Abnormal sound detection system and abnormal sound detection method

Abstract

To highly accurately detect an abnormal sound under an operation of a moving body emitting an abnormal sound.SOLUTION: An abnormal sound detection system is configured by: a microphone that measures a surrounding sound; a sound propagation characteristic identification unit that identifies a sound propagation characteristic between the moving body and the microphone on the basis of a reference sound reproduced from reference sound information on a reference sound a moving body emits, and a sound obtained by measuring the reference sound the moving body emits with the microphone; an estimation operation sound generation unit that generates an estimation operation sound estimating a sound obtained by measuring an operation sound the moving body emits with the microphone on the basis of an operation sound reproduced from operation sound information on an operation sound the moving body emits and the sound propagation characteristic; and an abnormal sound detection unit that cancels the estimation operation sound from the sound obtained by measuring with the microphone, and detects an abnormal sound from a remaining sound.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to an abnormal sound detection system and an abnormal sound detection method.

Background Art

[0002] Conventionally, an abnormal sound detection system that detects the occurrence of an abnormal situation based on the occurrence of an abnormal sound has been used. The abnormal sound detection system continuously measures the sound in a warning area to be detected for an abnormal situation by a microphone, and detects the occurrence of an abnormal situation based on the measurement of an abnormal sound that is not normally measured. In the warning area, a moving body such as a drone is often patrolled and used for warning activities. If the operation sound of the moving body is large, it may affect the detection of abnormal sounds. In order to detect abnormal sounds in an environment where a moving body is used, it is desirable to reduce the operation sound of the moving body or cancel the operation sound of the moving body from the measured sound by the microphone.

[0003] For example, Patent Document 1 discloses a noise reduction technique for reducing the operation sound of a drone. By providing a noise reduction unit using a resonator corresponding to each of the rotating blades of the drone, the noise caused by the rotation of the rotating blades is reduced. For example, Patent Document 2 discloses a noise cancellation technique using sound interference. The noise is measured by a sensor arranged near the noise source, and a cancellation sound having an opposite phase to the obtained noise is generated. By arranging a plurality of speakers inside and in the vicinity of the device serving as the noise source and reproducing the cancellation sound, a part of the noise is canceled and the noise is reduced.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, technologies such as attaching a noise-canceling unit to a mobile device, as described in Patent Document 1, lead to an increase in the size and weight of the mobile device. Adopting larger rotors or motors to compensate for the increased weight of the mobile device could create a vicious cycle of increased noise. To achieve effectiveness with the noise cancellation technology described in Patent Document 2, it is necessary to accurately measure the operating sound of the mobile device and generate a cancellation sound with the opposite phase. Microphones that detect abnormal sounds in a warning area measure ambient noise. Since the measured sound includes many sounds other than the operating sound of the mobile device, it is difficult to accurately obtain the operating sound of the mobile device from the measured sound.

[0006] This disclosure has been made in view of the prior art, including the problems described above, and one of its purposes is to provide an abnormal sound detection system and an abnormal sound detection method that can detect abnormal sounds with high accuracy even when a mobile device is in operation. [Means for solving the problem]

[0007] The abnormal sound detection system according to this disclosure includes a microphone for measuring ambient sound, an acoustic propagation characteristic identification unit for identifying acoustic propagation characteristics between the moving body and the microphone based on a reference sound reproduced from reference sound information relating to a reference sound emitted by the moving body and a sound obtained by measuring the reference sound emitted by the moving body with the microphone, an estimated operating sound generation unit for generating an estimated operating sound by estimating a sound obtained by measuring the operating sound emitted by the moving body with the microphone based on an operating sound reproduced from operating sound information relating to an operating sound emitted by the moving body and the acoustic propagation characteristics, and an abnormal sound detection unit for canceling the estimated operating sound from the sound obtained by measuring with the microphone and detecting abnormal sounds from the remaining sound.

[0008] In the above configuration, the sound propagation characteristic identification unit identifies sound propagation characteristics that include conversion information to convert the time waveform of a reference sound emitted by the moving body into a time waveform of a sound obtained by measuring the reference sound emitted by the moving body with the microphone; the estimated operation sound generation unit generates the estimated operation sound time waveform by converting the time waveform of the operation sound emitted by the moving body reproduced from the operation sound information using the conversion information; and the abnormal sound detection unit may detect abnormal sounds from the time waveform of the sound remaining after canceling the time waveform of the estimated operation sound from the time waveform of the sound obtained by measuring with the microphone.

[0009] In the above configuration, the system further includes a communication unit that receives mobile body information from the mobile body, including information regarding the rotational speed of a drive unit included in the mobile body, and the estimated operating sound generation unit may generate the estimated operating sound from the operating sound corresponding to the rotational speed of the drive unit identified based on the mobile body information, by referring to the operating sound information which includes information relating the rotational speed of the drive unit to the operating sound emitted by the mobile body.

