CN111480196A

CN111480196A - Voice measurement system and parameter generation device

Info

Publication number: CN111480196A
Application number: CN201780097594.4A
Authority: CN
Inventors: 阿部芳春
Original assignee: Mitsubishi Electric Corp
Current assignee: Mitsubishi Electric Corp
Priority date: 2017-12-22
Filing date: 2017-12-22
Publication date: 2020-07-31
Anticipated expiration: 2037-12-22
Also published as: CN111480196B; KR102174553B1; JP6419392B1; JPWO2019123633A1; KR20200075884A; WO2019123633A1

Abstract

A sound generating body (5) is provided in the vicinity of a diagnostic device (4) to be measured. A measurement unit (6) acquires a measurement signal obtained from a sound-receiving body (2) by propagating a test signal, which is a unit signal having one frequency component at each time and a time-varying center frequency of the frequency component, between the sound-generating body (5) and the sound-receiving body (2) on a time axis. An estimation unit (7) estimates the propagation characteristics of sound from the sound emitting body (5) to the sound receiving body (2) based on the relationship between the time and the intensity of the unit signal.

Description

Voice measurement system and parameter generation device

Technical Field

The present invention relates to a sound measurement system for measuring a propagation characteristic of sound, and a parameter generation device for generating a parameter for determining whether a grasped object is in a normal state or an abnormal state using the sound measurement system.

Background

For example, a device disclosed in patent document 1 is known for measuring the propagation characteristics of sound. The device uses a Time Spread Pulse (TSP) signal for measuring the propagation Time of an acoustic wave between a speaker and a microphone.

The propagation characteristics of sound are a general term for impulse response, transfer function, propagation time, distance attenuation, and the like.

Documents of the prior art

Patent document

Patent document 1: japanese patent laid-open publication No. 2004-193782

Disclosure of Invention

Problems to be solved by the invention

In the conventional apparatus described above, since the installation positions of the speaker and the microphone are fixed, there is a problem that the propagation characteristics that change with time cannot be estimated with high accuracy.

The present invention has been made to solve the above-described problems, and an object of the present invention is to provide a sound measurement system capable of accurately estimating a propagation characteristic that changes with time.

Means for solving the problems

The sound measurement system of the present invention includes: a sounding body provided to a measurement target; a sound receiving body provided at a sound receiving point; a measurement unit that obtains a measurement signal obtained from a sound-receiving body by propagating a test signal, which is a unit signal having one frequency component at each time and having a center frequency of the frequency component varying with time, between the sound-generating body and the sound-receiving body on a time axis; and an estimating unit that estimates a propagation characteristic of sound from the sound emitting body to the sound receiving body based on a relationship between time and intensity of the unit signal included in the measurement signal.

Effects of the invention

A sound measurement system of the present invention estimates the propagation characteristics of sound from a sound generating body to a sound receiving body by propagating a test signal, which is a unit signal having one frequency component at each time and a center frequency of the frequency component changing with time, between the sound generating body and the sound receiving body, on a time axis. This makes it possible to estimate the propagation characteristics that change with time with high accuracy.

Drawings

Fig. 1 is a configuration diagram showing an application example of a sound measurement system according to embodiment 1 of the present invention.

Fig. 2 is a configuration diagram showing a sound measurement system and a parameter generation device according to embodiment 1 of the present invention.

Fig. 3 is a structural diagram of a sounding body in the sound measurement system according to embodiment 1 of the present invention.

Fig. 4 is a configuration diagram of a computer that realizes the sound measurement system and the parameter generation device according to embodiment 1 of the present invention.

Fig. 5 is a flowchart showing the operation of the sound measurement system according to embodiment 1 of the present invention.

Fig. 6 is a flowchart showing the operation of the estimation unit in the sound measurement system according to embodiment 1 of the present invention.

Fig. 7 is an explanatory diagram of a process of estimating propagation characteristics from a sound-receiving signal in an estimation unit of a sound measurement system according to embodiment 1 of the present invention.

Fig. 8 is a flowchart showing the operation of the parameter generation device according to embodiment 1 of the present invention.

Fig. 9 is an explanatory diagram showing a threshold value determination method in the parameter generation device according to embodiment 1 of the present invention.

Fig. 10 is an explanatory diagram showing another example of a threshold value determining method in the parameter generating device according to embodiment 1 of the present invention.

Fig. 11 is an explanatory diagram showing a unit signal arrangement in the sound measurement system according to embodiment 2 of the present invention.

Fig. 12 is a flowchart showing the operation of an estimating unit in the sound measuring system according to embodiment 2 of the present invention.

