JP2019074373A - Marine engine speed estimating device, marine engine speed estimation method, and marine engine speed estimating program - Google Patents

Abstract

To estimate the revolution speed of a propulsion engine with high accuracy from inboard noise.SOLUTION: A marine engine speed estimating device 1 of an embodiment comprises: a Fourier transform unit 54 for Fourier transforming sound frame data acquired from recorded data in which inboard noise is recorded and thereby generating a sound frame spectrum; an averaging processing unit 56 for generating an average sound frame spectrum from a prescribed number of consecutive sound frame spectrum; a sound volume filter processing unit 57 for estimating the operating frequency band of a propulsion engine 110 on the basis of a sound volume filter and the sound volume of the average sound frame spectrum, and acquiring a transmission spectrum, out of the average sound frame spectrum, that passes through the operating frequency band; a peak extraction unit 58 for extracting one or a plurality of peaks included in the transmission spectrum; and a revolution speed calculation unit 59 for calculating, when the number of extracted peaks is one, the revolution speed of the propulsion engine 110 on the basis of the frequency of the peak.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a ship engine rotation speed estimation device for estimating the rotation speed of a ship propulsion engine, a ship engine rotation speed estimation method, and a ship engine rotation speed estimation program.

  Conventionally, in the field of automobiles, there is disclosed a method of collecting an acoustic signal in a vehicle by a portable terminal, extracting an engine rotational speed from an analysis result of the acoustic signal, and estimating an accelerator opening (Patent Document 1) .

  On the other hand, in the field of ships, for large ships, a device for monitoring the number of revolutions of a propulsion engine (also referred to as a "main engine") is generally provided. On the other hand, in a small vessel (for example, a vessel of 19 tons or less or less than 24 meters), such a monitoring device may not be provided, and the number of revolutions of the propulsion engine can not be grasped.

JP, 2015-175288, A

  It is conceivable to estimate the number of revolutions of the propulsion engine by collecting and analyzing the sound in the ship as in the case of a car. However, in the case of a ship, unlike the case of a car, various noises exist in addition to the operation noise of the propulsion engine (hereinafter also referred to as "engine noise"). For example, it includes various sounds such as sounds emitted by various facilities (a steering plate, wireless communication, etc.) on the ship, operation sounds of a power generation engine for driving a generator, sounds (waves, wind, etc.) of an overboard. Be For this reason, it is not easy to extract the operation sound of the propulsion engine from the collected data.

  Even if the operation noise of the propulsion engine can be extracted from the inboard noise, the spectrum (power spectrum) of the engine noise includes a plurality of peaks. Specifically, as shown in FIG. 20, the power spectrum of the engine sound has 1.5 times the peak in addition to having 1 times (fundamental frequency), 2 times, 3 times. There are also peaks of half integer multiples of 2.5 times, 3.5 times and so on. As can be seen from the fact that the peak of 1.5 times is maximized in the example of FIG. 20, the peak of the fundamental frequency is not necessarily the maximum. Also, the power spectrum varies depending on the type of engine. Furthermore, even with the same engine, the power spectrum varies depending on the type of ship (for example, the shape of the hull) and the state of the ship (for example, the load of cargo such as fishing gear). Therefore, even if the power spectrum of the engine sound can be acquired, it is not easy to estimate the number of revolutions of the propulsion engine immediately.

  The present invention has been made based on the above technical recognition, and an object thereof is a ship engine rotational speed estimating device capable of estimating the rotational speed of a propulsion engine from inboard noise with high accuracy, a ship engine A rotation speed estimation method and a ship engine rotation speed estimation program are provided.

The ship engine rotational speed estimation device according to the present invention is
A ship engine rotation speed estimation device for estimating the rotation speed of a ship propulsion engine, comprising:
A data acquisition unit for acquiring sound frame data having a predetermined time width for each predetermined data acquisition time from sound recording data obtained by recording inboard noise including the engine sound of the propulsion engine;
A Fourier transform unit that generates a sound frame spectrum by Fourier transforming the sound frame data;
An averaging processing unit that generates an average sound frame spectrum by taking an average of a predetermined number of consecutive sound frame spectra;
A volume filter created based on a volume profile indicating the relationship between the number of revolutions of the propulsion engine and the volume of the inboard noise, and the operating frequency band of the propulsion engine based on the volume of the average sound frame spectrum A volume filter processing unit for estimating a passing spectrum that passes through the operating frequency band in the average sound frame spectrum;
A first peak extraction unit that extracts one or more peaks included in the pass spectrum as a first peak group;
A number-of-rotations calculation unit that calculates the number of rotations of the propulsion engine based on the frequency of the peak when the number of peaks included in the first peak group is one;
And the like.

Further, in the ship engine rotation speed estimation device,
Based on the number of cylinders of the propulsion engine, the number of cycles and the rated number of revolutions, the fundamental frequency at the rated rotational speed of the propulsion engine is calculated, and the upper limit of the frequency to be analyzed is set based on the fundamental frequency Setting section,
A spectrum removal unit that removes frequency components above the upper limit of the sound frame spectrum before taking an average of the predetermined number of sound frame spectra;
May further be provided.

Further, in the ship engine rotation speed estimation device,
The data acquisition time may be shorter than the time width of the sound frame data.

Further, in the ship engine rotation speed estimation device,
The predetermined number of sound frame spectra may be three or more.

Further, in the ship engine rotation speed estimation device,
A moving average processing unit that generates a sound frame difference spectrum by obtaining a moving average spectrum of the average sound frame spectrum and subtracting the moving average spectrum from the average sound frame spectrum;
A second peak extraction unit that extracts a plurality of peaks as a second peak group from the sound frame difference spectrum;
When a plurality of peaks are included in the first peak group, an integer multiple of the fundamental frequency of the propulsion engine from the first peak group based on the first peak group and the second peak group And a multiple frequency processing unit that extracts half and half integer multiple peaks,
When the number of peaks extracted by the double frequency processing unit is one, the rotation speed calculation unit may calculate the rotation speed of the propulsion engine based on the frequency of the peak.

Further, in the ship engine rotation speed estimation device,
The moving average processing unit may delete a portion below the predetermined bias line in the sound frame difference spectrum.

Further, in the ship engine rotation speed estimation device,
The frequency doubler processing unit
A product of a first matrix whose element is the inverse of the frequency of each peak contained in the first peak group and a second matrix whose element is the frequency of each peak contained in the second peak group Calculate
A third matrix is created by determining whether each element of the matrix obtained by the calculation is within the bandwidth of the harmonic multiple band pass filter of integral multiple and half integral multiple of the fundamental frequency, the third matrix The peaks of integral multiples and half integral multiples of the fundamental frequency of the propulsion engine may be extracted based on the matrix of

Further, in the ship engine rotation speed estimation device,
When a plurality of peaks are extracted by the frequency doubler, a variation coefficient indicating variation of peak power of the predetermined number of sound frame spectra is calculated for each frequency of each peak, and the frequency of each peak is calculated. The apparatus further comprises a variation coefficient filter processing unit that determines the fundamental frequency of the propulsion engine based on the plurality of variation coefficients calculated.
The rotation speed calculation unit may calculate the rotation speed of the propulsion engine based on the peak frequency determined by the variation coefficient filter processing unit.