[0010] In the above configuration, the sound propagation characteristic identification unit may update the sound propagation characteristics by measuring the reference sound emitted by the moving body at predetermined time intervals using the microphone, the estimated operation sound generation unit may update the estimated operation sound using the updated sound propagation characteristics, and the abnormal sound detection unit may detect an abnormal sound from the remaining sound after canceling the updated estimated operation sound from the sound measured by the microphone.

[0011] The abnormal sound detection method according to this disclosure is an abnormal sound detection method executed by an abnormal sound detection system that detects abnormal sounds, and includes the steps of: measuring a reference sound emitted by a moving body with a microphone; identifying acoustic propagation characteristics between the moving body and the microphone based on a reference sound reproduced from pre-prepared reference sound information and a sound obtained by measuring the reference sound with the microphone; generating an estimated operating sound by estimating a sound obtained by measuring the operating sound emitted by the moving body with the microphone based on the operating sound of the moving body reproduced from pre-prepared operating sound information and the acoustic propagation characteristics; and canceling the estimated operating sound from the sound obtained by measuring with the microphone and detecting an abnormal sound from the remaining sound. [Effects of the Invention]

[0012] According to the abnormal sound detection system and abnormal sound detection method described herein, abnormal sounds can be detected with high accuracy even under the operation of a moving object, while suppressing the influence of the operating noise of the moving object. [Brief explanation of the drawing]

[0013] [Figure 1] Figure 1 is a diagram illustrating the overview of the abnormal sound detection system according to this embodiment. [Figure 2] Figure 2 is a diagram illustrating the estimated operating sound generation process performed by the abnormal sound detection device. [Figure 3] Figure 3 is a diagram illustrating the abnormal sound detection process performed by the abnormal sound detection device. [Figure 4] Figure 4 is a block diagram showing an example configuration of an abnormal sound detection device. [Figure 5] Figure 5 is a flowchart showing an example of the process performed by the abnormal sound detection device. [Figure 6] Figure 6 shows an example of how the abnormal sound detection system can be used. [Figure 7] Figure 7 is a diagram illustrating the acoustic propagation characteristics. [Figure 8] Figure 8 is a diagram illustrating the sound obtained by the abnormal sound detection device.

Best Mode for Carrying Out the Invention

[0014] Hereinafter, embodiments of an abnormal sound detection system and an abnormal sound detection method according to the present disclosure will be described while referring to the accompanying drawings. FIG. 1 is a diagram for explaining the outline of an abnormal sound detection system 1 according to the present embodiment.

[0015] The abnormal sound detection system 1 detects an abnormal sound by an abnormal sound detection device 10 from the sound measured by a microphone 20 in an environment where a moving body 100 is operated. The moving body 100 includes an unmanned aerial vehicle that monitors the ground while flying in the air and an unmanned vehicle that monitors the surroundings while traveling on the ground. The type of the moving body 100 is not particularly limited, but hereinafter, a multi-copter type unmanned aerial vehicle called a drone will be continued to be described as an example. The moving body 100 moves in the air by rotating rotary wings attached to a plurality of motors respectively. While moving in the air, an operation sound corresponding to the rotational speed of the motor is generated from the moving body 100.

[0016] [Specification of Acoustic Propagation Characteristics] The moving body 100 includes a speaker and reproduces a preset reference sound with the speaker. The abnormal sound detection device 10 measures the reference sound emitted from the moving body 100 with the microphone 20 and specifies the acoustic propagation characteristics using the obtained sound (S11).

[0017] The reference sound is a sound whose frequency, sound pressure, and reproduction method are preset so as not to be buried in other sounds measured by the microphone 20. That is, a sound measurable by the microphone 20 is set as the reference sound without being buried in the operation sound of the moving body 100 or other sounds measured at the location where the microphone 20 is installed.

[0018] For example, a sound with a predetermined sound pressure at a predetermined frequency is set as a reference sound. The reference sound emitted by the moving body 100 changes under the influence of various influencing factors such as obstacles and climate before reaching the microphone 20. For example, due to effects such as reflection, absorption, diffraction, and attenuation, the sound obtained by the microphone 20 measuring the reference sound may be different from the reference sound emitted by the moving body 100. Also, for example, when the moving body 100 emits a reference sound while moving, the sound measured by the microphone 20 is affected by the Doppler shift according to the relative speed between the moving body 100 and the microphone 20. The abnormal sound detection device 10 identifies the acoustic propagation characteristics between the moving body 100 and the microphone 20 by comparing a preset reference sound with the sound actually measured by the microphone 20.

[0019] The acoustic propagation characteristics are information necessary to identify the sound measured by the microphone 20 from the sound emitted by the moving body 100. For example, the acoustic propagation characteristics are information that can obtain the time waveform of the sound measured by the microphone 20 from the time waveform of the sound emitted by the moving body 100. The time waveform of the sound is, for example, a waveform representing the sound pressure on the vertical axis and time on the horizontal axis, and includes information on frequency and phase. By performing a Fourier transform on the time waveform, the frequency and phase of the sound indicated by the time waveform can be obtained.

