that the beamforming processing is performed according to the type of abnormality to be detected. The beamforming control unit 31 switches between a microphone to be operated and a channel (frequency band) used by the beamforming processing unit 32 according to the type of abnormality.

More specifically, the beamforming control unit 31 turns on each microphone 21, 22, 23, for example, in order to select the microphone to be used from the microphones 21, 22, 23, depending on the type of abnormality. Or control off. The beamforming control unit 31 may select the sound data used by the beamforming processing unit 32 from the sound data obtained by collecting sounds from the microphones 21, 22, and 23.

The beamforming control unit 31 controls the target frequency (frequency band) of the beamforming processing by the beamforming processing unit 32 according to the type of abnormality. The beam forming control unit 31 sets the target frequency to, for example, one of a low frequency band, a medium frequency band, and a high frequency band according to the type of abnormality.

The abnormality that occurs in the target device 5 generally has different frequency characteristics depending on the type of abnormality and the type of abnormality such as the site where the abnormality occurs. From this, the electronic device 1 can detect an abnormality of the first type using, for example, audio data of a certain frequency band, and uses audio data of another frequency band to be different from the abnormality of the first type. Two types of abnormalities can be detected. That is, voice data in a plurality of frequency bands can be used for different purposes related to abnormality detection.

FIG. 5 shows a configuration example of mode information regarding a mode set in the electronic device 1 in order to detect an abnormality in the target device 5. Here, in each mode, a case where voice in a certain frequency band is detected by using two of the three microphones 21, 22, and 23 will be illustrated. The mode information includes a plurality of records corresponding to the plurality of modes. Each record includes, for example, a first microphone, a second microphone, a third microphone, a distance between microphones, a target frequency, and an application.

In a record corresponding to a mode, the "first microphone" indicates whether the first microphone 21 is set to the on state or the off state in that mode. When the "first microphone" indicates the on state, the first microphone 21 is capable of collecting sound, and therefore, the voice signal (voice data) obtained by collecting the sound by the first microphone 21 is set to be available. To. Further, when the "first microphone" indicates an off state, the first microphone 21 cannot collect sound, or the audio signal (audio data) obtained by collecting sound by the first microphone 21 cannot be used. Can be set. Similarly, the "second microphone" indicates whether the second microphone 22 is set to the on state or the off state in the mode. Further, the "third microphone" indicates whether the third microphone 23 is set to the on state or the off state in the mode. In the example shown in FIG. 5, in each mode, any two of the "first microphone", the "second microphone", and the "third microphone" are set to the on state.

"Distance between microphones" indicates the distance between microphones used in that mode. The "distance between microphones" indicates, for example, the distance between two microphones set to the on state in the mode.

The "target frequency" indicates the frequency band of the voice signal indicated by the voice data generated for detecting the abnormality of the target device 5 in the mode. That is, the voice data generated for detecting the abnormality of the target device 5 is generated so as to include the signal component of the frequency band indicated by the “target frequency”. For example, by applying a bandpass filter (BPF) to the audio data acquired by the microphone in the on state, the signal component of the target frequency band can be extracted. The voice data has directivity in the direction in which the target device 5 exists.

"Use" indicates the use of the mode. “Use” indicates, for example, the type of abnormality detected by the mode. When an abnormality is detected, for example, the content shown in this "use" can be notified to the administrator or the like.

The beamforming control unit 31 can, for example, set the electronic device 1 to any one mode and control the beamforming processing unit 32 as described above according to the record of the mode information corresponding to the mode.

The beamforming processing unit 32 performs beamforming processing using two or more audio data obtained by two or more microphones under the control of the beamforming control unit 31. More specifically, first, the beamforming processing unit 32 extracts audio data in the target frequency band by applying a bandpass filter (BPF) to each audio data. Then, the beamforming processing unit 32 uses the extracted voice data to generate voice data having directivity in the direction in which the target device 5 exists.

The beamforming processing unit 32 uses two or more voice data obtained by two or more microphones to generate voice data having directivity in the direction in which the target device 5 exists, and then generates the voice data. The audio data in the target frequency band may be extracted by applying the above-mentioned bandpass filter to the generated audio data.

With such a configuration, the beamforming processing unit 32 can acquire audio data in the target frequency band having directivity in the direction in which the target device 5 exists.

The recording unit 33 can store the acquired audio data as a file (for example, a WAV format file) including waveform data in a storage medium such as the non-volatile memory 103. The recording unit 33 may store the audio data obtained by collecting sounds by the microphones 21, 22, and 23 in the storage medium as it is as a file including the waveform data. The recording unit 33 includes, for example, the sound collected by the first microphone 21, the sound collected by the second microphone 22, the sound collected by the third microphone 23, the sound collection result of the high frequency band, and the middle. At least one of the sound collection result of the frequency band and the sound collection result of the low frequency band can be stored in the storage medium.

The calculation unit 34 calculates various parameters related to the voice data by analyzing the voice data acquired by the beamforming processing unit 32 or the voice data obtained by collecting sounds by the microphones 21, 22, and 23. This parameter includes, for example, the power of sound collected by at least one of the first microphone 21, the second microphone 22, and the third microphone 23, and the high frequency band, medium frequency band, and low frequency band. Includes the power and spectrum of a particular frequency band for one of the sound collection results. More specifically, the parameters include the power of sound in all frequency bands collected by at least one microphone 21, 22, 23, the power of sound in a specific frequency band (target frequency) specified by the user, and short. The frequency spectrum of time and the like are included. The calculation unit 34 calculates, as the power, a value based on the amplitude, for example, the cumulative sum of the squares of the amplitudes of the signals in a certain section included in the voice data. Further, the calculation unit 34 calculates the frequency spectrum by performing a Fourier transform (for example, FFT) on the signal in a certain section included in the voice data acquired by the beamforming processing unit 32.