Further, in the ship engine rotation speed estimation device,
The variation coefficient filter processing unit obtains an average value of peak powers of the predetermined number of sound frame spectra before calculating the variation coefficient, and the predetermined number of peak powers are calculated based on a deviation width from the average value. A part of them may be removed from the calculation target of the variation coefficient.

The ship engine rotational speed estimation method according to the present invention is
A ship engine rotational speed estimation method for estimating the rotational speed of a ship propulsion engine, comprising:
Acquiring sound frame data having a predetermined time width from sound recording data obtained by recording inboard noise including the engine sound of the propulsion engine at predetermined data acquisition times;
Generating a sound frame spectrum by Fourier transforming the sound frame data;
Generating an average sound frame spectrum by averaging a predetermined number of successive sound frame spectra;
A volume filter created based on a volume profile indicating the relationship between the number of revolutions of the propulsion engine and the volume of the inboard noise, and the operating frequency band of the propulsion engine based on the volume of the average sound frame spectrum Estimating a passing spectrum that passes through the operating frequency band in the average sound frame spectrum;
Extracting one or more peaks included in the pass spectrum as a first peak group;
Calculating the number of revolutions of the propulsion engine based on the frequency of the peak if the number of peaks included in the first peak group is one;
And the like.

The ship engine rotational speed estimation program according to the present invention is
A ship engine rotational speed estimation program for estimating the rotational speed of a ship propulsion engine, comprising:
Acquiring sound frame data having a predetermined time width from sound recording data obtained by recording inboard noise including the engine sound of the propulsion engine at predetermined data acquisition times;
Generating a sound frame spectrum by Fourier transforming the sound frame data;
Generating an average sound frame spectrum by averaging a predetermined number of successive sound frame spectra;
A volume filter created based on a volume profile indicating the relationship between the number of revolutions of the propulsion engine and the volume of the inboard noise, and the operating frequency band of the propulsion engine based on the volume of the average sound frame spectrum Estimating a passing spectrum that passes through the operating frequency band in the average sound frame spectrum;
Extracting one or more peaks included in the pass spectrum as a first peak group;
Calculating the number of revolutions of the propulsion engine based on the frequency of the peak if the number of peaks included in the first peak group is one;
On a computer.

  According to the present invention, the rotational speed of the propulsion engine can be estimated with high accuracy from the inboard noise.

It is a figure which shows schematic structure of the ship 100 which concerns on an example. BRIEF DESCRIPTION OF THE DRAWINGS It is a figure which shows schematic structure of the ship engine rotation speed estimation apparatus 1 which concerns on embodiment. It is a figure which shows schematic structure of the control part 50 which concerns on 1st Embodiment. It is an example of the graph which shows the volume profile of the ship noise measured in advance. It is an example of the graph for demonstrating the method to set a limiter line to a volume profile. It is an example of the graph which shows the produced volume filter. It is a graph for demonstrating the production method of the volume filter concerning a modification. It is a figure for demonstrating the method to acquire sound frame data from the sound recording data of the noise in a ship. It is an example of the graph which shows an average sound frame spectrum. It is a figure for demonstrating the method to pass an average sound frame spectrum through a volume filter, and to acquire a pass spectrum. It is a flowchart for demonstrating the processing flow of prior preparation. It is a flowchart for demonstrating the ship engine rotation speed estimation method which concerns on 1st Embodiment. It is an example of the graph which shows the passage spectrum which has a plurality of peaks. It is a figure showing a schematic structure of control part 50A concerning a 2nd embodiment. It is an example of a graph which shows an average sound frame spectrum, and a spectrum which carried out moving average processing of the average sound frame spectrum. It is an example of the graph which shows a sound frame difference spectrum and a threshold (bias line). It is an example of the graph which shows the spectrum cut by the bias line. (A) shows each element of matrix X, (b) is a figure which shows an example of each element of matrix Y. FIG. About five sound frame spectrum, the reference value (frequency _{f p)} power value of a frequency that matches the ^{_{P fp j (j = 1,2,}} ···, 5) are diagrams showing a. It is a figure for demonstrating the process which removes a part of extracted power value, before calculating a coefficient of variation. It is a flowchart for demonstrating the ship engine rotation speed estimation method which concerns on 2nd Embodiment. It is an example of the power spectrum of the engine sound of a propulsion engine.

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

First Embodiment
First, with reference to FIG. 1, a schematic configuration of a ship 100 according to an example will be described. As shown in FIG. 1, the ship 100 is provided with a propulsion engine 110 for rotating a propeller 140 and a power generation engine 120 for rotating a generator 130. The power generation engine 120 and the generator 130 are not essential components. For example, in the case where a battery is used to drive an in-board electrical installation (air conditioning, lighting, winch, etc.), the power generation engine 120 and the generator 130 may not be provided on the ship 100.

  As shown in FIG. 1, the ship 100 is provided with a ship engine rotation speed estimation device 1. The ship engine rotation speed estimation device 1 is configured to estimate the rotation speed of the propulsion engine 110 of the ship 100, as will be described in detail later. The ship engine rotational speed estimation device 1 may be an information processing device (such as a personal computer) installed on the ship 100, or may be an information processing terminal (such as a smartphone or a tablet terminal) carried by a crew member. .

  In the present embodiment, the ship 100 is a small ship whose total length is less than 24 meters, or whose total tonnage is 19 tons or less. The ship engine rotation speed estimation device 1 is not only applied to small vessels, but may be used to estimate the rotation speeds of propulsion engines for mid-sized vessels and large vessels.

<Ship engine speed estimation device>
Next, with reference to FIG. 2, a schematic configuration of the ship engine rotation speed estimation device 1 will be described.

  As shown in FIG. 2, the ship engine rotation speed estimation device 1 includes a sound input unit 10, an information input unit 20, an information output unit 30, a storage unit 40, and a control unit 50. The ship engine rotation speed estimation device 1 may have a communication unit (not shown) for transmitting and receiving information to and from another information processing device via a communication network such as the Internet. Further, the ship engine rotation speed estimation device 1 may have a position detection unit such as a GPS (Global Positioning System) or an acceleration sensor.

  Hereinafter, each configuration of the ship engine rotation speed estimation device 1 will be described.

The sound input unit 10 is a device for inputting the sound around the ship engine rotation speed estimation device 1. The sound input unit 10 is configured of, for example, a microphone mounted on an information processing terminal such as a smartphone.

  The sound input unit 10 inputs inboard noise as the sound around the ship engine rotation speed estimation device 1. “Inboard noise” is noise in the ship 100 including the engine noise of the propulsion engine 110. For example, in addition to the engine noise of the propulsion engine 110, the noises on the inside of the ship include sounds emitted by various equipment (rudder board, wireless devices, navigation instruments, etc.) on the ship, sounds outside the ship (wind, waves, etc.), The operation noise of the power generation engine 120 and the like are included.