[0020] Since the acoustic propagation characteristics can be obtained using conventional techniques known in the fields of signal processing and acoustic processing, detailed description is omitted. For example, a transfer function can be obtained by setting the time waveform of the reference sound as the input signal and the time waveform of the measured sound obtained by measuring the reference sound with the microphone 20 as the output signal, and this can be used as the acoustic propagation characteristics.

[0021] When obtaining the transfer function, a sound unrelated to the reference sound may be canceled in advance from the measured sound of the microphone 20 as noise. Since the reference sound is a sound set to be distinguishable from other sounds measured by the microphone 20, it can be extracted relatively easily by canceling other sounds as noise. Since noise cancellation techniques are conventionally known, detailed description is omitted. For example, sounds in a frequency range unrelated to the reference sound may be canceled by filtering.

[0022] The abnormal sound detection device 10 utilizes acoustic propagation characteristics to estimate the sound obtained by measuring the operating sound of the mobile body 100 with the microphone 20. As long as acoustic propagation characteristics that allow estimation of the operating sound of the mobile body 100 can be obtained, the content and reproduction method of the reference sound emitted by the mobile body 100 are not particularly limited.

[0023] The reference tone may be a single-frequency sound, a discrete sound containing multiple frequencies, or a sound in a predetermined frequency range consisting of multiple frequencies. In order to obtain acoustic propagation characteristics that show the relationship between the frequency, amplitude, and phase of the operating sound emitted by the mobile body 100 and the sound measured by the microphone 20, it is preferable to reproduce reference tones of multiple frequencies corresponding to the operating sound of the mobile body 100. For example, if the mobile body 100 emits an operating sound with a frequency characteristic having peaks at 1kHz and 2kHz, a single-frequency sound of 1.5kHz, which is close to the peak frequency of the operating sound and not buried by the operating sound, can be used as the reference tone. For example, multiple frequencies may be set as reference tones in the frequency range that includes the peak frequency of the operating sound, such as 800Hz and 1.2kHz corresponding to the peak frequency of 1kHz, and 1.8kHz and 2.2kHz corresponding to the peak frequency of 2kHz. Alternatively, a reference tone may be set in a frequency range that includes the peak frequency of the operating sound, such as 500Hz to 2.5kHz.

[0024] If the reference tone consists of sounds of multiple frequencies, each sound may be played simultaneously or one by one in sequence. If the reference tone is a sound in a predetermined frequency range, the reference tone may be played while sweeping through the frequency range. The reference tone may be played only once, or it may be played repeatedly at predetermined time intervals or at predetermined distance intervals corresponding to the distance traveled by the mobile body 100. Depending on the placement of the microphone 20, one or more positions for playing the reference tone may be set in advance, and the reference tone may be played when the mobile body 100 passes through the set positions. The reference tone may be played while the mobile body 100 is moving through the air, or while it is stopped in the air. To suppress the effects of Doppler shift, it is preferable that the reference tone be played when the mobile body 100 is stopped or moving at a low speed. The content and playback method of the reference tone may be set appropriately based on the operating sounds emitted by the mobile body 100, background sounds in the environment where the microphone 20 is installed, abnormal sounds to be detected, etc.

[0025] For example, the moving object 100 plays a reference sound multiple times in a row at predetermined time intervals or predetermined travel distance intervals, such as every 5 seconds or every 20 meters. The abnormal sound detection device 10 identifies and updates the acoustic propagation characteristics each time it measures the reference sound. As a result, the abnormal sound detection system 1 can identify the latest acoustic propagation characteristics even when the moving object 100 moves and the acoustic propagation characteristics change.

[0026] [Estimation of the sound of a moving object] The abnormal sound detection device 10 communicates wirelessly with the mobile object 100 to acquire mobile object information. Based on the mobile object information, the abnormal sound detection device 10 estimates the sound obtained by measuring the operating sound of the mobile object 100 with the microphone 20 (S12 in Figure 1). The mobile object information includes information for reconstructing the time waveform of the operating sound emitted by the mobile object 100.

[0027] The mobile device 100, which flies by rotating its rotor blades with a motor, generates different operating sounds depending on the motor's rotation speed. Therefore, the rotation speed of the motor acting as the drive unit and the waveform information of the operating sounds generated at each rotation speed are associated and prepared in advance.

[0028] The motor's rotational speed and the waveform information of the operating sound generated at each rotational speed may be stored in the mobile unit 100's memory. In this case, the mobile unit 100 transmits the waveform information corresponding to the current motor's rotational speed to the abnormal sound detection device 10 as mobile unit information. Upon receiving the mobile unit information, the abnormal sound detection device 10 can reconstruct the time waveform of the operating sound of the mobile unit 100 from the waveform information.

[0029] The motor's rotational speed and the waveform information of the operating sound at each rotational speed may be stored in the memory of the abnormal sound detection device 10 or the server device. In this case, the mobile body 100 transmits information indicating the motor's rotational speed to the abnormal sound detection device 10 as mobile body information. Upon receiving the mobile body information, the abnormal sound detection device 10 refers to the memory based on the rotational speed and can reconstruct the time waveform of the operating sound of the mobile body 100 from the obtained waveform information.