Fig. 13 is an explanatory diagram showing time integration processing in the sound measurement system according to embodiment 2 of the present invention.

Fig. 14A is an explanatory view when the time integration processing is not performed on the unit signal arrays, and fig. 14B is an explanatory view when the time integration processing is performed.

Fig. 15 is an explanatory diagram showing a unit signal arrangement in the sound measurement system according to embodiment 3 of the present invention.

Fig. 16 is a flowchart showing the operation of an estimating unit in the sound measuring system according to embodiment 3 of the present invention.

Fig. 17 is an explanatory diagram showing time integration processing in the sound measurement system according to embodiment 3 of the present invention.

Fig. 18A to 18D are explanatory views showing measurement examples of propagation characteristics of a unit signal array based on a multiplexing degree of 8 in a sound measurement system according to embodiment 3 of the present invention.

Detailed Description

Hereinafter, in order to explain the present invention in more detail, a mode for carrying out the present invention will be described with reference to the drawings.

Embodiment mode 1

Fig. 1 is a configuration diagram showing an elevator system as an application example of a parameter generation device of the present embodiment.

The parameter generating device is composed of AN acoustic sensor 2 and a computer 3 mounted above a car 1, and a sounding body 5 provided in the vicinity of a measurement object 4, the car 1 is a car of AN elevator, the acoustic sensor 2 is composed of a microphone, the computer 3 has a USB terminal and AN L AN terminal, the acoustic sensor 2 is connected to the USB terminal via AN audio interface circuit not shown, a device controlled by the computer 3 is connected to a L AN terminal, and the parameter generating device generates a parameter for AN abnormal sound diagnostic device of the elevator system shown in the figure, for example.

The measurement target 4 is a diagnostic target device in the abnormal sound diagnostic apparatus, and is a device located in the hoistway of the elevator as the diagnostic target device. For example, there are a sheave provided at the top of a rope for driving the car 1, a sheave for supporting the car 1 from below, a car rail for preventing the car from swinging laterally, a counterweight for balancing the weight of the car 1, a governor for adjusting the speed of the car, and the like. The sounding body 5 is constituted by a speaker or the like.

Fig. 2 is a configuration diagram of a sound measurement system according to embodiment 1 and a parameter generation device using the sound measurement system. The sound measurement system 21 is constituted by the sound sensor 2, the sounding body 5, the measurement unit 6, and the estimation unit 7. The parameter generation device 22 includes a pseudo-acoustic synthesis unit 8, an acoustic source database (acoustic source DB)9, a simulation unit 10, and a parameter storage unit 11. The sound sensor 2 is a sound receiver in the sound measurement system 21, and is configured using a microphone. The sounding body 5 is provided near the measurement object 4, and is configured to generate a test sound corresponding to the test signal supplied from the measurement unit 6.

Fig. 3 is a configuration diagram of the sounding body 5, and as shown in the drawing, the sounding body 5 is configured by a control unit 51, a communication interface (communication I/F)52, and a speaker 53, the control unit 51 is configured by a microcomputer, and has a function of performing wireless communication with the measuring unit 6 via the communication interface 52, receiving a test signal, and driving the speaker 53 based on the received test signal to perform output control of the test sound, the communication interface 52 has AN interface of wireless L AN, and has a function of performing communication control with the measuring unit 6, and the speaker 53 is a speaker for sending the test sound to the air in the hoistway of the elevator measuring unit.

Returning to fig. 2, the measuring unit 6 has the following functions: the test sound is sent from the sound generating body 5, and the test sound propagated in the hoistway is acquired by the sound sensor 2. Here, as shown in fig. 7 described later, the test sound is obtained by arranging unit signals having one frequency component at each time and having a center frequency of the frequency component varying with time on a time axis. As the unit signal, a Time Spread Pulse (TSP) signal can be used.

The estimating unit 7 has the following functions: the propagation characteristics of the sound from the sounding body 5 to the acoustic sensor 2 are estimated from the relationship between the time and the intensity of the unit signal included in the test signal. The pseudo sound synthesizing unit 8 has a function of synthesizing a pseudo sound by generating an abnormal sound using a sound source stored in the sound source database 9. The simulation unit 10 has a function of determining parameters based on the synthesized pseudo sound generated by the pseudo sound synthesis unit 8. The parameter storage unit 11 is a storage unit for the parameters determined by the simulation unit 10.

