JP2592999B2 - Cavitation noise detection method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a cavitation noise detecting method for detecting cavitation noise generated by rotation of a propeller by using a signal from a sensor provided near a propeller of a ship or the like.

[0002]

2. Description of the Related Art Conventionally, as a method of detecting cavitation noise generated due to rotation of a propeller of a ship or the like, a method of detecting cavitation noise based on the level of a wideband signal received by a sensor near the propeller is known. In the detection method, for example, a detection device as shown in FIG. 2 is used.

FIG. 2 is a functional block diagram showing an example of a configuration of a cavitation noise detection device in a conventional cavitation noise detection method.

This cavitation noise detection device has a sensor 1 provided near a propeller.
, An amplifier 2, a band-limiting filter 3, a sampler 4, a square calculator 5, an integrator 6, a comparator 7, and an output terminal 8 are connected in cascade.

In a conventional cavitation noise detection method using such a cavitation noise detection device, cavitation noise is detected as follows, for example.

When a propeller of a ship or the like rotates, a broadband signal radiated from the vicinity of the propeller is received by the sensor 1. The received broadband signal is amplified to an appropriate level by the amplifier 2, and unnecessary frequency components are removed by the band limiting filter 3. The output of the band limiting filter 3 is converted into a digital signal by the sampler 4,
It is sent to the square calculator 5.

[0007] The square calculator 5 calculates the square value of the output signal of the sampler 4 to obtain the intensity (this is referred to as instantaneous power) P (k) at time k of the wideband signal generated from the vicinity of the propeller, and integrates it. To the container 6. Integrator 6
By integrating the instantaneous power P (k), determine the long integrated power P _a of the received wideband signal (k),
Send to comparator 7.

[0008] The comparator 7, the integrator 6 long integrated power P _a which is output from the (k), and compares with a threshold thl for determining cavitation noise occurs, the long time integral power P _a (k ) Exceeds the threshold thl, a cavitation noise detection signal is output to the output terminal 8.

[0009]

However, in the cavitation noise detection method having the above configuration, the occurrence of cavitation noise is detected using only the long-term integrated power P _a (k) of the received broadband signal. There was such a problem.

[0010] (1) when there is a high level of background noise, such as machinery noise and underwater noise such as ships, even if no cavitation noise occurs, a long period of time integral power P _a of the received wideband signal (k) is Since the threshold value thl of the comparator 7 is exceeded, false alarms increase.

(2) When characteristics of a ship or the like change due to aging, the threshold value thl set at the time of manufacture does not match, so that false alarms increase and the detection rate of cavitation noise decreases.

(3) When a pulse-like signal or the like that does not depend on cavitation noise is received, a false alarm may occur. In particular, in this case, for example, if the integrator 6 is configured by an IIR filter (infinite impulse response filter) or the like in order to improve the processing function of the operation,
The generated false alarms are continuous.

The present invention has a problem that the prior art has a problem that the number of false alarms increases when the background noise level is high, and that the threshold value does not match the cavitation noise generation level due to the secular change of characteristics of a ship or the like. Provided is a cavitation noise detection method that solves the problem that the number of false alarms increases or the detection rate decreases, and that a false alarm is generated when a pulse-like signal or the like that does not depend on cavitation noise is received. It is.

[0014]

In order to solve the above-mentioned problems, the present invention divides a wideband signal received by a sensor provided near a propeller into a low-frequency component and a high-frequency component,
In the cavitation noise detection method of calculating a level variation ratio between the divided low frequency component and high frequency component and detecting cavitation noise generated by rotation of the propeller based on the level variation ratio, a range in which the level variation ratio can be taken Is divided into a plurality of regions, a weighting factor is assigned to each of the regions, a weighting factor corresponding to the calculated level variation ratio is output, and a predetermined number of the weighting factor outputs are cumulatively added in the time direction. The cavitation noise is detected by comparing the result with a predetermined threshold value.