[0030] The mobile unit 100 may be equipped with a microphone and transmit the time waveform of the operating sound measured by the microphone to the abnormal sound detection device 10 as waveform information. In this case, the abnormal sound detection device 10 can use the received waveform information of the operating sound as is. Alternatively, the mobile unit 100 may transmit the sound data recorded by the microphone directly to the abnormal sound detection device 10.

[0031] If multiple mobile bodies 100 are used, the motor rotation speed and waveform information for each rotation speed can be associated with identification information to uniquely identify each mobile body 100. If a mobile body 100 transmits the identification information and rotation speed included in the mobile body information, the abnormal sound detection device 10 can identify the mobile body 100 based on the identification information and acquire the waveform information of this mobile body 100.

[0032] The abnormal sound detection device 10, which reproduces the time waveform of the operating sound of the mobile object 100 based on the mobile object information, uses the previously identified acoustic propagation characteristics to generate a time waveform of the estimated operating sound, which is the operating sound of the mobile object 100 measured by the microphone 20.

[0033] Figure 2 is a diagram illustrating the estimated operating sound generation process performed by the abnormal sound detection device 10. As shown in Figure 2, the acoustic propagation characteristics are determined from the relationship between the time waveform of a preset reference sound Sr and the time waveform of a measured sound S1 obtained by measuring the reference sound Sr emitted by the moving body 100 with the microphone 20.

[0034] The abnormal sound detection device 10 generates an estimated operating sound Se time waveform based on the time waveform of the operating sound Sd of the mobile body 100 reproduced using mobile body information and the acoustic propagation characteristics obtained using a reference sound. The estimated operating sound Se is an estimate of the time waveform of the sound obtained by measuring the operating sound Sd emitted by the mobile body 100 with the microphone 20. For example, if a transfer function that outputs the time waveform of the measured sound S1 by inputting the time waveform of a reference sound Sr is used as the acoustic propagation characteristics, the estimated operating sound Se time waveform can be obtained by inputting the time waveform of the operating sound Sd reproduced from waveform information into the transfer function.

[0035] [Detection of abnormal sounds] The abnormal sound detection device 10 cancels out the estimated operating sound from the sound measured by the microphone 20 and detects the abnormal sound from the remaining sound (S13 in Figure 1). Figure 3 is a diagram illustrating the abnormal sound detection process performed by the abnormal sound detection device 10. As shown in Figure 3, the microphone 20 of the abnormal sound detection device 10 measures the ambient sound Sc and the operating sound Sd of the moving object 100. The ambient sound Sc is normally only the background sound Sb (Sc=Sb), but in the event of an abnormal situation, the ambient sound Sc includes the background sound Sb in addition to the abnormal sound Sa (Sc=Sa+Sb).

[0036] The operating sound Sd of the moving object 100 changes into a different measurement operating sound Sf when measured by the microphone 20 due to the effects of reflection, absorption, diffraction, attenuation, Doppler shift, etc. The measurement sound S2 from the microphone 20 includes the ambient sound Sc and the measurement operating sound Sf (S2 = Sc + Sf). The ambient sound Sc can be obtained by canceling the measurement operating sound Sf from the measurement sound S2 (Sc = S2 - Sf).

[0037] The estimated operating sound Se is the sound obtained by estimating the measured operating sound Sf based on the operating sound Sd of the mobile body 100 and the acoustic propagation characteristics between the mobile body 100 and the microphone 20 (Se ≈ Sf). The abnormal sound detection device 10 uses the estimated operating sound Se to obtain a sound obtained by canceling the measured operating sound Sf from the measured sound S2. Specifically, the abnormal sound detection device 10 uses the sound obtained by canceling the estimated operating sound Se from the measured sound S2 as the sound obtained by canceling the measured operating sound Sf from the measured sound S2. By canceling the estimated operating sound Se, which is a pseudo-measured operating sound obtained by estimating the measured operating sound Sf, from the measured sound S2, ambient sound Sc can be obtained (Sc = S2 - Se). For example, the abnormal sound detection device 10 generates a time waveform of the cancellation sound with the opposite phase from the time waveform of the estimated operating sound Se and combines it with the time waveform of the measured sound S2 to obtain the time waveform of ambient sound Sc.

[0038] If only the measured operating sound Sf can be extracted from the measurement sound S2 of the microphone 20, the measured operating sound Sf can be directly canceled from the measurement sound S2. Furthermore, the acoustic propagation characteristics between the mobile body 100 and the microphone 20 can be identified from the operating sound Sd and the measured operating sound Sf extracted from the measurement sound S2. However, in reality, the measurement sound S2 contains various sounds other than the measured operating sound Sf. Therefore, it can be difficult to extract only the measured operating sound Sf from the measurement sound S2 to cancel it or identify the acoustic propagation characteristics. The abnormal sound detection device 10 can identify the acoustic propagation characteristics using a reference sound Sr that can be easily extracted from the measurement sound S1 of the microphone 20, and thereby generate an estimated operating sound Se, which estimates the measured operating sound Sf from the operating sound Sd and the acoustic propagation characteristics. By using the estimated operating sound Se, the abnormal sound detection device 10 can effectively cancel the measured operating sound Sf from the measurement sound S2.