These sound measurement system and parameter generation device are configured using a computer 3. Fig. 4 shows a hardware configuration diagram of the computer 3. As shown in the figure, the computer 3 is constituted by a processor 31, a memory 32, an input/output interface (I/F) 33, and a storage 34. The processor 31 is a processor for executing programs stored in the memory 32 or the memory 34, thereby realizing the functions of the measurement unit 6, the estimation unit 7, the simulated sound synthesis unit 8, and the simulation unit 10, and is configured using a CPU. The memory 32 is constituted by a RAM or the like, and is a memory for temporarily storing data or the like and constituting a work area of the processor 31. The input/output interface 33 is an interface for transmitting and receiving signals between the acoustic sensor 2 and the sounding body 5 and for communicating with other external devices. The memory 34 is a storage unit for storing various data and storing programs corresponding to the functions of the measurement unit 6, the estimation unit 7, the pseudo-acoustic synthesis unit 8, and the simulation unit 10. Further, the memory 34 realizes the sound source database 9 and the parameter storage unit 11.

Next, the operation of the sound measurement system and the parameter generation device according to embodiment 1 will be described.

Fig. 5 is a flowchart showing the operation of the sound measurement system.

First, the measuring unit 6 sends a test signal to the sounding body 5 to generate a test sound from the sounding body 5 (step ST 1). Next, the sound sensor 2 receives a test sound (step ST3) when the car 1 of the elevator is reciprocated between the lowermost floor and the uppermost floor (step ST2), and the test sound is sent to the measurement unit 6. The test sound received by the sound sensor 2 is sent from the measurement unit 6 to the estimation unit 7, the propagation characteristics are estimated by the estimation unit 7 (step ST4), and the propagation characteristics as the estimation result are output (step ST 5).

Fig. 6 is a flowchart showing the operation of the estimating unit 7. Fig. 7 is an explanatory diagram of the process of estimating the propagation characteristics from the test signal in the estimating unit 7.

First, the estimation unit 7 performs time-frequency analysis on the received waveform to obtain an intensity distribution S (t, f) on the time axis (frame t) and the frequency axis (frequency width (bin) f) (step ST 11). The received waveform is divided into frames overlapping each other, and the intensity of each frequency width is obtained by FFT (fast fourier transform) for each frame, thereby performing time-frequency analysis. In fig. 7, a cycle 71 represents a unit signal cycle, and a spectrum 72 represents a spectrum of a measurement signal. The specific frequency band 73 is a specific frequency band for obtaining the signal intensity among all the frequencies of each frame.

Next, the estimating unit 7 obtains the intensity included in the specific frequency band 73 for each frame t from the intensity distribution S (t, f) as an intensity time series b (t) (step ST 12). Further, the estimating unit 7 detects a peak in the intensity time series b (t) (step ST 13). Peak detection is performed by detecting maxima of the intensity time series b (t). In fig. 7,

peaks

74a, 74b, 74c represent detected peaks. Here, when detecting a peak from the intensity time series b (t), components other than the unit signal component included in the sound-receiving signal may be erroneously detected as peaks, and therefore, the interval between the detected peaks is measured, and when the interval between the peaks deviates from the unit signal cycle, the peaks are removed (step ST 14). The processing of step ST14 may be performed as necessary, and may be omitted.

Finally, the estimating unit 7 extracts the peak envelope 75 (see fig. 7) connecting the detected peaks as the intensity of the abnormal sound (step ST15), and outputs the signal of the peak envelope 75 as the estimated propagation characteristic (step ST 16).

Next, the operation of the parameter generation device 22 will be described along the flowchart of fig. 8. In the following description, it is assumed that the propagation characteristics 21a are obtained by the sound measurement system 21, and the simulated sound synthesis unit 8 of the parameter generation device 22 already obtains the propagation characteristics 21 a.

Here, the parameters generated in the present embodiment are as follows.

The abnormal sound diagnosis device is a device as follows: the method comprises the steps of judging whether the action sound of the equipment is normal or abnormal according to the action sound when the equipment is in a normal state and the action sound when the equipment is in an abnormal state. In such an abnormal sound diagnosis apparatus, as a parameter for determining normality and abnormality, for example, there is a threshold value.

In addition to the abnormal sound diagnostic apparatus, for example, the deteriorated sound diagnostic apparatus, the abnormal portion estimating apparatus, and the deteriorated portion estimating apparatus also have parameters for diagnosing the deteriorated sound and estimating the abnormal portion and the deteriorated portion in the apparatuses, respectively. These parameters need to be adjusted to be best suited for each device. Therefore, in the present embodiment, synthetic simulated sound is used to design and adjust these parameters. In addition, in reality, in many cases, the frequency of failure of the device is low, and it is difficult to obtain a sample of abnormal sound or degraded sound, and therefore, it is necessary to use synthetic simulation sound.