[0015]

According to the present invention, since the cavitation noise detecting method is configured as described above, when a wideband signal near the propeller is received by the sensor, the wideband signal is divided into a high-frequency component and a low-frequency component. The level variation ratio between the two components is calculated based on the obtained high-frequency component and low-frequency component. The range in which the level variation ratio can be taken is divided into a plurality of regions in advance, and a weighting factor is assigned to each region. When the level variation ratio is calculated, a weight coefficient assigned to the area corresponding to the level variation ratio is output,
A predetermined number of the output weight coefficients are added in the time direction. The addition result is compared with a predetermined threshold value, and cavitation noise is detected based on the comparison result.

As a result, even when a pulse-like signal or the like that does not depend on cavitation noise is generated, no false alarm is generated, and cavitation noise can be stably detected regardless of the surrounding environment and aging.

Therefore, the above problem can be solved.

[0018]

FIG. 1 is a functional block diagram of a cavitation noise detection device in a cavitation noise detection method according to a first embodiment of the present invention.

This cavitation noise detecting device has a sensor 10 such as a hydrophone provided near the propeller. The sensor 10 is connected to a frequency dividing means 20 via an amplifier 11.

The frequency dividing means 20 has a function of dividing the output of the amplifier 11 into a low frequency component and a high frequency component. This frequency dividing means 20 has a high-frequency band-limiting filter 21 and a low-frequency band-limiting filter 22 for removing unnecessary frequency components of the received signal. Variance calculator 2
5 and 26 are respectively connected.

The high-frequency band-limiting filter 21 and the low-frequency band-limiting filter 22 are configured with a plurality of bandpass bands so that the characteristics of cavitation noise can be easily captured. The band limiting filters 21 and 22 may be configured by analog band pass filters, or may be configured as digital filters by digitizing the output of the amplifier 11.

Each of the instantaneous power calculators 23 and 24 calculates the instantaneous power based on the output of each of the band limiting filters 21 and 22. Each of the variance calculators 25 and 26 calculates the average value and the variance of each frequency band using the instantaneous power output from each of the instantaneous power calculators 23 and 24.

On the output side of the frequency dividing means 20, a level variation ratio calculator 31, a weight coefficient determiner 32, an accumulator 33,
And a comparator 34 are sequentially connected, and an output terminal 35 is connected to an output side of the comparator 34.

The level variation ratio calculator 31 has a function of calculating a level variation ratio based on the outputs of the variance value calculators 25 and 26, and providing the level variation ratio to the weight coefficient determiner 32.

The weight coefficient judging unit 32 generates a weight coefficient corresponding to the level fluctuation ratio based on the level fluctuation ratio output from the level fluctuation ratio calculator 31, and serves to give the weight coefficient to the accumulator 33.

The accumulator 33 accumulates a predetermined number of weighting factors output from the weighting factor discriminator 32 in the time direction, and outputs the sum to a comparator 34.

The comparator 34 calculates the sum of the sum from the accumulator 33 and a predetermined threshold T for determining cavitation detection noise.
HL, and outputs a cavitation noise detection signal to the output terminal 35 based on the comparison result.

FIG. 3 is a diagram for explaining the weight coefficient judging unit 32 in FIG. 1. The level fluctuation ratio input on the horizontal axis and the weight coefficient ω _l (l = 1, 2, 2) output on the vertical axis. .., N).

The weight coefficient determiner 32 divides the range in which the level variation ratio can be taken into N areas, and assigns a weight coefficient ω _l (l = 1, 2,..., N) to each of the divided areas. , and it outputs a heavy viewed coefficient omega _l corresponding to an input level variation ratio. Here, the number of divided regions takes, for example, a value of about 3 to 4. FIG. 3 shows a case where N = 4.

The cavitation noise detection method of the present embodiment using the cavitation noise detection device configured as described above will be described with reference to FIGS.

In FIG. 1, when a propeller of a ship or the like rotates and a broadband signal is generated in the vicinity of the propeller, the broadband signal is received by a sensor 10 and amplified by an amplifier 11 to an appropriate level. Restriction filter 2
1 and input to the low frequency side band limiting filter 22. The high frequency side band limiting filter 21 and the low frequency side band limiting filter 22 remove unnecessary frequency components of the received signal,
It is divided into a plurality of frequency bands and sent to the instantaneous power calculators 23 and 24 for each band.