  The inboard noise inputted by the sound input unit 10 is converted into a digital signal and then stored in the storage unit 40 as recording data. The sound recording data of the inboard noise may be stored in an external storage device (not shown) communicably connected to the ship engine rotation speed estimation device 1.

  In order to analyze engine sound in detail, it is preferable to minimize frequency resolution when recording inboard noise. On the other hand, the smaller the frequency resolution, the longer the sampling time, which impairs real-time performance. In the present embodiment, 44.1 kHz used for audio recording is used as the sampling frequency Fs. In this case, assuming that the number of sampling points Ns is 2 bytes, the frequency resolution ΔF is 0.6729 [Hz] (= 44, 100/65, 536). Therefore, the sampling time (1 / ΔF) of the inboard noise is about 1.5 seconds. In the present embodiment, as described later, real-time property is secured by setting the recording data acquisition interval (data acquisition time) shorter than the time width (about 1.5 seconds) of the sound frame data.

  The information input unit 20 is an interface for the user to input information to the ship engine rotation speed estimation device 1 and is, for example, a touch panel, a keyboard, a mouse, a button or the like. Note that the sound input unit 10 may function as an information input unit. The user inputs, via the information input unit 20, values necessary for estimation of the number of revolutions of the propulsion engine 110, such as the rated number of revolutions of the propulsion engine 110, the number of cylinders, and the number of cycles.

  The information output unit 30 is an interface that outputs various information (for example, estimated rotation number) to the user, and is, for example, a display (liquid crystal display, organic EL display, or the like) that displays an image.

  The information output unit 30 may be a speaker that outputs information as sound. When the information input unit 20 is a touch panel, the information input unit 20 may double as the information output unit 30.

  In addition, the information output unit 30 may output a video signal or an audio signal to an image display unit or an audio output unit connected to the outside of the ship engine rotation speed estimation device 1, or may be connected to the outside. The data may be output to a printing device (printer), or may be output and stored in a storage device inside or outside the ship engine speed estimation device 1.

  The storage unit 40 is a storage device configured of a hard disk or a semiconductor memory. The storage unit 40 stores data necessary for information processing by the control unit 50, and stores information generated by the control unit 50.

  The control unit 50 controls the operation of the ship engine rotation speed estimation device 1, and is configured by a processor such as a CPU (central processing unit) as hardware.

  Next, the control unit 50 will be described in detail with reference to FIG.

  The control unit 50 includes a frequency upper limit setting unit 51, a volume filter generation unit 52, a data acquisition unit 53, a Fourier transform unit 54, a spectrum removal unit 55, an averaging processing unit 56, and a volume filter processing unit 57. , A peak extraction unit (first peak extraction unit) 58, and a rotation number calculation unit 59. Each part of the control unit 50 is realized by the processor in the marine vessel engine speed estimation device 1 executing a predetermined program (for example, an application of a smartphone). Note that at least one of these functional units may be realized by hardware in the ship engine speed estimation device 1.

  Hereinafter, each part of the control unit 50 will be described in detail.

  The frequency upper limit setting unit 51 acquires values of the rated rotational speed, the number of cylinders, and the number of cycles of the propulsion engine 110 via the information input unit 20. The frequency upper limit setting unit 51 may read out from the storage unit 40 the parameter value (at least one of the rated rotational speed, the number of cylinders, and the number of cycles) stored in advance in the storage unit 40.

The frequency upper limit setting unit 51 calculates the basic frequency at the rated rotation speed of the propulsion engine 110 based on the number of cylinders of the propulsion engine 110, the cycle number, and the rated rotation speed. Specifically, the basic frequency of the engine sound at the rated rotation speed is calculated using the equation (1).

Here, F _rated : basic frequency of engine noise at rated speed, N _rated : rated speed [rpm], C: number of cylinders of propulsion engine 110, S: number of cycles of propulsion engine 110 (stroke) is there.

The frequency upper limit setting unit 51 sets the upper limit (also referred to as “cutoff frequency”) of the analysis target frequency based on the calculated basic frequency at rated rotation. When the analysis target is a value up to ρ _k times the fundamental frequency at the time of rated rotation, the frequency upper limit setting unit 51 calculates and sets the cutoff frequency using Expression (2) and Expression (3). The set cutoff frequency is used by the spectrum removing unit 55 (details will be described later).

Here, F _{lim is a} cutoff frequency, and α is a margin.

For example, in the case of k = 3 and = 2 ₃ = 2, analysis is performed up to a frequency obtained by adding a margin α to twice the fundamental frequency at the rated rotation speed. The margin α may be a value defined in advance or may be a value set by the user via the information input unit 20.

  The volume filter creation unit 52 creates a volume filter (Sound Volume Filter: SVF) based on the volume profile. The sound volume profile is specific to the propulsion engine and the vessel, and indicates the relationship between the number of revolutions of the propulsion engine 110 and the volume of the noise on the ship. Here, the "volume" is a value (integrated value) obtained by integrating power (frequency component) equal to or lower than the cutoff frequency with respect to the power spectrum of the inboard noise in the frequency direction. When the cutoff frequency is not set, the volume is a value obtained by integrating the value of the power spectrum for all frequencies.

  Here, the method of acquiring the volume profile will be described. First, inboard noises are recorded at a plurality of rotational speeds from the idling operation state of the propulsion engine 110 to the rated speed operation state. At this time, the inboard noise is recorded a plurality of times for one rotation speed. Then, each recording data is subjected to Fourier transform to generate a power spectrum. Then, the volume of each power spectrum is calculated. Vertical bars in the graph of FIG. 4A indicate variations in volume. In FIG. 4A, the horizontal axis of the graph indicates the rotational speed of the propulsion engine 110, and the rotational speed of “100%” indicates the rated rotational speed (FIG. 4B and FIG. 5A described later). , (B) the same.).

  Next, an approximate curve is determined based on the volume data of each number of revolutions. This approximate curve shows the volume profile. The solid curve in FIG. 4A shows an example of the volume profile. In many vessels, it is possible to obtain such a volume profile, as it does not have a variable speed mechanism like the transmission of a car (ie only one gear).

  By the way, the volume increases as the rotation speed of the propulsion engine 110 increases. With the propulsion engine 110 alone, the relationship between the volume and the number of revolutions can be expressed by a relatively simple mathematical expression. However, when the propulsion engine is mounted on a ship, the sound volume is also affected by the fluid resistance of the ship 100 and the like, and therefore, changes in complex with the rotational speed. For example, although the approximate curve of the sound volume profile shown in FIG. 4A has a hump indicated by a symbol H, this hump appears due to the large fluid resistance of the hull at the rotation speed (ship speed). In the case of a small boat, when the boat speed (engine speed) is increased, it will slide after several humps (increase and decrease in ship resistance).

  The volume filter creation unit 52 creates a volume filter as follows based on the acquired volume profile.

First, as shown in FIG. 4A, the volume filter creation unit 52 obtains the divergence width ΔV _rated . The deviation width ΔV _rated is the deviation width of the sound volume value most deviated from the value of the average sound volume at a predetermined rotational speed. In the example of FIG. 4A, the predetermined number of revolutions is the rated number of revolutions. The deviation width at the rated speed is often larger than the deviation width at other speeds. For this reason, the estimation precision of rotation speed can be raised by using the divergence range in rated rotation speed. In addition, the volume filter creation unit 52 may obtain the divergence width at rotation speeds other than the rated rotation speed.