The display control unit 35 displays at least one of the parameters calculated by the calculation unit 34 on the screen of the display 105A via the display controller 105. The parameters are displayed on the screen in real time in various display forms such as numerical values, meters, and graphs. For example, by displaying the power of sound among the parameters, it is possible to notify the user that the target device 5 is being sensed and that the target device 5 is operating. Further, since the frequency spectrum shows the frequency characteristics of the audio data acquired by the beamforming processing unit 32, the user is notified that the signal of the target frequency can be sensed by displaying the frequency spectrum. Can be done.

The abnormality detection unit 36 detects an abnormality in the target device 5 by using the voice data acquired by the beamforming processing unit 32. For example, the abnormality detection unit 36 learns the voice data of the target device 5 during normal operation, which has been collected in advance, so that the voice data indicates the normal operation sound of the target device 5. Generate a classifier that identifies whether it is showing an abnormal operating noise. The abnormality detection unit 36 uses this classifier to identify whether the voice data acquired by the beamforming processing unit 32 indicates a normal operating sound of the target device 5 or an abnormal operating sound. By doing so, the abnormality of the target device 5 is detected. Examples of machine learning and identification will be described later with reference to FIGS. 10 and 11.

Further, the abnormality detection unit 36 acquires a power spectrum by Fourier transforming the voice data, for example, and detects an abnormality in the target device 5 based on the change in the power spectrum of the voice data in the first period. May be good. The abnormality detection unit 36 determines that an abnormality has occurred in the target device 5, for example, when the amount of change in the power spectrum or the amount of change in the specific frequency band is equal to or greater than the threshold value.

The abnormality detection unit 36 notifies the user of information regarding the detected abnormality. This information differs depending on the detected abnormality, and indicates the type of the abnormality, the countermeasure for the abnormality, and the like. Further, this information may be displayed on the screen of the display 105A by the display control unit 35, or may be transmitted by the abnormality detection unit 36 to a computer or the like used by the administrator via the communication interface 106.

FIG. 5 shows an example in which a first mode, a second mode, and a third mode are defined in order to detect three types of abnormalities. Here, a case is illustrated in which the abnormality is so serious that the abnormality of the target device 5 is detected in a lower frequency band.

In the first mode, the beamforming control unit 31 and the beamforming processing unit 32 use the sound data collected by the first microphone 21 and the third microphone 23 having a distance between the microphones of 3 cm, and use the target device. Generates audio data in a frequency band (low frequency band) of 6 kHz to 12 kHz having directivity to 5. The abnormality detection unit 36 determines the presence or absence of an abnormality in the target device 5 using the generated voice data. If there is an abnormality, notification of the failure status and prompt implementation of parts replacement are urged.

In the second mode, the beamforming control unit 31 and the beamforming processing unit 32 use the sound data collected by the second microphone 22 and the third microphone 23 having a distance between the microphones of 2 cm, and use the target device. Generates audio data in a frequency band (medium frequency band) of 12 kHz to 18 kHz having directivity to 5. The abnormality detection unit 36 determines the presence or absence of an abnormality in the target device 5 using the generated voice data. Then, if there is an abnormality, notification of failure detection and parts arrangement are urged.

In the third mode, the beamforming control unit 31 and the beamforming processing unit 32 use the sound data collected by the first microphone 21 and the second microphone 22 having a distance between the microphones of 1 cm, and use the target device. Generates audio data in a frequency band (high frequency band) of 18 kHz to 24 kHz having directivity to 5. The abnormality detection unit 36 determines the presence or absence of an abnormality in the target device 5 using the generated voice data. Then, if there is an abnormality, the sign of failure is found at an early stage, so that the notification of the sign of failure and the budget for parts replacement are promoted.

In this way, the abnormality detection unit 36 detects the abnormality of the target device 5 by using at least one sound collection result of the low frequency band, the medium frequency band, and the high frequency band. When the abnormality detection unit 36 detects an abnormality in the target device 5 by using the sound collection result of any one of the low frequency band, the medium frequency band, and the high frequency band, the abnormality detection unit 36 is of the first type of the target device 5. Detect as abnormal. Further, the abnormality detection unit 36 detects an abnormality in the target device 5 by using another sound collection result different from any one of the above among the low frequency band, the medium frequency band, and the high frequency band. If this is the case, it is detected that the target device 5 has an abnormality of the second type different from the first type.

The three modes described above are dynamically switched, for example, by the beamforming control unit 31 in order from the mode targeting a high frequency, and when an abnormality is detected in the mode targeting a certain frequency band, the mode is described. You can switch to a mode that targets a frequency band lower than the frequency band. That is, when an abnormality is detected in the third mode targeting the high frequency band, the mode is switched to the second mode targeting the middle frequency band, and when an abnormality is detected in this second mode, the value is low. It is possible to switch to the first mode for the frequency band. In this case, for example, the abnormality detection unit 36 notifies the beamforming control unit 31 that an abnormality has been detected in the third mode, and the beamforming control unit 31 responds to this notification by the beamforming processing unit 32. The operation is controlled to switch to the second mode. Further, the abnormality detection unit 36 notifies the beamforming control unit 31 that an abnormality has been detected in the second mode, and the beamforming control unit 31 operates the beamforming processing unit 32 in response to this notification. Control to switch to one mode. Thereby, it is possible to determine the presence or absence of abnormality detection in order from the mode in which the severity of deterioration is low.