The high frequency side instantaneous power calculator 23 calculates the square value P _Hj (k) of the output of the high frequency side band limiting filter 21 for each band (this is called the instantaneous power. J is the band number, j = 1, 2,..., NH)) and outputs the result to the variance value calculator 25. In the variance calculator 25, the instantaneous power P _Hj (k) from the instantaneous power calculator 23 is calculated.
, The average value m _Hj (k) of each band and the average variance δ ² _Hj (k) of each band are obtained by the following equation [Equation 1] using the exponential smoothing method.

[0033]

(Equation 1)

_{M Hj} (k) = (1−α) m _Hj (k−1) + α · P _Hj (k) δ ² _Hj (k) = (1−α) δ ² _Hj (k−1) + α [ P _Hj (k) −m _Hj (k)] ² + (1−α) [m _Hj (k) −m _Hj (k−1)] ² where α is a constant, j = 1, 2,. Further, the variance calculator 25 calculates the average δ of the variance of each band.
^{Based on 2} _Hj (k), the average of the dispersion on the high frequency side δ
² _aH (k) is obtained, and the δ ² _aH (k) is output to the level variation ratio calculator 31.

[0035]

(Equation 2)

[0036] Similarly, the output of the low frequency side band limiting filter 22 is also calculated by the instantaneous power calculator 24 as the square value P _Lj (k) (j =
1, 2,..., NL). Further, the variance value calculator 26 calculates the average value m _Lj (k) of each band and the average δ ^{2 of} the variance.
After determining the _lj (k), and the low-frequency side of the variance of the average [delta] ² _aL (k), and outputs the [delta] ² _aL (k) of the level variation ratio calculator 31.

The level fluctuation ratio calculator 31 calculates the level fluctuation ratio γ (k) based on the outputs δ ² _aH (k) and δ ² _aL (k) of the variance value calculators 25 and 26 as _follows : And outputs the level variation ratio γ (k) to the weight coefficient determination unit 32.

[0038]

(Equation 3)

Γ (k) = δ ² _aH (k) / δ ² _aL (k) (or δ ² _aL (k) / δ ² _aH (k)) The weight coefficient determination unit 32 includes a level variation ratio calculator 31 Weighting factor ω _l (k) corresponding to the level variation ratio γ (k) from
Is read out, for example, based on the correspondence shown in FIG. The way in which the level variation ratio corresponds to the weighting factor is, for example, as shown in FIG. 3, a range in which the level variation ratio can be taken is divided into N regions, and the weighting factor ω _l (l = 1, 2,..., N). Note that N is
For example, a value of about 3 to 4 is taken, and here, for example, N = 4
And

The accumulator 33 accumulates M weight coefficients ω _l (k) output from the weight coefficient determiner 32 in the time direction in accordance with the following equation (Equation 4), and compares the accumulative result W (k) with a comparator. 34
Output to

[0041]

(Equation 4)

[0042] The comparator 34 compares the output W (k) from the accumulator 33 with a predetermined threshold value THL for detecting cavitation noise, and determines whether or not the following expression [Equation 5] is satisfied.

[0043]

(Equation 5)

W (k) ≧ THL (in the case of γ (k) = δ ² _aH (k) / δ ² _aL (k)) When the condition of Expression [5] is satisfied, the comparator 34 sets the cavitation noise The detection signal is output to the output terminal 35.

The present embodiment has the following advantages.

(A) In the cavitation noise detection method of the present embodiment, the wideband signal near the propeller received by the sensor 10 is divided into a high-frequency component and a low-frequency component, and a level variation ratio γ (k) of both components is obtained. A conversion is performed from the level variation ratio having a continuous value to a weighting coefficient ω (k) having a discrete value, and the converted value is added by a predetermined number in the time direction, that is, an accumulation result, that is, an expression [Formula 4] Cavitation noise is detected based on W (k).

For this reason, each weight coefficient ω taking a discrete value
In setting (k) (that is, ω _l (k), l = 1, 2,..., N), even if a high level fluctuation ratio γ (k) enters sporadically, W (k) ) Does not exceed the threshold value THL, a false alarm can be prevented even when a pulse-like signal or the like not caused by cavitation noise occurs.