First, in a state where the test sound is not emitted from the sound emitting body 5, the sound sensor 2 acquires a waveform of a normal operation sound when the car 1 of the elevator reciprocates between the lowermost floor and the uppermost floor (step ST21) (step ST 22). The normal action sound is recorded in the memory 32. Next, the analog sound synthesizing unit 8 selects a sound source from the sound source database 9 (step ST23), controls the intensity of the sound source in accordance with the estimated propagation characteristics that change with time, and synthesizes a plurality of analog sounds having different abnormal/normal SN ratios (for example, SN ratios in a range of 0.1dB step size to 0-18 dB) overlapping the normal operation sound recorded in the memory 32 (step ST24), and outputs the synthesized analog sounds to the simulation unit 10 as synthesized analog sounds (step ST 25). For convenience of explanation, SN ratio 0 is SN ratio- ∞, and normal operation sound without abnormal sound components is synthetic sound.

Next, the simulation unit 10 obtains the relationship between the parameter, the detection rate, and the false detection rate in the abnormal sound diagnosis apparatus, for example, using the synthesized pseudo sound generated by the pseudo sound synthesis unit 8 (step ST 26). Here, the detection rate and the false detection rate are as follows. The detection rate is a rate at which the operating sound of the device in the abnormal state is correctly determined to be abnormal. On the other hand, the false detection rate is a rate at which the operating sound of the device in the normal state is erroneously determined to be abnormal. In addition, in order to obtain the detection rate and the false detection rate with high accuracy, it is necessary to perform simulation using a large amount of normal sounds and abnormal sounds.

The simulation unit 10 adjusts, for example, a threshold value to be referred to by the abnormal sound diagnostic apparatus as a parameter that affects the detection rate and the false detection rate. The abnormal sound diagnosis device analyzes the operation sound during the diagnosis operation, obtains the degree of abnormality, and compares the degree of abnormality with a threshold value to determine whether there is an abnormality. Therefore, the threshold value is an important parameter for determining the detection rate and the false detection rate as the performance of the abnormal sound diagnosis apparatus.

When a vector representing a threshold is θ, a vector representing an abnormality degree is a, and an index indicating an element of two vectors is K (K is 0,1,2, …, and K is a dimension), the simulation unit 10 determines that a K is abnormal if a [ K ] > Θ [ K ] is true for any K, and otherwise, the simulation unit 10 determines that the K is normal (see the following expression).

Here, the abnormality degree vector a is calculated as follows.

A＝(Y-μ)/σ

Y is a feature vector obtained by analyzing the operating sound to be diagnosed, μ is the average vector thereof, and σ is the standard deviation vector. μ and σ are feature quantities (feature vectors) X obtained by analyzing N normal operation sounds₁,X₂,…,X_N(N is the number of operating sounds in normal operation) and the standard deviation.

Next, the simulation unit 10 obtains a parameter that yields the maximum detection rate within the allowable range of the false detection rate as an optimum parameter (step ST 27). Currently, the threshold θ [ k ] of the index k is used as a parameter, and when the horizontal axis represents this value and the vertical axis represents the error rate (0 to 100%), the characteristics shown in fig. 9 are obtained. Here, the leakage rate 91 is assumed to be (1-detection rate), and the explosion rate 92 is assumed to be false detection rate. The larger the threshold θ [ k ], the lower the explosion rate 92, while the higher the leak rate 91. Further, a limit value 93 as an allowable range of the false detection rate is set to the explosion rate 92. The limit value 93 is, for example, an explosion rate 92 of 5% or less when S/N is 6 dB.

As one method of determining the optimum value of the threshold value θ [ k ], it is possible to determine the leakage rate 91 to be the minimum at the limit value 93, and in this case, the point θ × k in the figure becomes the optimum value.

As shown in fig. 10, the relationship between S/N and a certain threshold (ath) can be obtained using a synthetic simulated sound in which S/N is changed, and the modulation threshold 101 for finding a sign of an abnormality is set to a threshold of S/N6dB, and the abnormality threshold 102 for abnormality determination is determined to a threshold in which S/N is 6+ α dB (α is, for example, 3 dB).

Then, the simulation unit 10 outputs the obtained optimum parameters (step ST28), and stores them in the parameter storage unit 11.

Next, an example in which the simulation unit 10 uses the synthetic simulated acoustic learning sound source position estimation parameter will be described.