More specifically, the abnormality detection unit 36 first detects an abnormality from the voice in the high frequency band among the low frequency band, the middle frequency band, and the high frequency band, and then the voice in the middle frequency band or the low frequency band. When an abnormality is detected from the above, it is detected as a third type abnormality of the target device 5. Further, when an abnormality is first detected from the voice of the middle frequency band or the low frequency band among the low frequency band, the middle frequency band and the high frequency band, the abnormality detection unit 36 is of the third type of the target device 5. It is detected as a fourth type of abnormality different from the abnormality. At least one of the first to fourth types indicates, for example, any one of a sign of failure, a failure state, and a difference in the length of remaining life. Further, at least one of the first to fourth types of abnormalities is a failure of the first member provided in the target device 5, a failure of the second member provided in the target device 5, a failure cause of the first member, and a first. It may indicate any one of the differences in the causes of failure of the members.

That is, as shown in FIG. 6, one type of abnormality can be discriminated by using the voice data of one target frequency band. The type of abnormality is represented, for example, as a failure mode according to the severity of deterioration. The three failure modes shown in FIG. 6 are common in that they are desired to have sharp directivity with respect to the target sound, but the target frequency bands are different. Therefore, by changing the beamforming process by changing the distance between the microphones, it is possible to acquire audio data with sharp directivity for each target frequency band. By using such voice data, it is possible to realize more accurate abnormality detection.

Next, FIG. 7 shows an example of the directivity of the voice signal (voice data) acquired by the beamforming control unit 31 and the beamforming processing unit 32 in the first mode. Here, the signal arrival direction (angle) and gain of the 6 kHz audio signal when the audio signal obtained by using the first microphone 21 and the third microphone 23 having a distance between the microphones of 3 cm is subjected to beamforming processing. Relationship 41 with. For the beamforming process, for example, a delay sum array method is used.

As shown in FIG. 7, this audio signal has a sharp directivity in which the gain is maximized at 0 degrees, which is the direction of arrival of the target sound (target signal), that is, the sensitivity is particularly high at 0 degrees. It has a beam (lobe) that is directional. The target sound is, for example, the operating sound of the target device 5. Therefore, by the beamforming process, the beam is steered in the direction of the target sound, and the target sound included in the audio signal can be emphasized.

As described above, in the first mode, it is possible to acquire an audio signal having sharp directivity in a low frequency band including 6 kHz.

Next, FIGS. 8 and 9 show an example of the directivity of the audio signal acquired by the beamforming control unit 31 and the beamforming processing unit 32 in the third mode.

First, FIG. 8 shows the signal arrival direction (angle) of the 18 kHz audio signal when the audio signal obtained by using the first microphone 21 and the second microphone 22 having a distance between the microphones of 1 cm is subjected to beamforming processing. ) And the gain 42. As shown in FIG. 8, this audio signal has a sharp directivity in which the gain is maximized at 0 degrees, which is the direction of arrival of the target sound, that is, a beam having a particularly high sensitivity at 0 degrees. have. Therefore, by the beamforming process, the beam is steered in the direction of the target sound, and the target sound included in the audio signal can be emphasized.

Next, FIG. 9 shows the signal arrival direction of the 24 kHz audio signal when the audio signal obtained by using the first microphone 21 and the second microphone 22 having a distance between the microphones of 1 cm is subjected to beamforming processing. The relationship 43 between the angle) and the gain is shown. As shown in FIG. 9, this audio signal has a sharp directivity in which the gain is maximized at 0 degrees, which is the direction of arrival of the target sound, that is, a beam having a particularly high sensitivity at 0 degrees. have. Although the gain is high near 90 degrees and −90 degrees, it can be said that the target sound included in the audio signal can be sufficiently emphasized because there is a difference from the gain in the 0 degree direction.

As described above, in the third mode, it is possible to acquire an audio signal having sharp directivity in a high frequency band including 18 kHz and 24 kHz.

Next, FIG. 10 shows an example of machine learning and identification used for abnormality detection by the abnormality detection unit 36. The abnormality detection unit 36 learns in advance using the voice data of the target device 5 when it is normal, and generates a classifier that identifies the presence or absence of an abnormality in the target device 5. By this learning, the abnormality detection unit 36 determines, for example, the identification boundary surface 51 at which a certain voice should be determined to be a normal sound by using one or more feature quantities included in the normal voice data.

When the voice having the feature amount 52 located in the identification boundary surface 51 is acquired from the target device 5, the abnormality detection unit 36 determines that the target device 5 is normal. When the voice having the feature amount 53 located outside the identification boundary surface 51 is acquired from the target device 5, the abnormality detection unit 36 determines that the target device 5 is abnormal. The abnormality detection unit 36 can also estimate the degree of deterioration of the target device 5 based on the distance from the identification boundary surface 51 to the feature amount 53. The abnormality detection unit 36 estimates, for example, that the greater the distance, the greater the degree of deterioration (degree of abnormality).

Further, FIG. 11 shows an example of an autoencoder (Auto-Encoder) used for abnormality detection. The abnormality detection unit 36 learns the normal pattern using the normal voice data of the target device 5, and generates an autoencoder 55 having a network capable of reproducing the normal pattern. The autoencoder 55 has, for example, a network having a three-layer structure of an input layer, an intermediate layer, and an output layer. When the normal voice (normal pattern) 54 is input, the autoencoder 55 has learned the normal pattern, so that the normal voice 56 can be reproduced as an output. On the other hand, when the abnormal voice (abnormal pattern) 57 is input, the autoencoder 55 cannot reproduce the abnormal voice 57 as an output because the autoencoder 55 has not learned the abnormal pattern, and the voice is different from the input. Output 58.