(B) Further, in the present embodiment, the cavitation noise can be detected almost without being affected by the level of the ambient noise and the secular change of the characteristics of the ship and the like. For example, each weight coefficient ω _l (k) By setting the range of the level variation ratio γ (k) and the number of divided areas, etc., corresponding to the background noise and the change in the characteristics of the ship,
When the background noise level is high or when the characteristics of a ship or the like change due to aging, the effect of preventing false alarms can be further enhanced, and the occurrence of cavitation noise can be detected more effectively.

FIG. 4 is a functional block diagram of a cavitation noise detection apparatus in a cavitation noise detection method according to a second embodiment of the present invention. Elements common to those in FIG. 1 are denoted by the same reference numerals. ing.

In this cavitation noise detection device,
Instead of the frequency dividing means 20 of FIG. 1, a frequency dividing means 40 having a different configuration is provided.

The frequency dividing means 40 has a function of dividing the output of the amplifier 11 into a low frequency component and a high frequency component, and includes a band limiting filter 41 for removing unnecessary frequency components from the output of the amplifier 11. A sampler 42 for converting the output of the band-limiting filter 41 into a digital signal is connected to the output side of the band-limiting filter 41, and a frequency analyzer 43 is connected to the output side. The frequency analyzer 43 uses a technique such as a fast Fourier transform on the output of the sampler 42 to generate a plurality of frequencies bi with a predetermined analysis width.
It has a function of dividing into n, and a bin selector 44 is connected to the output side.

The bin selector 44 selects a frequency bin having the characteristic of cavitation noise from the output of the frequency analyzer 43, and switches the frequency bin between a high frequency side and a low frequency side to output. is there. bin selector 4
4 is connected to a variance calculator 47 via an instantaneous power calculator 45, and to the lower frequency output terminal is connected to a variance calculator 4 via an instantaneous power calculator 46.
8 are connected. The instantaneous power calculators 45 and 46 are:
It has a function of obtaining the instantaneous power from the output of the bin selector 44 and supplying the instantaneous power to the variance calculators 47 and 48. The variance value calculators 47 and 48 have the same configuration as the variance value calculators 25 and 26 in FIG. 1, and the level variation ratio calculator 31 is connected to the output side.

Next, the operation will be described.

The broadband signal generated by the rotation of the propeller is received by the sensor 10 and amplified by the amplifier 11 to an appropriate level. The output of the amplifier 11 is digitized by a sampler 42 and sent to a frequency analyzer 43 after an unnecessary frequency band is removed by a band limiting filter 41.

The frequency analyzer 43 divides the wideband signal digitized by the sampler 42 into a plurality of frequency bins with a predetermined analysis width by using a method such as fast Fourier transform, and the result is bin selector 44. Output to bin
The selector 44 selects a frequency bin having cavitation noise characteristics from the signal from the frequency analyzer 43, and when the bin belongs to the high frequency side, calculates the instantaneous power on the high frequency side. To the container 45, and b
When in belongs to the low frequency side, the result is output to the instantaneous power calculator 46 on the low frequency side.

Here, as an example of a method of selecting a frequency bin, there is a method in which cavitation noise is analyzed in advance, and a bin number in which characteristics of the cavitation noise appear largely is set in advance.

In the instantaneous power calculator 45 on the high frequency side, b
signal x (f _Hj ) (j = 1, 2, 2)
.., NH), the instantaneous power P _Hj (k) on the high frequency side is calculated by the following equation [Equation 6], and output to the variance calculator 47.

[0058]

(Equation 6)

P _Hj (k) = | x (f _Hj ) | ² where j = 1, 2,..., NH Similarly, the instantaneous power calculator 46 on the low frequency side has a bin
The signal x (f _Lj ) from the selector 44 (j = 1, 2,..., N
L) to determine the instantaneous power P _Lj (k) on the low frequency side,
Output to the variance calculator 48. These variance calculators 4
7 and 48, the same as in the first embodiment,
The level fluctuation ratio γ (k) is obtained by the level fluctuation ratio calculator 31, and the weight coefficient determiner 32, the accumulator 33, and the comparator 3
4 detects cavitation noise. Therefore, substantially the same advantages as those of the first embodiment can be obtained.