For example, when the parameter generating device is applied to the abnormal part estimating device, the sound source position to be estimated by the abnormal part estimating device is, for example, a car, a pit, a counterweight, a ceiling, or the like in an elevator. Here, the sound source position means an installation position of the device generating the abnormal sound in the hoistway, that is, a height from the hoistway bottom surface. The abnormal part estimation device estimates the sound source position by referring to the sound source position estimation parameter. Therefore, the simulation unit 10 optimizes the load and variation of the neural network, which are the sound source position estimation parameters referred to by the abnormal part estimation device, as parameters affecting the estimation of the sound source position.

As an example thereof, the abnormal part estimating device obtains an abnormality degree curve, which is a change curve of the abnormality degree corresponding to the car position of the elevator, from the abnormality degree vector obtained by analyzing the operation sound at the time of diagnosis, inputs the abnormality degree curve to the neural network, obtains the scores of "car", "pit", "counterweight" and "top", which are the estimated scores of the sound source position, and outputs the recognition result having the largest score as the estimation result of the sound source position. The sound source position estimation parameters of the neural network are composed of a load and a bias, and a synthetic model sound whose sound source position is known is used as training data to learn.

As described above, the sound measurement system according to embodiment 1 includes: a sounding body provided to a measurement target; a sound receiving body provided at a sound receiving point; a measurement unit that obtains a test signal obtained from a sound-receiving body by propagating a test signal between the sound-generating body and the sound-receiving body, the test signal being obtained by arranging unit signals having one frequency component at each time and having a center frequency of the frequency component varying with time on a time axis; and an estimating unit that estimates the propagation characteristic of sound from the sound emitting body to the sound receiving body based on the relationship between the time and the intensity of the unit signal included in the test signal.

Further, according to the voice measurement system of embodiment 1, since the unit signal is a time-spread pulse signal, the propagation characteristics of the voice can be estimated with high accuracy.

Further, according to the parameter generating device of embodiment 1, since the parameter for determining whether the measurement target is in the normal state or in the abnormal state is generated using the propagation characteristics estimated by the sound measurement system of embodiment 1, it is possible to obtain a parameter that can be determined with high accuracy even when the propagation characteristics change with time.

Further, the parameter generation device according to embodiment 1 includes: a simulated sound synthesizing unit that generates a synthesized simulated sound using the propagation characteristics estimated by the estimating unit; and a simulation unit that determines the parameters using the synthetic simulated sound, and therefore can obtain parameters that can be determined with higher accuracy.

Embodiment mode 2

The sound receiving signal includes, as noise, device noise (normal operation sound) and external noise in addition to a test signal component emitted from the sound emitting body. In particular, since frequency components of impulsive noise are concentrated in time, there is a high possibility that the impulsive noise is erroneously detected as a peak. Therefore, in embodiment 2, a sound measurement system that removes the influence of impulsive noise on propagation characteristic estimation will be described. The configuration of the sound measurement system and the parameter generation device shown in the drawings is the same as that shown in fig. 2, and therefore the description will be given using fig. 2.

The estimation unit 7 according to embodiment 2 is configured to perform frequency analysis on the unit signal, and obtain the propagation characteristics by shifting the time axis so that the components of the unit signal are at the same time for each frequency. The other configurations of the sound measurement system and the parameter generation device are the same as those of embodiment 1.

Next, the operation of embodiment 2 will be described.

Fig. 11 is an explanatory diagram showing a unit signal arrangement in embodiment 2. As shown in the figure, the estimation unit 7 acquires the arrangement of the unit signals 112 in the unit signal period 111. The estimation unit 7 performs time integration processing on the unit signal arrays. Fig. 12 is a flowchart showing the operation of the estimating unit 7.

First, the estimation unit 7 performs time-frequency analysis on the waveform of the acquired unit signal array to obtain an intensity distribution S (t, f) about the time axis (frame t) and the frequency axis (frequency width f) (step ST 31). Next, an intensity distribution S' (t, f) obtained by shifting the time axis by a time shift amount d (f) corresponding to the frequency f is obtained for each component of the frequency width f of the intensity distribution S (t, f) (step ST 32). Here, the time offset d (fc) for the frequency f ═ fc is calculated by the following equation.

d(fc)＝fc/(Fs/2)*Tw

Here, Fs is a sampling frequency, and Tw is a time length (corresponding to a period) of the unit signal. Further, a frame offset number nd (fc) in which the time offset d (fc) is converted into a frame number (discrete value) is calculated by the following equation.

nd(fc)＝int(d(fc)/fp+0.5)

Here, fp is a frame interval (frame period), int (×) is an integer function for an argument, and 0.5 is a number for reducing a pause error accompanying the integer.

Therefore, the intensity distribution S' (t, f ═ fc) obtained by shifting the time axis is calculated by the following equation.