Utilizing such characteristics of the autoencoder 55, the abnormality detection unit 36 determines the abnormality of the target device 5 based on the degree of data similarity between the input voice and the output voice (that is, the reproduced voice). Judge the presence or absence of. For example, the abnormality detection unit 36 determines that the target device 5 is normal when the similarity is equal to or higher than the threshold value, and determines that the target device 5 is abnormal when the similarity is less than the threshold value. The abnormality detection unit 36 can also estimate that the lower the degree of similarity, the greater the degree of deterioration (degree of abnormality).

The classifier shown in FIG. 10 and the autoencoder shown in FIG. 11 may be generated by a device other than the electronic device 1 (another computer or the like). By incorporating the function (instruction group) of the classifier or autoencoder generated by the other device into the abnormality detection unit 36 of the electronic device 1, the abnormality detection unit 36 detects the presence or absence of an abnormality from the voice of the target device 5. It can be determined.

Next, an example of the procedure of the abnormality detection process executed by the electronic device 1 will be described with reference to the flowchart of FIG. This abnormality detection process is realized, for example, by the CPU 101 of the electronic device 1 executing the instruction group included in the abnormality detection program 102B.

First, in the third mode for detecting an abnormality in the target device 5 using the voice in the high frequency band, the CPU 101 acquires the first voice data by collecting the sound using the first microphone 21 and the second microphone. The second voice data is acquired by collecting sound using 22 (step S101). The CPU 101 uses the first audio data and the second audio data to generate audio data in a high frequency band having directivity in the direction in which the target device 5 exists (step S102). The CPU 101 generates, for example, high frequency band audio data having directivity in the direction in which the target device 5 exists by beamforming processing using the first audio data and the second audio data.

The CPU 101 uses the acquired voice data in the high frequency band to determine whether or not there is an abnormality in the target device 5 (step S103). The CPU 101 uses, for example, a classifier that identifies the presence or absence of an abnormality, which is generated by machine learning using voice data in a high frequency band during normal operation of the target device 5. By using this classifier, the CPU 101 determines whether or not there is an abnormality in the target device 5 by using the acquired voice data in the high frequency band. If there is no abnormality (NO in step S103), the CPU 101 ends the process.

If there is an abnormality (YES in step S103), the CPU 101 notifies the failure sign and the budget for parts replacement (step S104). The CPU 101 may notify that an abnormality has been detected in the high frequency band. The CPU 101 can realize this notification by displaying the display 105A on the screen or transmitting information to the terminal used by the administrator via the communication interface 106.

Then, the CPU 101 shifts to the second mode for detecting the abnormality of the target device 5 using the voice in the middle frequency band, acquires the third voice data by collecting the sound using the second microphone 22, and the third voice data. 3 Acquire the fourth voice data by collecting sound using the microphone 23 (step S105). The CPU 101 uses the third audio data and the fourth audio data to generate audio data in the middle frequency band having directivity in the direction in which the target device 5 exists (step S106). The CPU 101 generates audio data in a medium frequency band having directivity in the direction in which the target device 5 exists, for example, by beamforming processing using the third audio data and the fourth audio data.

The CPU 101 uses the acquired voice data in the middle frequency band to determine whether or not there is an abnormality in the target device 5 (step S107). The CPU 101 uses, for example, a classifier that identifies the presence or absence of an abnormality, which is generated by machine learning using voice data in the middle frequency band during normal operation of the target device 5. By using this classifier, the CPU 101 determines whether or not there is an abnormality in the target device 5 by using the acquired voice data in the middle frequency band. If there is no abnormality (NO in step S107), the CPU 101 ends the process.

If there is an abnormality (YES in step S107), the CPU 101 notifies the failure detection and parts arrangement (step S108). The CPU 101 may notify that an abnormality has been detected in the middle frequency band.

Then, the CPU 101 shifts to the first mode for detecting the abnormality of the target device 5 by using the voice in the low frequency band, acquires the fifth voice data by collecting the sound using the first microphone 21, and the fifth voice data. 3 Acquire the sixth voice data by collecting sound using the microphone 23 (step S109). The CPU 101 uses the fifth voice data and the sixth voice data to generate low frequency band voice data having directivity in the direction in which the target device 5 exists (step S110). The CPU 101 generates, for example, low frequency band audio data having directivity in the direction in which the target device 5 exists by beamforming processing using the fifth audio data and the sixth audio data.

The CPU 101 uses the acquired voice data in the low frequency band to determine whether or not there is an abnormality in the target device 5 (step S111). The CPU 101 uses, for example, a classifier that identifies the presence or absence of an abnormality, which is generated by machine learning using voice data in a low frequency band during normal operation of the target device 5. By using this classifier, the CPU 101 determines whether or not there is an abnormality in the target device 5 by using the acquired voice data in the low frequency band. If there is no abnormality (NO in step S111), the CPU 101 ends the process.

If there is an abnormality (YES in step S111), the CPU 101 notifies the failure state and the execution of component replacement (step S112). The CPU 101 may notify that an abnormality has been detected in the low frequency band.

As described above, the sound for each frequency band having sharp directivity can be acquired by inexpensive hardware, and the abnormality of the target device 5 can be detected with high accuracy by using the sound. In the above-mentioned example, the presence / absence of one type of abnormality is determined by using the voice of one target frequency band, but the presence / absence of one type of abnormality is determined by using the voice of a plurality of target frequency bands. May be determined.