The present invention is not limited to the above embodiment, but can be variously modified. For example, there are the following modifications.

(A) In the first embodiment, for example, the average value m _Hj (k) and the average variance δ ² of the variance are calculated by the variance value calculators 25 and 26 in accordance with the equation [Equation 1] using the exponential smoothing method. _Hj
Although (k) is obtained, a simple averaging method or the like in which N samples are collected in the time direction and simply added and multiplied by 1 / N may be used.

(B) Frequency dividing means 2 in FIGS. 1 and 4
0 and 40 may have configurations other than those shown. 1 and 4 may be configured as individual circuits using an integrated circuit or the like, or may be configured to be executed by program control using a processor or the like. If executed under such program control, the circuit configuration can be simplified.

(C) In the above embodiment, when the range in which the level variation ratio can be taken is divided into a plurality of regions by the weight coefficient determining unit 32, the number of the divided regions is about 3 to 4 (for example, N = 4). However, the number of the divided areas, the range of the level variation ratio of each area, and the like can be appropriately set according to, for example, the level of the background noise. Also,
Regarding the addition of the weighting coefficients performed by the accumulator 33, the time range to be added or the number of weighting coefficients to be added can be set as appropriate.

[0064]

As described above in detail, according to the present invention, attention is paid to the fact that the level fluctuation on the high frequency side is larger than the level fluctuation on the low frequency side as a characteristic of cavitation noise, and the cavitation noise is received by the sensor. Divides the wideband signal in the vicinity of the propeller into a high-frequency component and a low-frequency component, obtains a level variation ratio of both components based on the divided high-frequency component and low-frequency component, and detects cavitation noise using the level variation ratio. I have to. That is, a weighting coefficient corresponding to the level variation ratio is output, a predetermined number of the weighting coefficient outputs are added in the time direction, the addition result is compared with a predetermined threshold, and cavitation noise detection is performed based on the comparison result. I do.

As described above, the conversion from the level variation ratio having a continuous value to the weighting factor having a discrete value is performed, and the result obtained by adding up the converted values is used for the determination of cavitation noise detection. In addition, even when a pulse-like signal or the like that does not depend on cavitation noise is generated, the generation of a false alarm can be prevented, and a highly reliable cavitation noise detection method can be provided.

Further, the cavitation noise can be detected without being affected by the level of the ambient noise and the aging of the characteristics of the ship and the like. In consideration of this, false alarms can be more effectively prevented when the level of background noise is high, or when the characteristics of ships and the like change due to aging, and the detection rate of cavitation noise generation is further improved. Can be done.

[Brief description of the drawings]

FIG. 1 is a functional block diagram of a cavitation noise detection device according to a first embodiment of the present invention.

FIG. 2 is a functional block diagram showing a configuration example of a cavitation noise detection device in a conventional detection method.

FIG. 3 is a diagram for explaining a weight coefficient determining unit in FIG. 1;

FIG. 4 is a functional block diagram of a cavitation noise detection device according to a second embodiment of the present invention.

[Description of Signs] 10 Sensor 20, 40 Frequency dividing means 21 High frequency side band limiting filter 22 Low frequency side band limiting filter 23, 24, 45, 46 Instantaneous power calculator 25, 26, 47, 48 Dispersion value calculator 31 Level Fluctuation ratio calculator 32 weight coefficient determiner 33 accumulator 34 comparator 41 band limiting filter 42 sampler 43 frequency analyzer 44 bin selector THL predetermined threshold

Claims (1)

(57) [Claims] 1. A broadband signal received by a sensor provided near a propeller is divided into a low frequency component and a high frequency component,
In the cavitation noise detection method of calculating a level variation ratio between the divided low frequency component and high frequency component and detecting cavitation noise generated by rotation of the propeller based on the level variation ratio, a range in which the level variation ratio can be taken Is divided into a plurality of regions, a weighting factor is assigned to each of the regions, a weighting factor corresponding to the calculated level variation ratio is output, and a predetermined number of the weighting factor outputs are cumulatively added in the time direction. A cavitation noise detection method, wherein the cavitation noise is detected by comparing a result with a predetermined threshold.