S’(t,f＝fc)＝S(t+nd(fc),f＝fc)

Fig. 13 is an explanatory diagram showing the processing of steps ST31, ST32 described above.

The time-frequency distribution of the original unit signals (the unit signal arrangement shown in fig. 11) is represented by diagonal stripes shown in fig. 13, and the time-integrated signals whose time axes are shifted are represented by vertical stripes in fig. 13. That is, the time axis of the unit signal 132 of the unit signal cycle 131 is shifted, thereby obtaining a time-aligned signal (shifted unit signal) 133. Here, the left arrow indicates the amount of time shift from the original unit signal in each frequency (the amount of time shift corresponds to nd (fc) calculated as described above when each frequency is fc).

Next, the estimating unit 7 obtains the intensity included in the specific frequency band for each frame t from the intensity distribution S' (t, f) of the time-integrated signal, and sets the intensity as an intensity time series b (t) (step ST 33). Further, the estimating unit 7 detects a peak in the intensity time series b (t) (step ST 34). Peak detection is performed by detecting maxima of the intensity time series b (t). Here, when detecting a peak from the intensity time series b (t), components other than the unit signal component included in the sound-receiving signal may be erroneously detected as peaks, and therefore, the interval between the detected peaks is measured, and when the interval between the peaks deviates from the unit signal cycle, the peaks are removed (step ST 35). The processing of step ST35 may be performed as necessary, and may be omitted.

Finally, the estimating unit 7 extracts the peak envelope of the peak detected by the connection (step ST36), corrects the time delay due to the time offset (step ST37), and outputs the corrected time delay as the estimated propagation characteristic (step ST 38).

Fig. 14A shows a time-frequency intensity distribution when impulsive noise (which may be regarded as interference estimated for propagation characteristics) is superimposed on an acoustic signal, and a peak envelope estimated from the time-frequency intensity distribution. The unit signal component 141a in the sound-receiving signal appears as a diagonal streak, and the impulsive noise (disturbance 142a) appears as a vertical streak. Thus, as the intensity in specific frequency band 143, peak 145a among peaks 144a to 146a has a value higher than peak 144a and peak 146a due to the influence of interference 142 a. Therefore, the peak envelope 147a differs from the estimation result 148 in the absence of interference due to the influence of the peak 145 a.

Fig. 14B shows a time-frequency distribution obtained by applying a frequency-dependent time offset to the time-frequency distribution of fig. 14A and a peak envelope estimated from the time-frequency distribution. The unit signal component 141b after the time shift is represented as a vertical streak, and the disturbance 142b after the time shift is represented as a diagonal streak. Accordingly, the peak 144b to 146b, which are the intensities of the specific frequency band 143, have no influence of the interference 142b after the time shift, and the peak envelope 147b is close to the estimation result 148 shown in fig. 14A when there is no interference.

As can be seen by comparing fig. 14A and 14B, the influence of the impulsive interference is removed from the estimated propagation characteristics as a result of the frequency-dependent time shift.

As described above, according to the sound measurement system of embodiment 2, the estimation unit obtains the propagation characteristics by shifting the time axis for each frequency so that the unit signals have the same intensity at the same time, and therefore, even when impulsive noise is present, for example, the propagation characteristics can be estimated with high accuracy.

Embodiment 3

In

embodiments

1 and 2 using the unit signal array of multiplexing degree 1, the interval of the peaks constituting the peak envelope becomes the cycle of the unit signal. When the change of the propagation characteristic with respect to the time is fast, it is necessary to measure the propagation characteristic at a period shorter in time than the period of the unit signal. Therefore, as embodiment 3, a sound measurement system capable of satisfactorily measuring the propagation characteristics even when the propagation characteristics change rapidly with respect to time will be described. In the present embodiment, in order to avoid the complexity of the description, a case will be described in which the multiplexing degree of the unit signal array is 2, but the present invention can also be applied to a case in which the multiplexing degree is 3 or more, for example, 8. The configuration of the sound measurement system and the parameter generation device shown in the drawings is the same as that shown in fig. 2, and therefore the description will be given using fig. 2.

The measurement unit 6 according to embodiment 3 is configured to use, as a test signal, a unit signal array in which a plurality of unit signals each having a different timing are multiplexed on a time axis. The estimation unit 7 is configured to divide the frequency according to the multiplexing degree of the multiplexed unit signal array, and to obtain the propagation characteristics by shifting the time axis for each frequency so that the intensity of the unit signal is the same for each division. The other configurations of the sound measurement system and the parameter generation device are the same as those of embodiment 1.

Next, the operation of embodiment 3 will be described.