FIG. 13 shows an example in which the presence or absence of one type of abnormality is determined using voices in a plurality of frequency bands. For example, the abnormality detection unit 36 determines whether or not there is an abnormality in the gear of the target device 5 by using the voice in the low frequency band and the voice in the middle frequency band. The abnormality detection unit 36 determines whether or not there is an abnormality in the bearing of the target device 5 by using the sound in the middle frequency band and the sound in the high frequency band. The abnormality detection unit 36 determines whether or not there is an abnormality in the motor of the target device 5 by using the voice in the low frequency band and the voice in the high frequency band. Further, the abnormality detection unit 36 determines whether or not there is an abnormality in the system of the target device 5 by using the sound in the low frequency band, the sound in the middle frequency band, and the sound in the high frequency band.

The abnormality detection unit 36 switches the type of abnormality (failure mode) to be determined on a time-division basis (for example, every minute). The abnormality detection unit 36 sets, for example, a failure mode for determining a system abnormality, a failure mode for determining a motor abnormality, a failure mode for determining an bearing abnormality, and a failure mode for determining a gear abnormality. By switching by division, the abnormality of each failure mode (type) is determined in order.

In such a case, an example of the procedure of the abnormality detection process executed by the electronic device 1 will be described with reference to the flowchart of FIG. Here, it is assumed that the electronic device 1 is set to a certain failure mode for detecting an abnormality (failure) at the start of the abnormality detection process.

First, in the set failure mode, the CPU 101 acquires voice data by collecting sound using a microphone corresponding to the failure mode (step S21). The CPU 101 may, for example, have first sound data collected by using the first microphone 21, second sound data collected by using the second microphone 22, and third sound collected by using the third microphone 23. Get at least two of the data.

The CPU 101 uses at least two acquired audio data to generate audio data in one or more frequency bands corresponding to the current failure mode, which has directivity in the direction in which the target device 5 exists (step). S22). For example, the CPU 101 generates audio data in a high frequency band having directivity in the direction in which the target device 5 exists by beamforming processing using the first audio data and the second audio data. The CPU 101 generates audio data in a medium frequency band having directivity in the direction in which the target device 5 exists by beamforming processing using the second audio data and the third audio data. Further, the CPU 101 generates audio data in a low frequency band having directivity in the direction in which the target device 5 exists by beamforming processing using the first audio data and the third audio data.

The CPU 101 determines whether or not there is an abnormality in the target device 5 by using the audio data of one or more frequency bands corresponding to the current failure mode among the generated audio data for each frequency band (step S23).

For example, in the failure mode for detecting an abnormality in the system shown in FIG. 13, the CPU 101 generates audio data in the low frequency band, the medium frequency band, and the high frequency band, and uses the audio data to generate the target device. It is determined whether or not there is an abnormality in the system of 5. The CPU 101 uses, for example, a discriminator that identifies the presence or absence of an abnormality in the system, which is generated by machine learning using voice data of each frequency band during normal operation of the target device 5. By using this classifier, the CPU 101 determines whether or not there is an abnormality in the system by using the acquired voice data of the three frequency bands.

Further, in the failure mode for detecting an abnormality of the gear, the CPU 101 generates audio data in the low frequency band and audio data in the middle frequency band, and uses the audio data to provide the gear provided in the target device 5. Determine if there is any abnormality in. The CPU 101 uses, for example, a discriminator that identifies the presence or absence of a gear abnormality, which is generated by machine learning using audio data in the low frequency band and audio data in the medium frequency band during normal operation of the target device 5. .. By using this classifier, the CPU 101 determines whether or not there is an abnormality in the gear by using the acquired voice data in the low frequency band and the voice data in the middle frequency band.

If there is an abnormality (YES in step S23), the CPU 101 notifies the abnormality corresponding to the current failure mode and the countermeasures thereof (step S24). The CPU 101 can realize this notification by displaying the display 105A on the screen or transmitting information to the terminal used by the administrator via the communication interface 106.

On the other hand, if there is no abnormality (NO in step S23), the procedure in step S24 is skipped.

Next, the CPU 101 determines whether or not the first time has elapsed after shifting to the current failure mode (step S25). If the first time has not elapsed (NO in step S25), the process returns to step S21. As a result, the abnormality detection process in the current failure mode is continued.

When the first time has elapsed (YES in step S25), the CPU 101 shifts to another failure mode (step S26). Then, the CPU 101 sets the microphone and the frequency band used for detecting the abnormality in the failure mode after the transition (step S27), and returns to the procedure of step S21. As a result, the abnormality detection process in the failure mode after the transition is started.

Therefore, the CPU 101 can execute a plurality of abnormality detection processes corresponding to a plurality of failure modes having different contents of the abnormalities to be determined in a time division every first hour. The content of the anomaly can be defined, for example, as a type, part, or degree of anomaly, or a combination thereof. For a plurality of failure modes, a mode for determining the presence or absence of a certain type of abnormality using the voice of one target frequency band and a mode for determining the presence or absence of another type of abnormality using the voice of a plurality of target frequency bands Modes to be used may be mixed.

With the above configuration, it is possible to acquire sound for each frequency band having sharp directivity with inexpensive hardware. The electronic device 1 of the present embodiment is, for example, the operation of the target device 5 with three microphones, three directional patterns (three target frequencies), switching target frequencies, sharp directivity, and a high SN ratio. Can collect sounds. Then, the electronic device 1 can determine the presence or absence of an abnormality in the target device 5 by using such an operating sound. Therefore, the electronic device 1 can acquire the sound for each frequency band having a sharp directivity with respect to the direction of the target device 5 with an inexpensive hardware configuration, and accurately detects the abnormality of the target device 5 from the sound. be able to.