[0039] To cancel the measured operating sound Sf from the measured sound S2 of the microphone 20, the abnormal sound detection device 10 cancels the estimated operating sound Se from the measured sound S2 to obtain the ambient sound Sc. As shown in Figure 3, the device further detects the abnormal sound Sa by canceling the background sound Sb from the ambient sound Sc. The method for detecting abnormal sounds is not particularly limited, but for example, a sound pressure threshold can be set in advance, and if a loud sound with a sound pressure exceeding the threshold is detected, it can be determined that an abnormal sound has been detected. Alternatively, for example, an AI (Artificial Intelligence) for abnormal sound detection may be used to detect abnormal sounds. Since the AI ​​used for detecting abnormal sounds is conventionally known, a detailed explanation will be omitted, but for example, if an AI that uses machine learning on normal sounds when no abnormal situation is occurring is used to detect abnormal sounds, abnormal sounds can be detected with high accuracy.

[0040] The abnormal sound detection device 10 continuously performs measurements using the microphone 20, and updates the acoustic propagation characteristics each time it detects the reference sound Sr of the mobile body 100, and updates the estimated operating sound Se using the updated acoustic propagation characteristics. Similarly, if the motor rotation speed of the mobile body 100 changes and the operating sound Sd changes, the abnormal sound detection device 10 detects this based on the mobile body information and updates the estimated operating sound Se. The abnormal sound detection device 10 uses the updated estimated operating sound Se to detect abnormal sounds Sa targeting ambient sound Sc obtained by canceling the measured operating sound Sf from the measured sound S2. Even if the acoustic propagation characteristics change or the operating sound Sd of the mobile body 100 changes, abnormal sounds can be detected with high accuracy using the estimated operating sound Se based on the changed acoustic propagation characteristics and operating sound Sd.

[0041] Although Figure 3 does not show the reference tone Sr, if the measurement tone S2 from the microphone 20 includes the sound of the measurement of the reference tone Sr, this sound can be canceled from the measurement tone S2 and the abnormal sound detection process can be executed. For example, similar to the cancellation process for the measurement operation sound Sf, an estimated reference tone can be generated by using the acoustic propagation characteristics and the reference tone Sr to estimate the sound obtained by measuring the reference tone Sr with the microphone 20, and the estimated reference tone can be canceled when an abnormal sound is detected. Alternatively, the process of canceling the estimated reference tone from the sound measured by the microphone 20 can be executed first, and the resulting sound can be used as the measurement tone S2 to execute each of the processes described above.

[0042] An abnormal sound detection device 10 that detects an abnormal sound can perform a notification process to notify the detection result. The notification method is not particularly limited, but for example, the abnormal sound detection device 10 may be equipped with a display device and notify by displaying information indicating an abnormal situation on the display device. The abnormal sound detection device 10 may be equipped with a speaker and notify by playing a sound indicating an abnormal situation through the speaker. The abnormal sound detection device 10 may also notify an external device of the abnormal situation, and the external device may notify the abnormal situation by display or sound.

[0043] [Configuration of the abnormal sound detection device] An example configuration of the abnormal sound detection device 10 will be described. Figure 4 is a block diagram showing an example configuration of the abnormal sound detection device 10. In addition to the microphone 20, the abnormal sound detection device 10 includes a control unit 30, a storage unit 40, and a communication unit 50. The abnormal sound detection device 10 may further include a display unit, an operation unit, a speaker, etc.

[0044] The microphone 20 measures the sound in the alert area, which is set as the target for detecting abnormal situations. The communication unit 50 communicates wirelessly with the mobile unit 100. The communication unit 50 may also communicate with external devices via wired or wireless connection.

[0045] The control unit 30 includes an operating sound identification unit 31, an acoustic propagation characteristic identification unit 32, an estimated operating sound generation unit 33, an estimated operating sound cancellation unit 34, an abnormal sound detection unit 35, and a notification unit 36. The storage unit 40 stores reference sound data 41 and operating sound data 42. The storage unit 40 is also used to store various data necessary for the operation of the control unit 30. The control unit 30 controls each unit while utilizing the various data stored in the storage unit 40, thereby realizing the functions and operation of the abnormal sound detection device 10 described in this embodiment.

[0046] The operating sound data 42 stores waveform information of the operating sound corresponding to the identification information of each of the multiple mobile bodies 100. Specifically, for each mobile body 100, the rotational speed of the motor equipped on the mobile body 100 and waveform information that can reproduce the time waveform of the operating sound generated by the mobile body 100 at each rotational speed are stored in association with each other.