Fig. 15 is an explanatory diagram showing a unit signal arrangement in embodiment 3. As shown in the figure, the estimating unit 7 acquires the arrangement of the unit signals 152 in the unit signal period 151 of the multiplexing degree 2. That is, a unit signal array is obtained by multiplexing 2 unit signals 152 in a unit signal period 151. The estimation unit 7 performs time integration processing on the unit signal arrays. Fig. 16 is a flowchart showing the operation of the estimating unit 7. Here, the degree of multiplexing is m.

First, the estimation unit 7 performs time-frequency analysis on the received waveform to obtain an intensity distribution S (t, f) on the time axis (frame t) and the frequency axis (frequency width f) (step ST 41). Next, an intensity distribution S' (t, f) obtained by shifting the time axis by a time shift amount d (f) corresponding to the frequency f is obtained for each component of the frequency width f of the intensity distribution S (t, f) (step ST 42). Here, the time offset d (fc) for the frequency f ═ fc is calculated by the following equation.

The index ix of the m-divided band to which fc belongs is obtained by m-dividing all bands, and the time shift amount d (fc) is calculated from ix as follows.

bw＝(Fs/2)/m

ix＝int(fc/bw)

d(fc)＝(fc-bw*ix)/(Fs/2)*Tw

Here, m is the multiplexing degree, bw is the bandwidth of the band into which m is divided, ix is the index of the band to which fc belongs, Fs is the sampling frequency, and Tw is the time length (coinciding with the period) of the unit signal.

Further, a frame offset number nd (fc) in which the time offset d (fc) is converted into a frame number (discrete value) is calculated by the following equation.

nd(fc)＝int(d(fc)/fp+0.5)

S’(t,f＝fc)＝S(t+nd(fc),f＝fc)

Fig. 17 is an explanatory diagram showing the processing of steps ST41, ST42 described above.

The time-frequency distribution of the original unit signals (the unit signal arrangement shown in fig. 15) is represented by diagonal stripes shown in fig. 17, and the time-integrated signals whose time axes are shifted are represented by vertical stripes in fig. 17. That is, a time axis of the unit signal 172 of the unit signal period 171 of multiplexing degree 2 is shifted, and a time-aligned signal (shifted unit signal) 173 of multiplexing degree 2 is obtained. Here, the left arrow indicates the amount of time shift from the original unit signal in each frequency (the amount of time shift corresponds to nd (fc) calculated as described above when each frequency is fc).

Next, the estimating unit 7 obtains the intensity included in the specific frequency band b for each frame t from the intensity distribution S' (t, f) of the time-integrated signal, and sets the intensity as an intensity time series b (t) (step ST 43). Further, the estimating unit 7 detects a peak in the intensity time series b (t) (step ST 44). Peak detection is performed by detecting maxima of the intensity time series b (t). Here, when detecting a peak from the intensity time series b (t), components other than the unit signal component included in the sound-receiving signal may be erroneously detected as peaks, and therefore, the interval between the detected peaks is measured, and when the interval between the peaks deviates from the unit signal cycle, the peaks are removed (step ST 45). The processing of step ST45 may be performed as necessary, and may be omitted.

Finally, the estimating unit 7 extracts the peak envelope of the peak detected by the connection (step ST46), corrects the time delay caused by the time offset (step ST47), and outputs the corrected time delay as the estimated propagation characteristic (step ST 48).

When the time-integrated signal (fig. 13) which is not multiplexed and which is described in embodiment 2 is compared with the time-integrated signal (fig. 17) which is multiplexed and which is embodiment 3, the maximum amount of time shift when the time-integrated signal 173 is obtained by multiplexing is m times smaller, and the overall delay is improved. Further, the sampling interval of the propagation characteristic is m times shorter than the unit signal period 171, and the sampling interval is also improved.

Fig. 18 is an explanatory diagram showing an example of measurement of the propagation characteristics of the unit signal array based on the multiplexing degree 8. Fig. 18A shows a time-frequency intensity distribution, fig. 18B shows a time shift result, fig. 18C shows a peak detection result, and fig. 18D shows a peak envelope (propagation characteristic estimation result). In these figures, only the frequency band of 0 to 8000Hz is shown among the frequency bands of 0 to 22050 Hz. Therefore, although the multiplexing degree seems to be about 2, the unit signal array of multiplexing degree 8 is actually used in the frequency band of 0 to 22050 Hz. In addition, in these figures, the horizontal axis shows time (seconds), the vertical axes of fig. 18A and 18B show frequency (Hz), and the vertical axes of fig. 18C and 18D show intensity of a unit signal component.