(Second Embodiment)
FIG. 15 shows an example of the functional configuration of the system according to the second embodiment. This system can also be realized as an electronic device 1 as in the first embodiment. Further, the electronic device 1 may have the appearance, configuration, function and the like described above in the first embodiment. Hereinafter, the differences from the electronic device 1 of the first embodiment will be mainly described.

The microphones 21, 22, and 23 provided in the electronic device 1 of the second embodiment have a function of switching the frequency characteristics of the signal to be collected. As shown in FIG. 15, the microphones 21, 22, and 23 are provided with microphone setting units 21A, 22A, and 23A, respectively, for switching the frequency characteristics of the signal to be collected. Each microphone 21, 22, 23 has a plurality of modes having different sound collecting characteristics. The microphone setting units 21A, 22A, and 23A can switch the mode used by the corresponding microphones 21, 22, and 23. The plurality of modes include, for example, a low frequency mode suitable for collecting sound in a low frequency band centered on a low frequency (pole) and a high frequency mode suitable for collecting sound in a high frequency band centered on a high frequency. Including.

Further, the abnormality detection program 102B executed on the electronic device 1 is used in the beamforming (BF) control unit 31, the beamforming processing unit 32, the recording unit 33, the calculation unit 34, the display control unit 35, and the abnormality detection unit 36. In addition, it includes a microphone mode control unit 37. The beamforming control unit 31 outputs a target frequency for determining the presence or absence of an abnormality in the target device 5 to the microphone mode control unit 37. The microphone mode control unit 37 determines the frequency characteristics of the signal collected by the microphones 21, 22, 23 for the microphone setting units 21A, 22A, 23A of each microphone 21, 22, 23 according to the target frequency. Request to switch.

For example, FIG. 16 shows a case where the microphones 21, 22, and 23 have a function of switching between the low frequency mode 61 and the high frequency mode 62. The microphones 21, 22, and 23 set in the low frequency mode 61 can collect the sound in the low frequency band with high power, that is, with high sensitivity. On the other hand, the microphones 21, 22, 23 set in the high frequency mode 62 can collect the sound in the high frequency band with high power, that is, with high sensitivity.

In this case, the microphone mode control unit 37 is set to either the low frequency mode 61 or the high frequency mode 62 with respect to the microphone setting units 21A, 22A, and 23A of the microphones 21, 22, and 23, depending on the target frequency. Request to set. The microphone setting units 21A, 22A, and 23A set the corresponding microphones 21, 22, and 23 in either the low frequency mode 61 or the high frequency mode 62 in response to this request.

FIG. 17 shows an example in which a first mode, a second mode, and a third mode are defined in order to detect three types of abnormalities.

In the first mode, the beamforming control unit 31 and the beamforming processing unit 32 collect sound using the first microphone 21 and the third microphone 23, each of which has a distance between microphones of 3 cm and is set to the low frequency mode. Using the generated audio data, audio data in a frequency band (low frequency band) of 10 kHz having directivity toward the target device 5 is generated. The abnormality detection unit 36 determines the presence or absence of an abnormality in the target device 5 using the generated voice data. Then, if there is an abnormality, notification of the failure state and implementation of parts replacement are urged.

In the second mode, the beamforming control unit 31 and the beamforming processing unit 32 use the second microphone 22 and the third microphone 23, each of which has a distance between microphones of 1.8 cm and is set to the high frequency mode. Using the collected audio data, audio data in a frequency band (medium frequency band) of 15 kHz having directivity toward the target device 5 is generated. The abnormality detection unit 36 determines the presence or absence of an abnormality in the target device 5 using the generated voice data. Then, if there is an abnormality, notification of failure detection and parts arrangement are urged.

In the third mode, the beamforming control unit 31 and the beamforming processing unit 32 use the first microphone 21 and the second microphone 22, which have a distance between microphones of 1.2 cm and are set to the high frequency mode, respectively. Using the collected audio data, audio data in a 20 kHz frequency band (high frequency band) having directivity toward the target device 5 is generated. The abnormality detection unit 36 determines the presence or absence of an abnormality in the target device 5 using the generated voice data. Then, if there is an abnormality, notification of a failure sign and budgeting for parts replacement are promoted.

By setting each microphone 21, 22, 23 to a frequency mode according to the target frequency, the power of the frequency component of the target frequency included in the sound collected by the microphones 21, 22, 23 is increased. Therefore, it is possible to determine with high accuracy whether or not there is an abnormality in the target device 5. It is preferable that the two or more microphones 21, 22, 23 used in one mode are all set to the same mode, which is either the low frequency mode or the high frequency mode. This is a sound suitable for determining the presence or absence of an abnormality by using the sound collected by the microphones 21, 22, 23 set in the same mode (for example, a sound having a sharp directivity to the target sound). This is because it is easy to obtain.

The flowchart of FIG. 18 shows an example of the procedure of the abnormality detection process executed by the electronic device 1. In this abnormality detection process, at the beginning of the procedure corresponding to the first mode, the second mode, and the third mode, the microphones 21, 22, and 23 are placed in either the low frequency mode or the high frequency mode according to the target frequency. Contains steps to set to.