[0047] The sound identification unit 31 receives mobile body information, including identification information and motor rotation speed, from the mobile body 100 via the communication unit 50. The sound identification unit 31 obtains waveform information corresponding to the sound of the mobile body 100 by referring to the sound data 42 based on the identification information and motor rotation speed.

[0048] The reference tone data 41 stores waveform information of a reference tone corresponding to the identification information of each of the multiple moving objects 100. The acoustic propagation characteristics identification unit 32 obtains the waveform information of the reference tone by referring to the reference tone data 41 based on the identification information of the moving objects 100, and generates a time waveform of the reference tone. The acoustic propagation characteristics identification unit 32 also obtains a time waveform of the sound obtained by measuring the reference tone with the microphone 20. The acoustic propagation characteristics identification unit 32 identifies the acoustic propagation characteristics based on the time waveform of the reference tone generated from the reference tone data 41 and the time waveform obtained by actually measuring the reference tone.

[0049] As described above, the operating sound data 42 may be stored in an external device such as a server, and the operating sound identification unit 31 may acquire waveform information of the operating sound from the external device, or the operating sound data 42 may not be stored in the storage unit 40, and the operating sound identification unit 31 may receive waveform information of the operating sound from the mobile body 100. Similarly, the reference sound data 41 may be stored in an external device such as a server, and the acoustic propagation characteristics identification unit 32 may acquire waveform information of the reference sound from the external device, or the reference sound data 41 may not be stored in the storage unit 40, and the acoustic propagation characteristics identification unit 32 may receive waveform information of the reference sound from the mobile body 100.

[0050] The estimated operating sound generation unit 33 generates an estimated operating sound based on the operating sound of the mobile body 100 generated by the operating sound identification unit 31 and the acoustic propagation characteristics identified by the acoustic propagation characteristics identification unit 32. The estimated operating sound generation unit 33 applies the acoustic propagation characteristics to the time waveform of the operating sound generated by the operating sound identification unit 31 to generate a time waveform that estimates the measured operating sound obtained by measuring the operating sound of the mobile body 100 with the microphone 20, and uses this as the estimated operating sound.

[0051] The estimated operating sound cancellation unit 34 cancels the estimated operating sound generated by the estimated operating sound generation unit 33 from the measurement sound of the microphone 20. In other words, the estimated operating sound cancellation unit 34 uses the estimated operating sound to obtain a sound from which the measurement operating sound has been canceled. The estimated operating sound cancellation unit 34 can generate a time waveform from which the estimated operating sound has been canceled by generating a time waveform of the canceled sound that is in opposite phase to the time waveform of the estimated operating sound and combining it with the time waveform of the measurement sound of the microphone 20. In other words, the estimated operating sound cancellation unit 34 can generate a time waveform from which the measurement operating sound has been canceled using the time waveform of the estimated operating sound.

[0052] The abnormal sound detection unit 35 detects abnormal sounds using the time waveform of the sound obtained by the estimated operation sound cancellation unit 34 canceling the estimated operation sound from the measurement sound of the microphone 20, that is, the time waveform of the sound after canceling the measured operation sound from the measurement sound. Abnormal sound detection is performed using a sound pressure threshold for abnormal sound detection, AI, etc. If the measurement sound of the microphone 20 includes a reference sound, the abnormal sound detection process is executed after the reference sound is canceled from the measurement sound.

[0053] The notification unit 36 ​​performs notification processing when the abnormal sound detection unit 35 detects an abnormal sound. The notification processing is performed by displaying information indicating the occurrence of an abnormal situation on the display unit, playing a sound, notifying external devices, etc.

[0054] [Abnormal sound detection processing] An example of the processing performed by the abnormal sound detection device 10 will be explained. Figure 5 is a flowchart showing an example of the processing performed by the abnormal sound detection device 10. Figure 6 is a diagram showing an example of the use of the abnormal sound detection system 1. Figure 7 is a diagram for explaining the acoustic propagation characteristics. Figure 8 is a diagram for explaining the sound obtained by the abnormal sound detection device 10. Each processing shown in Figure 5 will be explained with reference to Figures 4 and 6 to 8. Note that the time waveforms of sound shown in Figures 7 and 8 are schematic diagrams for explaining each processing and do not represent actual sounds.

[0055] The operating sound identification unit 31 of the abnormal sound detection device 10 monitors whether or not it has received mobile body information from the mobile body 100 to the communication unit 50 (Step S101; No). When it receives mobile body information from the communication unit 50 (Step S101; Yes), the operating sound identification unit 31 refers to the reference sound data 41 and operating sound data 42 in the storage unit 40 based on the identification information and motor rotation speed contained in the mobile body information, and obtains waveform information of the reference sound and operating sound emitted by the mobile body 100 (Step S102).