As shown in fig. 18A, an interference component (vertical stripe) 181 is mixed in around 6 seconds. As shown by the interference 182 after the time shift in fig. 18B, the interference component 181 is dispersed by dividing the time shift of the frequency as shown by an arrow 183. As a result, as shown in fig. 18C, a peak due to interference is not detected in the peak detection result, and as a result, a good peak envelope is estimated (see fig. 18D). Since the multiplexing degree is 8, the estimation result is a time-intensive 1/8 unit signal period.

As described above, according to the sound measurement system of embodiment 3, since the measurement unit uses, as the test signal, a unit signal array in which a plurality of unit signals each having a different timing are multiplexed on the time axis, it is possible to measure the propagation characteristics well even when the propagation characteristics change rapidly with respect to the time.

Further, according to the sound measurement system of embodiment 3, the estimation unit divides the frequency according to the multiplexing degree of the multiplexed unit signal array, and obtains the propagation characteristics by shifting the time axis for each frequency so that the intensity of the unit signal becomes the same time for each division, so that the propagation characteristics can be estimated with high accuracy even when the variation of the propagation characteristics with respect to the time is fast.

In each of the above embodiments, the acoustic sensor 2 as the sound receiver is provided at one location of the car 1, but may be provided at a plurality of locations to acquire test sounds from a plurality of acoustic sensors 2.

In the above embodiments, the example in which the sounding body 5 is provided on the side that does not move (fixed side) and the sound sensor 2 is provided on the side that moves (moving side) has been described, but the present invention is not limited to this, and can be similarly applied to a device in which the sounding body 5 is provided on the moving side and the sound sensor 2 is provided on the fixed side. For example, although the sound sensor 2 is provided at the traffic signal post of the intersection to monitor the vehicle accident sound in the monitoring device for the vehicle accident sound at the intersection, the same application can be applied to this device.

Further, although the above embodiments have been described as examples of applications to elevator systems, other configurations can be similarly applied to voice recognition of moving bodies in factories, voice recognition by mobile robots, and voice recognition of devices including moving bodies such as vehicles and escalators.

In the present application, it is possible to freely combine the respective embodiments, to modify any of the components of the respective embodiments, or to omit any of the components of the embodiments within the scope of the invention.

Industrial applicability

As described above, the sound measurement system and the parameter generation device according to the present invention are related to a configuration for obtaining a propagation characteristic when the propagation characteristic changes with time, and are applied to an abnormal sound diagnosis device for an elevator, for example.

Description of the reference symbols

1: boarding a car; 2: acoustic sensors (sound receptors); 3: a computer; 4: a measurement target; 5: a sound producing body; 6: a measurement section; 7: an estimation unit; 8: an analog sound synthesizing unit; 9: a sound source database; 10: a simulation unit; 11: a parameter storage unit; 21: a sound measurement system; 22: a parameter generating device; 21 a: propagation characteristics.

Claims

1. A sound measurement system is characterized by comprising:

a sounding body provided to a measurement target;

a sound receiving body provided at a sound receiving point;

a measurement unit that propagates a test signal between the sounding body and the sound-receiving body, and acquires a test signal obtained from the sound-receiving body, the test signal being obtained by arranging, on a time axis, unit signals each having one frequency component at each time and a center frequency of the frequency component changing with time; and

and an estimating unit that estimates a propagation characteristic of sound from the sound emitting body to the sound receiving body based on a relationship between time and intensity of a unit signal included in the test signal.

2. The sound measurement system according to claim 1, wherein the estimation unit obtains the propagation characteristic by shifting a time axis for each frequency of the unit signals so that the unit signals have the same intensity at the same time.

3. The sound measurement system according to claim 1,

the measurement unit uses, as the test signal, a unit signal array in which a plurality of unit signals each having a different timing are multiplexed on a time axis.

4. The sound measurement system according to claim 3,

the estimation unit divides the frequency according to the multiplexing degree of the multiplexed unit signal array, and obtains the propagation characteristics by shifting the time axis for each frequency so that the intensity of the unit signal is the same for each division.

5. The sound measurement system according to claim 1,

and the unit signal is a Time Spread Pulse (TSP) signal.

6. A parameter generation apparatus, characterized in that,

the parameter generation device generates a parameter for determining whether the measurement target is in a normal state or an abnormal state, using the propagation characteristic estimated by the sound measurement system according to any one of claims 1 to 5.

7. Parameter generation apparatus according to claim 6,

the parameter generation device has:

a simulated sound synthesizing unit that generates a synthesized simulated sound using the propagation characteristics estimated by the estimating unit; and

and a simulation unit which determines the parameter using the synthetic simulation sound.