More specifically, at the beginning of the procedure corresponding to the third mode, the CPU 101 sets the first microphone 21 and the second microphone 22 to the high frequency mode, respectively, according to the target frequency of 20 kHz (step S301). ). Subsequent procedures S302 to S305 of the third mode are the same as the procedures S101 to S104 described above with reference to the flowchart of FIG.

Then, at the beginning of the procedure corresponding to the second mode, the CPU 101 sets the second microphone 22 and the third microphone 23 to the high frequency mode, respectively, according to the target frequency of 15 kHz (step S306). Subsequent steps S307 to S310 of the second mode are the same as the steps S105 to S108 described above with reference to the flowchart of FIG.

Further, at the beginning of the procedure corresponding to the first mode, the CPU 101 sets the first microphone 21 and the third microphone 23 to the low frequency mode, respectively, according to the target frequency of 10 kHz (step S311). Subsequent steps S312 to S315 of the first mode are the same as the steps S109 to S112 described above with reference to the flowchart of FIG.

As described above, according to the present embodiment, it is possible to acquire sound for each frequency band having sharp directivity with inexpensive hardware. The system is a system for collecting the sound emitted by the target device 5, and is a system for collecting the sound emitted by the target device 5, the first microphone 21, the second microphone 22 which is the first distance away from the first microphone 21, and the second microphone 22 and the second distance. A third microphone 23 that is separated and separated from the first microphone 21 and a third distance is provided. The directivity of sound collection in the first frequency band using the first microphone 21 and the second microphone 22 is the first direction. The directivity of sound collection in the second frequency band using the second microphone 22 and the third microphone 23 is the second direction. The directivity of sound collection in the third frequency band using the first microphone 21 and the third microphone 23 is the third direction. The position of the target device 5 is the direction from the first microphone 21 and the second microphone 22 in the first direction, the direction from the second microphone 22 and the third microphone 23 in the second direction, and the first microphone 21. The positional relationship between the first microphone 21, the second microphone 22, the third microphone 23, and the target device 5 is determined so as to be in the third direction from the third microphone 23. The second distance is greater than the first distance. The third distance is greater than the second distance. The second frequency band is lower than the first frequency band. The third frequency band is lower than the second frequency band.

As a result, the sound of the first frequency band, the sound of the second frequency band, and the sound of the third frequency band, which have sharp directivity to the target device 5, can be acquired by the three microphones, respectively.

Further, each of the various functions described in the present embodiment may be realized by a circuit (processing circuit). Examples of processing circuits include programmed processors such as central processing units (CPUs). This processor executes each of the described functions by executing a computer program (instruction group) stored in the memory. This processor may be a microprocessor including an electric circuit. Examples of processing circuits also include digital signal processors (DSPs), application specific integrated circuits (ASICs), microcontrollers, controllers, and other electrical circuit components. Each of the components other than the CPU described in this embodiment may also be realized by a processing circuit.

Further, since various processes of the present embodiment can be realized by a computer program, the present embodiment and the present embodiment can be obtained only by installing and executing the computer program on a computer through a computer-readable storage medium in which the computer program is stored. A similar effect can be easily achieved.

Although some embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other embodiments, and various omissions, replacements, and changes can be made without departing from the gist of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are also included in the scope of the invention described in the claims and the equivalent scope thereof.

1 ... Electronic device, 11 ... Board, 12, 13, 14 ... Opening, 15 ... Processor unit, 16 ... Memory unit, 21, 22, 23 ... Microphone, 101 ... CPU, 102 ... Main memory, 102A ... OS, 102B ... Abnormality detection program, 103 ... Non-volatile memory, 104 ... Audio controller, 105 ... Display controller, 105A ... Display, 106 ... Communication interface, 31 ... Beam forming control unit, 32 ... Beam forming processing unit, 33 ... Recording unit, 34 ... Calculation unit, 35 ... Display control unit, 36 ... Abnormality detection unit, 5 ... Target device.

Claims (17)