Next, as shown in FIG. 4B, the volume filter creation unit 52 sets limiter lines UL and LL having widths of ± ΔV _rated on the volume profile. In FIG. 4B, the limiter line UL is an upper limiter line, and the limiter line LL is a lower limiter line.

  Thereafter, as shown in FIG. 4C, the volume filter creation unit 52 creates a volume filter by converting the horizontal axis (X axis) of the graph indicating the limiter lines UL and LL into a frequency. The volume filter created in this manner is stored in the storage unit 40.

In addition, the production method of a volume filter is not restricted above, It is also possible to adopt a simpler method. A method of creating a volume filter according to a modification will be described with reference to FIG. In this modification, creates a average volume _{V a-idle} idling speed _{N idle,} the average information volume _{V a-rated} 2 points during the rated operation speed _{N rated,} the approximate outline line volume profile. The line indicating the volume profile may be a curve or a straight line.

Specifically, first, as shown in FIG. 5A, the volume filter creation unit 52 calculates a point A indicating the average volume Va _-idle at the idling speed N _{idle and} an average at the rated speed N _rated operation. A curve C (volume profile) passing through two points with a point B indicating the volume Va _-rated is determined. The curve C is, for example, an exponential curve or a quadratic curve.

Next, as shown in FIG. 5B, the volume filter creation unit 52 obtains the divergence width ΔV _rated . The deviation range ΔV _rated is the deviation range of the volume value most diverging from the value of the average volume at a predetermined rotational speed. Here, the predetermined rotational speed is taken as the rated rotational speed. Thereafter, as shown in FIG. 5B, the volume filter creation unit 52 sets limiter lines UL and LL in which the width of ± ΔV _rated is given to the volume profile. Then, as shown in FIG. 5C, the volume filter creation unit 52 creates a volume filter by converting the horizontal axis (X axis) of the graph of the limiter lines UL and LL into a frequency.

  According to this modification, since the volume filter is created from only the information of the two points at the idling operation and at the rated operation, the estimation accuracy of the rotational speed is lowered, but the volume filter can be easily obtained.

  The data acquisition unit 53 acquires sound frame data having a predetermined time width for each predetermined data acquisition time from the recording data in which the inboard noise is recorded. The data acquisition unit 53 causes the storage unit 40 to store the sound frame data extracted from the recording data.

  The data acquisition time is preferably shorter than the time width of the sound frame data. In other words, as shown in FIG. 6, the data acquisition unit 53 may acquire sound frame data so as to overlap each other. As a result, it is possible to secure real time property while securing the time width (that is, frequency resolution) of the sound frame data.

  In the present embodiment, as shown in FIG. 6, the data acquisition unit 53 generates sound frame data SF1 to SF7 each having a length of about 1.5 seconds every 0.7 to 0.8 seconds from the recording data SRD. get. In FIG. 6, the sound frame data acquired most recently is SF7, and the data acquired immediately before that is SF6.

  The Fourier transform unit 54 Fourier transforms the sound frame data to generate a sound frame spectrum. More specifically, each time the data acquisition unit 53 acquires sound frame data, the Fourier transform unit 54 generates a sound frame spectrum of the sound frame data by fast Fourier transform.

  The Fourier transform unit 54 may store the data of the generated sound frame spectrum in the storage unit 40. The stored sound frame spectrum is used, for example, in the variation coefficient filtering described in the second embodiment.

The spectrum removing unit 55 sets frequency components higher than the cutoff frequency (F _lim ) of the sound frame spectrum acquired by the data acquiring unit 53 before the averaging processing unit 56 described later averages the plurality of sound frame spectra. Remove. This can reduce the amount of calculation involved in the subsequent data processing. Furthermore, since the frequency component irrelevant to the engine sound is removed, the estimation accuracy of the rotation speed of the propulsion engine 110 can be improved. However, the spectrum removing unit 55 is not an essential component.

  The spectrum removing unit 55 may store, in the storage unit 40, data of the sound frame spectrum from which frequency components higher than the cutoff frequency have been removed. The stored sound frame spectrum is used, for example, in the variation coefficient filtering described in the second embodiment. Also, the spectrum removing unit 55 may remove the operating frequency of the power generation engine 120. Since the power generation engine 120 rotates at a constant rotational speed, the operating frequency is also constant and known.

  The averaging processing unit 56 generates an average sound frame spectrum by taking an average of a predetermined number of continuous sound frame spectra. In the present embodiment, the averaging processing unit 56 performs the averaging process using the five most recent sound frame spectra. FIG. 7 shows an example of the average sound frame spectrum when the arithmetic average is taken. As shown in FIG. 7, the average sound frame spectrum includes a plurality of peaks including the peak of the fundamental frequency. In the present embodiment, the average sound frame spectrum does not have frequency components higher than the cutoff frequency.

  From the viewpoint of improving the estimation accuracy of the rotation speed of the propulsion engine 110, it is preferable that the number (predetermined number) of sound frame spectra to be averaged is as large as possible. The predetermined number is preferably three or more from the viewpoint of performing variation coefficient filter processing described later. In this embodiment, an average sound frame spectrum is generated by averaging five sound frame spectra. Therefore, as can be seen from FIG. 6, the average sound frame spectrum is generated by using the recording data from about 4.5 seconds ago to the present.

  When the spectrum removing unit 55 is not provided, the averaging processing unit 56 averages the sound frame spectrum generated by the Fourier transform unit 54 as it is. Further, the averaging processing unit 56 may generate the average sound frame spectrum by other averaging processing such as geometric averaging, not limited to arithmetic averaging.

  The volume filter processing unit 57 estimates the operating frequency band of the propulsion engine 110 based on the volume filter created by the volume filter creation unit 52 and the volume of the average sound frame spectrum. In FIG. 8, the band described as “pass band” indicates the operating frequency band, and the measured volume line SL indicates the volume (that is, the integral value of the power spectrum) of the average sound frame spectrum generated by the averaging processing unit 56. It shows. As shown in FIG. 8, the lower limit frequency of the pass band is the frequency at the intersection of the measured volume line SL and the upper limiter line UL, and the upper limit frequency of the pass band is the measured volume line SL and the lower limiter line LL. It is the frequency of the intersection.

  The volume of the average sound frame spectrum is calculated by the volume filter processing unit 57 based on the average sound frame spectrum generated by the averaging processing unit 56. Not limited to this, other functional units of the control unit 50 such as the averaging processing unit 56 may calculate the volume of the average sound frame spectrum.

  After estimating the operating frequency band, the volume filter processing unit 57 obtains a passing spectrum that passes through the estimated operating frequency band in the average sound frame spectrum. The passing spectrum has one or more peaks and includes peaks of the fundamental frequency of the propulsion engine 110.