[0056] The acoustic propagation characteristics identification unit 32 monitors whether or not the reference tone of the mobile body 100 has been detected (step S103). For example, based on the mobile body information received from the mobile body 100 to the communication unit 50, the acoustic propagation characteristics identification unit 32 refers to the reference tone data 41 in the storage unit 40 to recognize the characteristics of the reference tone, such as its frequency and reproduction method, and based on these characteristics, monitors whether or not the reference tone is included in the sound measured by the microphone 20. If it detects that the reference tone is included in the measured sound, the acoustic propagation characteristics identification unit 32 identifies the acoustic propagation characteristics from the relationship between the time waveform reproduced from the waveform information contained in the reference tone data 41 and the time waveform of the sound actually obtained by measuring the reference tone with the microphone 20 (step S104).

[0057] For example, as shown in Figure 6, various obstacles 201 and 202 between the moving body 100 emitting the reference sound and the microphone 20 cause reflection, absorption, diffraction, and attenuation of a portion of the reference sound emitted by the moving body 100. As a result, the sound obtained by measuring the reference sound with the microphone 20 will not perfectly match the reference sound emitted by the moving body 100. As shown in Figure 7, when the time waveform 301 of the reference sound reproduced from the reference sound data 41 becomes a different time waveform 302 due to the influence of obstacles 201 and 202, the acoustic propagation characteristic identification unit 32 identifies the acoustic propagation characteristics from the relationship between time waveform 301 and time waveform 302.

[0058] The estimated operating sound generation unit 33 generates an estimated operating sound based on the time waveform of the operating sound of the moving body 100 reproduced by the operating sound identification unit 31 and the acoustic propagation characteristics obtained by the acoustic propagation characteristics identification unit 32 (step S105).

[0059] The estimated operating sound cancellation unit 34 acquires ambient sound by canceling the estimated operating sound generated by the estimated operating sound generation unit 33 from the measured sound of the microphone 20 (step S106). In other words, the estimated operating sound cancellation unit 34 acquires ambient sound by canceling the measured operating sound from the measured sound. For example, as shown by the dashed line in Figure 8(b), the time waveform 405 of the measured operating sound obtained by measuring the operating sound shown in Figure 8(a) with the microphone 20 is different from the time waveform 401 of the operating sound. By applying acoustic propagation characteristics to the time waveform 401 of the operating sound, the time waveform 402 of the estimated operating sound shown by the solid line in Figure 8(b) is obtained. The time waveform 402 of the estimated operating sound is a waveform that approximates the time waveform 405 of the measured operating sound. By canceling the time waveform 402 of the estimated operating sound from the time waveform 403 of the measured sound of the microphone 20 shown in Figure 8(c), the time waveform of the ambient sound shown in Figure 8(d) is obtained. As shown in Figure 8(b), the time waveform 402 of the estimated operating sound is approximately the same as the time waveform 405 of the measured operating sound. Therefore, the time waveform 404 of the ambient sound obtained by canceling the estimated operating sound from the measured sound can be used as the time waveform of the ambient sound obtained by canceling the measured operating sound from the measured sound.

[0060] The abnormal sound detection unit 35 performs abnormal sound detection processing on the ambient sound time waveform 404 obtained by the estimated operation sound cancellation unit 34 (step S107). The ambient sound time waveform 404 is the waveform obtained by canceling the estimated operation sound time waveform 402, that is, the time waveform 405 of the measured operation sound measured by the microphone 20, from the sound actually measured by the microphone 20. Therefore, the abnormal sound detection unit 35 can detect abnormal sounds with high accuracy without being affected by the operation sound of the mobile body 100. When the abnormal sound detection unit 35 detects an abnormal sound, the notification unit 36 ​​performs notification processing to inform of the abnormal situation.

[0061] If the measurement sound from microphone 20 includes a reference sound emitted by the mobile body 100, the reference sound is canceled from the measurement sound before the abnormal sound detection process is executed. Also, although Figure 5 shows one process, the abnormal sound detection device 10 executes each process continuously. When the abnormal sound detection device 10 detects a reference sound, the acoustic propagation characteristics are updated, and when the acoustic propagation characteristics are updated, the estimated operating sound is updated. When the abnormal sound detection device 10, which has received mobile body information, detects that the operating sound of the mobile body 100 has changed, the operating sound is updated, and when the operating sound is updated, the estimated operating sound is updated. The abnormal sound detection device 10 continues to measure sound with microphone 20, cancels the updated estimated operating sound, i.e., the updated measurement operating sound, from the measurement sound of microphone 20, and executes the abnormal sound detection process on the remaining sound.

[0062] In this embodiment, an example using the time waveform of sound has been described, but it is also possible to use the frequency waveform (frequency characteristics) of sound instead of the time waveform. In order to improve the accuracy of detecting abnormal sounds, it is preferable to use a time waveform that includes phase information in addition to frequency, but it is also possible to perform the above-described processing on the frequency characteristics. Even when the frequency characteristics are targeted, the operating sound of the mobile body 100 can be effectively canceled from the measurement sound of the microphone 20.