A system for collecting sounds emitted from an object,
With the first microphone
The first microphone and the second microphone separated by the first distance,
A third microphone separated from the second microphone by a second distance and separated from the first microphone by a third distance is provided.
The directivity of sound collection in the first frequency band using the first microphone and the second microphone is the first direction.
The directivity of sound collection in the second frequency band using the second microphone and the third microphone is the second direction.
The directivity of sound collection in the third frequency band using the first microphone and the third microphone is the third direction.
The position of the object is
The direction from the first microphone and the second microphone in the first direction.
From the second microphone and the third microphone in the second direction.
From the first microphone and the third microphone so as to be in the direction of the third direction.
The positional relationship between the first microphone, the second microphone, the third microphone, and the object is determined.
The second distance is larger than the first distance,
The third distance is larger than the second distance,
A system in which the second frequency band is lower than the first frequency band and the third frequency band is lower than the second frequency band. A system for collecting sounds emitted from an object,
With the first microphone
The first microphone and the second microphone separated by the first distance,
A third microphone separated from the second microphone by a second distance and separated from the first microphone by a third distance is provided.
The directivity of sound collection in the first frequency band using the first microphone and the second microphone, and the directivity of sound collection in the second frequency band using the second microphone and the third microphone. The directivity of sound collection in the third frequency band using the first microphone and the third microphone is substantially the same in the first direction.
The second distance is larger than the first distance,
The third distance is larger than the second distance,
A system in which the second frequency band is lower than the first frequency band and the third frequency band is lower than the second frequency band. Claim 1 or claim further comprising a detection means for detecting an abnormality of the object by using at least one sound collection result of the first frequency band, the second frequency band, and the third frequency band. The system according to 2. When the detection means detects an abnormality of the object by using the sound collection result of any one of the first to third frequency bands, the detection means detects the abnormality as the first type of the object. And
When an abnormality of the object is detected using a sound collection result different from any one of the first to third frequency bands, the object is different from the first type of the object. The system according to claim 3, which detects an abnormality of the second type. The detection means
When an abnormality is first detected in the first frequency band of the first to third frequency bands and then an abnormality is detected in the second or third frequency band, the third type of the object is described. Detected as abnormal,
Claim that when an abnormality is first detected in the second or third frequency band of the first to third frequency bands, it is detected as a fourth type abnormality different from the third type of the object. The system according to 4. The system according to claim 5, wherein at least one of the first to fourth types shows any one of a sign of failure, a failure state, and a difference in the length of remaining life. At least one of the first to fourth types of abnormalities is a failure of the first member of the object, a failure of the second member of the object, a cause of failure of the first member, and a failure of the first member. The system according to claim 5, which shows any one of the differences in causes. The invention according to any one of claims 1 to 7, further comprising a setting means for setting at least two microphones among the first to third microphones to a high frequency band mode or a low frequency band mode. System. The sound collected by the first microphone, the sound collected by the second microphone, the sound collected by the third microphone, the sound collection result of the first frequency band, and the sound of the second frequency band. The system according to any one of claims 1 to 8, further comprising a recording means for storing at least one of the sound collection result and the sound collection result of the third frequency band. The power of the sound collected by at least one of the first to third microphones and the power and spectrum of a specific frequency band for the sound collection result of one of the first to third frequency bands. The system according to any one of claims 1 to 9, further comprising a display means for displaying at least one of them on a screen. By a method using a first microphone, a second microphone that is first distance away from the first microphone, and a third microphone that is second distance away from the second microphone and third distance away from the first microphone. There,
The directivity of sound collection in the first frequency band using the first microphone and the second microphone is set as the first direction.
The directivity of sound collection in the second frequency band using the second microphone and the third microphone is set as the second direction.
The directivity of sound collection in the third frequency band using the first microphone and the third microphone is set as the third direction.
The position of the object that emits sound,
The direction from the first microphone and the second microphone in the first direction.
From the second microphone and the third microphone in the second direction.
From the first microphone and the third microphone so as to be in the direction of the third direction.
The positional relationship between the first microphone, the second microphone, the third microphone, and the object is determined.
The second distance is larger than the first distance,
The third distance is larger than the second distance,
The second frequency band is lower than the first frequency band, and the third frequency band is lower than the second frequency band.
A method of detecting the sound of the object using at least two of the first microphone, the second microphone, and the third microphone. It is a method of using the first microphone, the second microphone that is first distance away from the first microphone, and the third microphone that is second distance away from the second microphone and third distance away from the first microphone. hand,
The directivity of sound collection in the first frequency band using the first microphone and the second microphone, and the directivity of sound collection in the second frequency band using the second microphone and the third microphone. The directivity of sound collection in the third frequency band using the first microphone and the third microphone is set to substantially the same first direction.
The second distance is larger than the first distance,
The third distance is larger than the second distance,
The second frequency band is lower than the first frequency band, and the third frequency band is lower than the second frequency band.
Detecting the sound of Target product using at least two of said first microphone and said second microphone third microphone, methods. The 11th or 12th claim further includes detecting an abnormality of the object by using at least one sound collection result of the first frequency band, the second frequency band, and the third frequency band. The method described. When an abnormality of the object is detected using the sound collection result of any one of the first to third frequency bands, it is detected as an abnormality of the first type of the object.
When an abnormality of the object is detected using a sound collection result different from any one of the first to third frequency bands, the object is different from the first type of the object. The method according to claim 13, further comprising detecting an abnormality of the second type. When an abnormality is first detected in the first frequency band of the first to third frequency bands and then an abnormality is detected in the second or third frequency band, the third type of the object is described. Detected as abnormal,
When an abnormality is first detected in the second or third frequency band of the first to third frequency bands, it is detected as a fourth type abnormality different from the third type of the object. The method of claim 14, further comprising. Using the first microphone, the second microphone that is first distance away from the first microphone, and the third microphone that is second distance away from the second microphone and third distance away from the first microphone. A system for collecting the sound emitted from an object,
The directivity of sound collection in the first frequency band using the first microphone and the second microphone is the first direction.
The directivity of sound collection in the second frequency band using the second microphone and the third microphone is the second direction.
The directivity of sound collection in the third frequency band using the first microphone and the third microphone is the third direction.
The position of the object is
The direction from the first microphone and the second microphone in the first direction.
From the second microphone and the third microphone in the second direction.
From the first microphone and the third microphone so as to be in the direction of the third direction.
The positional relationship between the first microphone, the second microphone, the third microphone, and the object is determined.
The second distance is larger than the first distance,
The third distance is larger than the second distance,
A system in which the second frequency band is lower than the first frequency band and the third frequency band is lower than the second frequency band. Using the first microphone, the second microphone that is first distance away from the first microphone, and the third microphone that is second distance away from the second microphone and third distance away from the first microphone. A system for collecting the sound emitted from an object,
The directivity of sound collection in the first frequency band using the first microphone and the second microphone, and the directivity of sound collection in the second frequency band using the second microphone and the third microphone. The directivity of sound collection in the third frequency band using the first microphone and the third microphone is substantially the same in the first direction.
The second distance is larger than the first distance,
The third distance is larger than the second distance,
A system in which the second frequency band is lower than the first frequency band and the third frequency band is lower than the second frequency band.

Images