  The peak extraction unit 58 extracts one or more peaks included in the pass spectrum as a first peak group. The pass spectrum is a set of numerical values (power levels) equally spaced in the frequency direction in the operating frequency band. The peak extraction unit 58 extracts the frequency of the peak (local maximum value) from this set. That is, the data of the first peak group is a set of frequency values of each peak.

When the number of peaks included in the first peak group extracted by the peak extraction unit 58 is one, the rotation speed calculation unit 59 calculates the rotation speed of the propulsion engine 110 based on the frequency of the peak. Specifically, the rotational speed of the propulsion engine 110 is calculated using the equation (4).

  Here, N: rotational speed [rpm] of propulsion engine 110, C: number of cylinders, S: cycle number (stroke), F: frequency of extracted peak.

<Ship engine speed estimation method>
The processing operation of the marine engine rotational speed estimation device 1 according to the first embodiment will be described along the flowcharts of FIG. 9 and FIG. FIG. 9 shows a process flow according to preparation in advance, and FIG. 10 shows a process flow according to speed estimation of the propulsion engine 110.

  First, with reference to FIG. 9, the process flow concerning prior preparation will be described.

The frequency upper limit setting unit 51 calculates the basic frequency at the rated rotation speed of the propulsion engine 110 as described above, and sets the cutoff frequency (F _lim ) based on the calculated basic frequency (step S1). .

  Next, the volume filter creation unit 52 creates a volume filter based on the volume profile as described above (step S2).

  The order of step S1 and step S2 may be switched.

  Next, referring to FIG. 10, a process flow relating to the estimation of the number of revolutions of the propulsion engine 110 will be described.

  First, the data acquisition unit 53 acquires sound frame data having a predetermined time width for each predetermined data acquisition time from the recording data in which the inboard noise is recorded (step S11).

  Next, the Fourier transform unit 54 Fourier transforms the sound frame data acquired in step S11 to generate a sound frame spectrum (step S12).

Next, the spectrum removing unit 55 removes frequency components higher than the cutoff frequency (F _lim ) of the sound frame spectrum acquired in step S12 (step S13).

  Next, the averaging processing unit 56 generates an average sound frame spectrum by taking an average of a predetermined number of continuous sound frame spectra (step S14).

  Next, the volume filter processing unit 57 estimates the operating frequency band of the propulsion engine 110 based on the volume filter created in step S2 and the volume of the average sound frame spectrum generated in step S14, and averages Of the sound frame spectrum, a pass spectrum passing through the estimated operating frequency band is acquired (step S15).

  Next, the peak extraction unit 58 extracts one or more peaks included in the pass spectrum acquired in step S15 as a first peak group (step S16).

  Next, the control unit 50 determines whether the number of peaks included in the first peak group extracted in step S16 is one (step S17). Then, when the number of peaks included in the first peak group is one (S17: Yes), the rotation speed calculation unit 59 determines, based on the frequency of the peak (specified peak), the propulsion engine 110. The number of rotations is calculated (step S18). Since the frequency of the identified peak is the fundamental frequency of the engine sound, the number of rotations of the propulsion engine 110 can be calculated.

  On the other hand, when the number of peaks included in the first peak group is plural (S17: No), narrowing-down processing is performed (step S20). The contents of this process will be described in the second embodiment.

  As described above, in the first embodiment, the volume filter processing unit 57 estimates the operating frequency band of the propulsion engine 110 based on the volume filter created in advance and the volume of the average sound frame spectrum. The sound volume filter processing unit 57 acquires a passing spectrum passing through the estimated operating frequency band in the average sound frame spectrum, and the rotation number calculating unit 59 selects the propulsion engine 110 based on the peak frequency of the passing spectrum. Calculate the rotation speed of. As described above, according to the first embodiment, the number of revolutions of the propulsion engine 110 can be estimated from the inboard noise with high accuracy by using the pre-learned vessel-specific volume filter.

Second Embodiment
Next, a second embodiment according to the present invention will be described. In the present embodiment, as shown in FIG. 11, when the first peak group extracted from the pass spectrum includes a plurality of peaks, the peaks are narrowed down to identify the peak of the fundamental frequency. Hereinafter, the second embodiment will be described focusing on differences from the first embodiment. About components other than control part 50 among ship engine rotation speed estimating devices 1, since it is the same as that of a 1st embodiment, detailed explanation is omitted.

  The control unit 50A according to the second embodiment will be described with reference to FIG.

  Control unit 50A includes moving average processing unit 61, peak extraction unit (second peak extraction unit) 62, and frequency doubler in addition to each unit (frequency upper limit setting unit 51 to rotation speed calculation unit 59) of control unit 50. A processing unit 63 and a variation coefficient filter processing unit 64 are included. In FIG. 12, the configuration common to the control unit 50 is not shown. Each part of control part 50A is realized when a processor in vessel engine speed estimation device 1 runs a predetermined program (for example, application of a smart phone). Note that at least one of these functional units may be realized by hardware in the ship engine speed estimation device 1.

  Hereinafter, each unit newly added by the control unit 50A will be described in detail.

  The moving average processing unit 61 obtains a moving average spectrum of the average sound frame spectrum generated by the averaging processing unit 56. Then, the moving average processing unit 61 generates a sound frame difference spectrum by subtracting the obtained moving average spectrum from the average sound frame spectrum. FIG. 13 shows the average sound frame spectrum (arithmetic average) by a broken line and the moving average spectrum by a solid line. FIG. 14 shows the sound frame difference spectrum as a solid line.

  Furthermore, the moving average processing unit 61 removes a portion below the predetermined bias line in the sound frame difference spectrum. In FIG. 14, the bias line is indicated by an alternate long and short dash line. In the present embodiment, the bias line is set based on the volume of the propulsion engine 110 during idling operation. As described above, by performing a cutoff of the sound frame difference spectrum with the bias line, a power spectrum having a plurality of relatively high peaks can be obtained as shown in FIG. Note that partial removal of the sound frame difference spectrum by the bias line is not essential. Also, the bias line may be set by another method.

  The peak extraction unit 62 extracts a plurality of peaks as a second peak group from the sound frame difference spectrum generated by the moving average processing unit 61. The sound frame difference spectrum is a set of numerical values (power levels) arranged at equal intervals in the frequency direction. The peak extraction unit 62 extracts the frequency of the peak (local maximum value) from this set.

  The second peak group extracted by the peak extraction unit 62 is an integral multiple and a half integral multiple of the fundamental frequency of the propulsion engine 110 (that is, 1.0, 1.5, 2.0, ...) Peak of the frequency of

  In the case where the first peak group extracted by the peak extraction unit 58 described in the first embodiment includes a plurality of peaks, the frequency doubler 63 is based on the first peak group and the second peak group. Then, from the first peak group, peaks of integral multiples and half integral multiples of the fundamental frequency of the propulsion engine 110 are extracted.

  That is, the frequency doubler 63 processes the plurality of peaks (candidates of the fundamental frequency) included in the first peak group and the plurality of peaks (the fundamental frequency and its integral and half integer multiples) included in the second peak group. The plurality of peaks included in the first group of peaks are narrowed down to groups of peaks having frequencies that are integral and half-integer multiples of the fundamental frequency.