[0063] In this embodiment, an example of a mobile body 100 that flies while rotating its rotor blades with a motor has been described. However, the mobile body 100 may also be configured to rotate its rotor blades with an engine, or it may move using a jet engine without using rotor blades. The mobile body 100 may also be a vehicle that moves on the ground using wheels, powered by a motor, engine, or jet engine. If an engine is used, the operating sounds of the mobile body 100 can be prepared in advance for each engine rotation speed, and estimated operating sounds can be generated. If a jet engine is used, the operating sounds of the mobile body 100 can be prepared in advance for each turbine rotation speed, and estimated operating sounds can be generated. Furthermore, the configuration in which the operating sounds of the mobile body 100 are prepared for each rotation speed is not limited to a configuration in which they are prepared for each mobile body 100's moving speed. In this case, the moving speed can be included in the mobile body information, the operating sounds can be identified from the moving speed, and the above-described processing can be performed.

[0064] As described above, according to the abnormal sound detection system and abnormal sound detection method of this embodiment, the abnormal sound detection device can determine the acoustic propagation characteristics between the moving object and the microphone by utilizing the reference sound emitted by the moving object. The abnormal sound detection device can estimate the sound obtained by measuring the operating sound of the moving object with the microphone from the information of the operating sound of the moving object and the acoustic propagation characteristics, and cancel it out from the sound measured by the microphone. As a result, the abnormal sound detection device can suppress the influence of the operating sound of the moving object and detect abnormal sounds with high accuracy from the sound measured by the microphone. [Industrial applicability]

[0065] As described above, the abnormal sound detection system and abnormal sound detection method described herein are useful because they can detect abnormal sounds with high accuracy even under the operation of a moving object. [Explanation of Symbols]

[0066] 1. Abnormal Sound Detection System 10 Abnormal sound detection device 20 microphones 30 Control Unit 31. Operation sound identification unit 32. Acoustic propagation characteristics identification unit 33 Estimated operation sound generator 34 Estimated operating noise cancellation unit 35 Abnormal sound detection unit 36 Hochi Department 40 Storage section 50 Communications Department 100 Mobile Units

Claims

1. A microphone to measure ambient sounds, An acoustic propagation characteristics identification unit identifies the acoustic propagation characteristics between the moving body and the microphone based on a reference tone reproduced from reference tone information relating to a reference tone emitted by a moving body and a sound obtained by measuring the reference tone emitted by the moving body with the microphone, An estimated operating sound generation unit generates an estimated operating sound by estimating the sound obtained by measuring the operating sound emitted by the moving body with the microphone, based on the operating sound reproduced from the operating sound information relating to the operating sound emitted by the moving body and the acoustic propagation characteristics. An abnormal sound detection unit cancels the estimated operating sound from the sound obtained by measuring with the microphone and detects abnormal sounds from the remaining sound. An abnormal sound detection system characterized by having the following features.

2. The acoustic propagation characteristic identification unit identifies acoustic propagation characteristics that include conversion information for converting the time waveform of a reference sound emitted by the moving body into a time waveform of a sound obtained by measuring the reference sound emitted by the moving body with the microphone, The estimated operating sound generation unit generates the estimated operating sound time waveform by converting the time waveform of the operating sound emitted by the moving body, which has been reproduced from the operating sound information, using the conversion information. The abnormal sound detection unit detects abnormal sounds from the time waveform of the remaining sound after canceling the time waveform of the estimated operating sound from the time waveform of the sound measured by the microphone. The abnormal sound detection system according to feature 1.

3. A communication unit receives mobile body information from the mobile body, including information regarding the rotation speed of the drive unit included in the mobile body. Furthermore, The estimated operating sound generation unit refers to the operating sound information, which includes information relating the rotational speed of the drive unit to the operating sound emitted by the moving body, and generates the estimated operating sound from the operating sound corresponding to the rotational speed of the drive unit identified based on the moving body information. The abnormal sound detection system according to feature 1.

4. The acoustic propagation characteristics identification unit updates the acoustic propagation characteristics by measuring the reference sound emitted by the moving body at predetermined time intervals using the microphone. The estimated operating sound generation unit updates the estimated operating sound using the updated acoustic propagation characteristics. The abnormal sound detection unit cancels the updated estimated operating sound from the sound measured by the microphone and detects the abnormal sound from the remaining sound. The abnormal sound detection system according to feature 1.

5. An abnormal sound detection system is executed to detect abnormal sounds, and an abnormal sound detection method is provided. The process involves measuring the reference sound emitted by a moving object using a microphone, A step of determining the acoustic propagation characteristics between the moving body and the microphone based on a reference tone reproduced from pre-prepared reference tone information and a sound obtained by measuring the reference tone with the microphone, A step of generating an estimated operating sound by estimating the sound obtained by measuring the operating sound emitted by the moving body with the microphone, based on the operating sound of the moving body reproduced from pre-prepared operating sound information and the acoustic propagation characteristics, The process involves canceling the estimated operating sound from the sound obtained by measuring with the aforementioned microphone, and detecting an abnormal sound from the remaining sound. A method for detecting abnormal sounds, characterized by including the following:

Images