  Here, the frequency doubling processing by the frequency doubling processing unit 63 will be described in detail.

First, a matrix F is created from data of the first peak group (a set of frequency values of each peak). The matrix F is a 1-by-n matrix, which is expressed by equation (5).

Here, f ₁ , f ₂ ,..., F _n are frequency values of peaks included in the first peak group, and n is the number of peaks included in the first peak group.

  For example, the matrix F is F = (72.7, 76.0, 78.7, 81.4, 82.8, 86.8, 88.8).

Next, a matrix A having elements of the inverse number of elements of the matrix F is created. This matrix A is an n-by-1 matrix and is expressed by equation (6).

Next, a matrix B is created from the data of the second peak group (the set of frequency values of each peak). This matrix B is a 1-row m-column matrix, and is expressed by equation (7).

Here, b ₁ , b ₂ ,..., B _m are frequency values of peaks included in the second peak group, and m is the number of peaks included in the second peak group.

  For example, the matrix B is B = (57.9, 61.2, 72.7, 86.8, 88.8, 101.6, 103.0, 115.9, 129.9, 135.9, 144.7, 159.5, 173.6, 175.6, 188.4).

Next, the product of the matrix A and the matrix B is calculated, and the matrix X is obtained. This matrix X is a matrix of n rows and m columns, as shown in equation (8).

Next, for the matrix X, a _{k k} harmonic overtone filter (harmonic overtone band pass filter) is prepared. The bandwidth (passband) of the _{k k} harmonic filter in the case where the tolerance error is ± e% is expressed by equation (9).

As a _{k k} harmonics filter, k (= 2 用意_k -1) number of band pass filters are prepared. For example, when the analysis target is up to twice the fundamental frequency at the rated rotation speed (ie, in the case of _{3 3} = 2), three overtone band pass filters (1.0 times, 1, 5 times, etc.) Prepare a 2.0x filter). Although the middle side of equation (9) indicates the (n, m) component of matrix X, this is an example, and each component of matrix X enters the middle side as described later, and ρ _k harmonics It is determined whether to pass the filter.

From equation (9), the bandwidth of the 1.0-fold filter is 0.95 ≦ b _m / f _n ≦ 1.05. Similarly, the bandwidth of the 1.5 × filter is 1.425 ≦ b _m / f _n ≦ 1.575, and the bandwidth of the 2.0 × filter is 1.90 ≦ b _m / f _n It is ≦ 2.10.

  Next, the matrix X is passed through each of the prepared overtone band pass filters. Specifically, it is determined whether the value of each element of the matrix X falls within the bandwidth of each overtone band pass filter. Then, the matrix Y is created by leaving the numerical values of the elements that have passed through any overtone band pass filter unchanged, while setting the elements that do not pass through the bandwidth of any overtone band pass filter to be 0. FIG. 16A shows each element of the matrix X, and FIG. 16B shows an example of each element of the matrix Y. In FIG. 16 (b), the element which does not pass the bandwidth of any harmonic over band pass filter is zero.

Next, the matrix Y is decomposed into _n matrices Y ₁ , Y ₂ ,..., Y _n row by row. Each of the decomposed matrices is a 1-by-m matrix. In the example of FIG. 16B, the matrix Y ₂ = (b ₁ / f ₂ , 0, 0, ..., 0) Next, the matrix Y _p (p = 1, 2, ... n) The elements from the first column to the m-th column are summed up, and the matrix Y _p is discarded (that is, outside the process subject to calculation processing) when the total value falls below a predetermined threshold t.

The threshold value t is given, for example, by the following equations (10) and (11). By using these formulas, it is possible to reduce the weight and accuracy of the calculation process.

For example, in the case of _{3 3} = 2, e = 5, t = 2.375.

On the other hand, when the total value of the respective elements from the first column to the m-th column becomes equal to or more than the threshold value t, the element f _{p of the} matrix F with respect to the matrix Y _{p as} shown in equation (12) Multiply the value) In the example of Equation (12), two elements b ₁ and b _m are extracted as candidates of the fundamental frequency.

Perform the above calculation processing for the matrix Y _n from the matrix Y _1, if not elements 0 is narrowed down to only one, the value is identified as the fundamental frequency of the propulsion engine 110.

By the above-mentioned double frequency processing, peaks of integral multiples and half integral multiples of the fundamental frequency are extracted from the plurality of peaks included in the first peak group. For example, in the case where up to a frequency twice as high as the fundamental frequency at the rated rotation speed is set as the analysis target (that is, in the case of k = 3, _{3 3} = 2), among the plurality of peaks included in the first peak group A peak corresponding to a peak 1.0 times, 1.5 times or 2.0 times the fundamental frequency is extracted.

As described above, the frequency doubler 63 processes the first matrix A ^T whose element is the inverse of the frequency of each peak included in the first peak group, and the frequency of each peak included in the second peak group And the elements b _m / f _{n of} the matrix X obtained by this calculation are set in predetermined numerical ranges (passbands) set for each harmonic band pass filter The matrix Y _p (p = 1, 2,..., N) of one row and m columns is created by determining whether or not it is within. Then, the matrix Y _p for which the sum of elements is equal to or greater than the threshold value t is multiplied by the scalar value f _p . As a result, peaks of integral multiples and half integral multiples of the fundamental frequency of the propulsion engine 110 are extracted. If only one peak is extracted by frequency doubling processing, that peak is identified as the fundamental frequency of the propulsion engine 110.

  That is, when the number of peaks included in the first peak group is narrowed to one by the above-described frequency doubler processing, the rotation speed calculation unit 59 uses the equation (4) as described in the first embodiment. The rotational speed of the propulsion engine 110 is calculated based on the frequency of the peak.

  Next, the variation coefficient filter processing unit 64 will be described.

  The variation coefficient filter processing unit 64 specifies the fundamental frequency by performing variation coefficient filter processing when the frequency can not be reduced to one peak by the multiple frequency processing. In the variation coefficient filtering process, a predetermined number of sound frame spectra stored in the storage unit 40 are referred to, and those with large peak power (level) fluctuations are excluded from the candidates for the fundamental frequency.

More specifically, when the multiple frequency processing unit 63 extracts a plurality of peaks, the variation coefficient filter processing unit 64 detects a predetermined number (five in the present embodiment) of sound frame spectrum peaks for each frequency of each peak. A coefficient of variation (CV _fp to be described later) indicating variation of power is calculated. Then, the fundamental frequency of the propulsion engine 110 is determined based on the plurality of variation coefficients calculated for the frequency of each peak.

  The variation coefficient filter process will be described in detail below.

The variation coefficient filter processing unit 64 sets an element other than 0 (a frequency of the fundamental frequency candidate) remaining in the matrix Y _p created by the frequency doubler processing unit 63 as a reference value of the peak frequency included in the sound frame spectrum. Then, the power of the frequency matching the reference value is extracted from each of a predetermined number of sound frame spectra. FIG. 17 shows power values P _fp ^j (j = 1, 2, 3, 4, 5) of frequencies matching the reference value (frequency f _p ) for five sound frame spectra. The sound frame spectrum used here is, for example, the one stored in the storage unit 40 by the Fourier transform unit 54 or the spectrum removal unit 55.

The variation coefficient filter processing unit 64 calculates the variation coefficient CV _fp using the reference value described above and the power value extracted from the sound frame spectrum. The variation coefficient CV _fp represents the variation of power at the frequency f _p .

First, the variation coefficient filter processing unit 64 calculates the average value of the power of the frequency f _p using Equation (13).

Here, J is a predetermined number of sound frame spectra, P _fp ^j is the power at the frequency f _p of the j-th (j = 1, 2,..., J). In the present embodiment, J = 5.

Next, the variation coefficient filter processing unit 64 calculates the variation coefficient CV _fp representing the variation of the power at the frequency f _{p according} to equation (14).

Then, the variation coefficient filter processing unit 64 calculates a final evaluation scale CV _{p according} to equation (15).

Here, CV _{fp fp} represents a coefficient of variation of ρ times the fundamental frequency candidate f _p . For example, a variation coefficient of 1/2 octave higher frequency than _{f p} (i.e. 1.5 times the frequency of _{f p)} represents the _{CV 1.5fp,} 1 octave higher frequency than _{f p} (i.e. 2.0 times the _{f p} The variation coefficient of the frequency of _f) is expressed as CV _{2.0 fp} . The function min is a function that returns the smallest value among the arguments.

Variation coefficient filter process unit 64, evaluation measure CV _p determines the fundamental frequency candidate f _p is the minimum as the fundamental frequency of the propulsion engine 110.

  When the fundamental frequency is determined by the variation coefficient filtering process as described above, the rotation number calculation unit 59 is determined by the variation coefficient filter processing unit 64 using the equation (4) described in the first embodiment. The rotational speed of the propulsion engine 110 is calculated based on the peak frequency (fundamental frequency).

Incidentally, when the frequency f _p of the fundamental frequency candidate is close to the fundamental frequency of the generator engine 120, by beat phenomenon occurs, it may adversely affect the above variation coefficient filtering. That is, in the frequency region around the fundamental frequency of the power generation engine 120, the extracted power value largely fluctuates, as a result, the value of the variation coefficient becomes large, and the fundamental frequency of the propulsion engine 110 is accurately determined. Will be difficult.

Therefore, among the power in the frequency matching the reference value, that divergence from the power average value of the frequency is large, that excluded from the calculation target of the coefficient of variation CV _fp, to calculate the coefficient of variation CV _fp subsequently preferable. For example, before calculating the variation coefficient CV _fp , the maximum and the minimum of the extracted power values may be excluded from the calculation target. In this case, J in equation (14) is J-2 because two power values are excluded. FIG. 18 (a) shows five power values extracted from the sound frame spectrum, and FIG. 18 (b) shows three power values excluding the maximum power value (j = 2) and the minimum power value (j = 3). Power values are shown. In this example, the coefficient of variation is calculated using power values of j = 1, 4, 5.

As described above, before calculating the variation coefficient CV _fp , the variation coefficient filter processing unit 64 obtains an average value of peak powers of a predetermined number of sound frame spectra, and the predetermined number is calculated based on the difference width from this average value. A part of the peak power of is removed from the calculation object of the variation coefficient CV _fp . A noise is generated between the propulsion engine 110 and the power generation engine 120 by calculating the coefficient of variation CV _fp after excluding power values (maximum value, minimum value, etc.) having a large deviation from the average value. Even in such a case, it is possible to suppress the influence of the noise of the power generation engine 120 and to determine the basic frequency of the propulsion engine 110 with high accuracy.

<Ship engine speed estimation method>
The processing operation (stop-down processing) of the marine engine rotational speed estimation device 1 according to the second embodiment will be described along the flowchart of FIG.

  First, the moving average processing unit 61 obtains a moving average spectrum of the average sound frame spectrum and subtracts the moving average spectrum from the average sound frame spectrum to generate a sound frame difference spectrum (step S21).

  Next, the peak extraction unit 62 extracts a plurality of peaks as a second peak group from the sound frame difference spectrum generated in step S21 (step S22).

  Next, based on the first peak group extracted in step S16 and the second peak group extracted in step S22, the frequency doubler processing unit 63 generates the propulsion engine 110 of the first peak group. Peaks of integral multiples and half integral multiples of the fundamental frequency are extracted (step S23).

  Next, the control unit 50A determines whether the number of peaks extracted in step S23 is one (step S24). Then, if the number of peaks is one (S24: Yes), the process returns to step S18, and the rotation speed calculation unit 59 determines the number of rotations of the propulsion engine 110 based on the frequency of the peak (specified peak). calculate.

  On the other hand, when the number of peaks is more than one (S24: No), the variation coefficient filter processing unit 64 performs variation coefficient filter processing (step S25). The variation coefficient filtering process identifies the fundamental frequency. After step S25, the process returns to step S18, and the number of rotations of the propulsion engine 110 is calculated based on the fundamental frequency specified by the variation coefficient filtering process.

  As described above, in the second embodiment, the fundamental frequency is specified from the plurality of frequency components included in the pass spectrum described in the first embodiment by performing the frequency doubling process and the variation coefficient filtering process. In the double frequency processing, the candidate of the fundamental frequency is narrowed down using the knowledge that the power spectrum of the engine sound of the propulsion engine has integral multiples and half integral multiples of the fundamental frequency. In the variation coefficient filtering process, while the sound of the fundamental frequency is constantly generated, the other sounds (waves, wind, sounds generated by inboard equipment, etc.) are generated or stopped. Narrow down the candidates for the fundamental frequency. Further, even if the operating frequency of the power generation engine 120 is close to that of the propulsion engine 110 and a beat occurs, the beat frequency component is removed by the variation coefficient filtering process because the level fluctuation is relatively large. It is possible.

  Therefore, according to the second embodiment, even if the first peak group includes a plurality of peaks, the peak of the fundamental frequency is specified, and the number of revolutions of the propulsion engine 110 is estimated with high accuracy. be able to.

  As described above, according to the present invention, the rotational speed of the propulsion engine can be accurately estimated from the inboard noise. This makes it possible, for example, to detect an abnormality in the propulsion engine based on the speed of the ship and the estimated number of revolutions.

  Although one skilled in the art may conceive of additional effects and various modifications of the present invention based on the above description, aspects of the present invention are not limited to the embodiments described above. Various additions, modifications and partial deletions are possible without departing from the conceptual idea and spirit of the present invention derived from the contents defined in the claims and the equivalents thereof.

  At least a part of the control units 50 and 50A of the ship engine rotation speed estimation device 1 described in the above-described embodiment may be configured by hardware or may be configured by software. When configured by software, a program for realizing at least a part of the functions of the control units 50 and 50A may be stored in a recording medium such as a flexible disk or a CD-ROM, read by a computer, and executed. The recording medium is not limited to a removable medium such as a magnetic disk or an optical disk, and may be a fixed recording medium such as a hard disk drive or a memory.

  Further, a program for realizing at least a part of the functions of the control units 50 and 50A may be distributed via a communication line (including wireless communication) such as the Internet. Furthermore, the program may be encrypted, modulated, compressed, or stored in a recording medium via a wired line or a wireless line such as the Internet or may be distributed.

  Moreover, you may comprise the ship engine rotation speed estimation apparatus 1 by the several information processing apparatus mutually connected by communication. For example, an information processing apparatus communicably connected to the ship engine rotation speed estimation apparatus 1 may have at least a part of the data processing function of the ship engine rotation speed estimation apparatus 1.

1 ship engine rotation speed estimation device 10 sound input unit 20 information input unit 30 information output unit 40 storage unit 50 control unit 51 frequency upper limit setting unit 52 volume filter creation unit 53 data acquisition unit 54 Fourier transform unit 55 spectrum removal unit 56 averaging Processing unit 57 volume filter processing unit 58 peak extraction unit 59 rotation number calculation unit 61 moving average processing unit 62 peak extraction unit 63 multiple frequency processing unit 64 coefficient of variation filter processing unit 100 ship 110 propulsion engine 120 power generation engine 130 generator 140 Propeller Flim Cutoff frequency LL, UL Limiter line SL Measured volume line SF1 to SF7 Sound frame data

Claims (11)

A ship engine rotation speed estimation device for estimating the rotation speed of a ship propulsion engine, comprising:
A data acquisition unit for acquiring sound frame data having a predetermined time width for each predetermined data acquisition time from sound recording data obtained by recording inboard noise including the engine sound of the propulsion engine;
A Fourier transform unit that generates a sound frame spectrum by Fourier transforming the sound frame data;
An averaging processing unit that generates an average sound frame spectrum by taking an average of a predetermined number of consecutive sound frame spectra;
A volume filter created based on a volume profile indicating the relationship between the number of revolutions of the propulsion engine and the volume of the inboard noise, and the operating frequency band of the propulsion engine based on the volume of the average sound frame spectrum A volume filter processing unit for estimating a passing spectrum that passes through the operating frequency band in the average sound frame spectrum;
A first peak extraction unit that extracts one or more peaks included in the pass spectrum as a first peak group;
A number-of-rotations calculation unit that calculates the number of rotations of the propulsion engine based on the frequency of the peak when the number of peaks included in the first peak group is one;
A ship engine rotational speed estimation device comprising: Based on the number of cylinders of the propulsion engine, the number of cycles and the rated number of revolutions, the fundamental frequency at the rated rotational speed of the propulsion engine is calculated, and the upper limit of the frequency to be analyzed is set based on the fundamental frequency Setting section,
A spectrum removal unit that removes frequency components above the upper limit of the sound frame spectrum before taking an average of the predetermined number of sound frame spectra;
The ship engine rotational speed estimation device according to claim 1, further comprising:   The ship engine rotation speed estimation device according to claim 1, wherein the data acquisition time is shorter than the time width of the sound frame data.   The ship engine rotation speed estimation device according to any one of claims 1 to 3, wherein the predetermined number of the sound frame spectra is three or more. A moving average processing unit that generates a sound frame difference spectrum by obtaining a moving average spectrum of the average sound frame spectrum and subtracting the moving average spectrum from the average sound frame spectrum;
A second peak extraction unit that extracts a plurality of peaks as a second peak group from the sound frame difference spectrum;
When a plurality of peaks are included in the first peak group, an integer multiple of the fundamental frequency of the propulsion engine from the first peak group based on the first peak group and the second peak group And a multiple frequency processing unit that extracts half and half integer multiple peaks,
When the number of peaks extracted by the double frequency processing unit is one, the rotation speed calculation unit calculates the rotation speed of the propulsion engine based on the frequency of the peak. The ship engine rotational speed estimation apparatus in any one of -4.   The ship engine rotation speed estimation device according to claim 5, wherein the moving average processing unit deletes a portion below the predetermined bias line in the sound frame difference spectrum. The frequency doubler processing unit
A product of a first matrix whose element is the inverse of the frequency of each peak contained in the first peak group and a second matrix whose element is the frequency of each peak contained in the second peak group Calculate
A third matrix is created by determining whether each element of the matrix obtained by the calculation is within the bandwidth of the harmonic multiple band pass filter of integral multiple and half integral multiple of the fundamental frequency, the third matrix The ship engine rotation speed estimation device according to claim 5 or 6, wherein peaks of integral multiples and half integral multiples of the fundamental frequency of the propulsion engine are extracted based on the matrix of. When a plurality of peaks are extracted by the frequency doubler, a variation coefficient indicating variation of peak power of the predetermined number of sound frame spectra is calculated for each frequency of each peak, and the frequency of each peak is calculated. The apparatus further comprises a variation coefficient filter processing unit that determines the fundamental frequency of the propulsion engine based on the plurality of variation coefficients calculated.
The said rotation speed calculation part calculates the rotation speed of the said engine for propulsion based on the frequency of the peak determined by the said variation coefficient filter process part, It is characterized by the above-mentioned. Ship engine speed estimation device.   The variation coefficient filter processing unit obtains an average value of peak powers of the predetermined number of sound frame spectra before calculating the variation coefficient, and the predetermined number of peak powers are calculated based on a deviation width from the average value. The ship engine rotation speed estimation device according to claim 8, wherein a part of the movement coefficient is removed from the calculation target of the variation coefficient. A ship engine rotational speed estimation method for estimating the rotational speed of a ship propulsion engine, comprising:
Acquiring sound frame data having a predetermined time width from sound recording data obtained by recording inboard noise including the engine sound of the propulsion engine at predetermined data acquisition times;
Generating a sound frame spectrum by Fourier transforming the sound frame data;
Generating an average sound frame spectrum by averaging a predetermined number of successive sound frame spectra;
A volume filter created based on a volume profile indicating the relationship between the number of revolutions of the propulsion engine and the volume of the inboard noise, and the operating frequency band of the propulsion engine based on the volume of the average sound frame spectrum Estimating a passing spectrum that passes through the operating frequency band in the average sound frame spectrum;
Extracting one or more peaks included in the pass spectrum as a first peak group;
Calculating the number of revolutions of the propulsion engine based on the frequency of the peak if the number of peaks included in the first peak group is one;
A ship engine rotational speed estimation method comprising: A ship engine rotational speed estimation program for estimating the rotational speed of a ship propulsion engine, comprising:
Acquiring sound frame data having a predetermined time width from sound recording data obtained by recording inboard noise including the engine sound of the propulsion engine at predetermined data acquisition times;
Generating a sound frame spectrum by Fourier transforming the sound frame data;
Generating an average sound frame spectrum by averaging a predetermined number of successive sound frame spectra;
A volume filter created based on a volume profile indicating the relationship between the number of revolutions of the propulsion engine and the volume of the inboard noise, and the operating frequency band of the propulsion engine based on the volume of the average sound frame spectrum Estimating a passing spectrum that passes through the operating frequency band in the average sound frame spectrum;
Extracting one or more peaks included in the pass spectrum as a first peak group;
Calculating the number of revolutions of the propulsion engine based on the frequency of the peak if the number of peaks included in the first peak group is one;
Ship engine speed estimation program that causes a computer to execute.

Images