JP4669283B2 - Ship automatic steering system - Google Patents

Description

  The present invention relates to a marine automatic steering apparatus having a wave identifier in order to identify a wave parameter of a wave model used in the estimator.

  The automatic marine steering system is a device that controls the rudder so that the direction from the gyrocompass follows the set course, and the control system obtains the deviation and turning angular velocity from the input of the set course and heading and feeds back. The command steering angle, which is a control amount, is output to the steering machine by multiplying the gain. Since the steering machine moves the rudder angle proportional to the command rudder angle, the hull generates an angular velocity depending on the rudder angle, and the azimuth changes.

As shown in FIG. 9, the closed-loop system of the marine vessel automatic steering apparatus includes a control target including a hull model and a wave model, and a control system including an estimator 20 and a feedback loop 15. Since the steering machine is omitted for simplification, the command steering angle is equivalent to the steering angle. In the figure, s is the Laplace operator,
The preset course each detected azimuth, azimuth, the waves, r is the angular velocity, [delta] _c is the command steering angle, [nu is the white noise N (0, 1), respectively. The unit of azimuth, rudder angle and noise is [deg], and that of angular velocity is [deg / s]. K _sn , T _sn , and T _s3n are nominal values of the hull parameters, and the turning force gain [1 / s], hull time constant [s], and rudder sensitivity time constant [s] are respectively ζ _wn , ω _wn , K _wn is the nominal value of the wave parameter, indicating the attenuation coefficient, natural frequency [rad / s], and gain [1 / s], and K _P and K _D are the feedback gain, indicating the proportional gain and differential gain [s], respectively. . The subscripts (·) _s and (·) _w are the hull and wave, respectively. (·) _N and (·) _a are the nominal and actual values, respectively.
Means estimated values, respectively.

  During stormy weather, disturbance components included in the direction are input to the rudder angle through the control system, so the steering machine performs invalid steering, leading to equipment fatigue and wear, and fuel loss. In order to prevent such a rudder, various filters and weather adjustment functions have been proposed (Non-Patent Document 1, Non-Patent Document 2). Among them, the effectiveness of the disturbance removal effect of an estimator incorporating a wave model as a disturbance model has been reported (Non-patent Documents 3 and 4).

  Patent Document 1 discloses an automatic steering apparatus including a Kalman filter that uses a steady Kalman filter to estimate optimum values of a turning angular velocity and an azimuth error and perform weather adjustment.

However, the hull model to be controlled has nominal parameter uncertainty due to changes in loading, and K _sn , T _sn , and T _s3n in FIG. 9 change to K _sa , T _sa , and T _s3a . For this reason, an estimator configured on the basis of a nominal value model causes an error in the estimated value due to a change in loading. For this reason, the closed loop stability due to the state feedback of the estimated value deteriorates due to the estimation error, and there is a risk that the hull will meander (yaw).

  However, in designing a conventional estimator, no consideration has been given to such parameter uncertainty.

  In view of such a situation, the present inventors actively take into account the parameter uncertainty of the nominal value of the hull parameter so that the estimation error caused by the parameter uncertainty can be reduced. Japanese Patent Application No. 2004-226670 proposes a design method for a marine vessel automatic steering device and its estimator.

One of the design methods of the estimator proposed in this Japanese Patent Application No. 2004-226670 is a natural frequency for estimating a state quantity of a hull model as an estimator without a wave model.
And the damping coefficient ζ _e, and then, as an estimator with a wave model, the natural frequency for estimating the state quantity of the hull model so that the estimation error due to the hull parameters is almost the same as the case without the wave model
The natural frequency ω _ew and the attenuation coefficient ζ _ew for estimating the state quantity of the wave model are designed.

  In this way, the present inventors incorporate a wave model into the design of the estimator, and remove the wave component included in the estimated azimuth and estimated angular velocity estimated by the estimator, thereby reducing the estimator designed without the wave model. It has been found that disturbance removal performance can be improved.

Otsu, Hasegawa, IX. Autopilot Evaluation and Prospect, 3rd Maneuverable Symfodium Text, Japan Society of Shipbuilding Research, Tank Committee, p.243 / 279 (1981) Fossen, T. I., "Guidance and Control of Ocean Vehicles", John Wiley & Sons Ltd., p.222 / 245 (1994) Fossen, T. I., "High Performance Ship Autopilot With Wave Filter", Proceedings of the 10th International Ship Control Systems Symposium (SCSS'93), p.2.271 / 2.285 (1993) Grimble, M. J., Patton, R. J., and Wise, D. A., "The Design of Dynamic Positioning Control System Using Stochastic Optimal Control Thory", Optimal Control Applications and Methods, Vol.1, p.167 / 202 (1980) Japanese Patent Publication No. 4-5598

  As described above, in order to incorporate the wave model into the estimator, it is necessary to specify the wave parameters such as the wave frequency and the attenuation coefficient of the wave model. Since the wave model is based on the relative motion between the sea state and the hull, the value of the wave parameter should be different depending on the hull.

  The present invention can accurately identify a wave frequency and a wave attenuation coefficient of a wave model, and has an estimator that outputs an estimated azimuth and an estimated angular velocity using the identified wave frequency and attenuation coefficient. The object is to provide a steering device.

To achieve the above object, the invention according to claim 1 of the present invention includes a wave identification unit for identifying the wave frequency and wave attenuation coefficient from biopsy attitude position,
Detection direction
And the command rudder angle δc, and using the wave frequency and the wave attenuation coefficient identified by the wave identifier,
And estimated angular velocity
An estimator that outputs and
A feedback gain loop for outputting the command steering angle δc by applying a feedback gain to the deviation between the set course and the estimated azimuth and the estimated angular velocity,
In the marine vessel automatic steering apparatus having a steering machine that changes the steering angle according to the command steering angle δc output from the feedback gain loop , the wave identifier includes:
A spectrum calculation unit that converts time-series data of the detected orientation signal into spectrum data;
Identification data that is an envelope point cloud from spectral data
An identification data extraction unit that extracts (f _i represents a frequency and S _i − represents a spectrum value);
A center frequency calculator for determining the wave frequency f _wn from the center frequency based on the identification data;
Spectral values identifying data (S _i -) and, to minimize the evaluation value based on the deviation between the spectral values of wave model corresponding to the wave frequency based on the wave frequency determined by said center frequency calculation unit (S _i) And an attenuation coefficient calculation unit that _sets the attenuation coefficient of the spectrum S of the wave model to be the wave attenuation coefficient ζ _wn .

  The invention according to claim 2 is characterized in that the wave identifier according to claim 1 further includes a normalization unit that normalizes the spectrum data converted by the spectrum calculation unit.

  According to a third aspect of the present invention, in the first or second aspect of the invention, the identification data extraction unit is directed to a frequency having a maximum spectrum value in an area from an arbitrary start frequency in a predetermined area of the spectrum data. A plurality of envelope points obtained on each of the lower frequency side and the higher frequency side than the frequency at which the maximum spectral value is obtained. Among them, the envelope point group closest to the frequency at which the start frequency takes the maximum spectrum value is extracted as identification data from among the envelope point group having the maximum number of envelope points.

The invention of claim 4, wherein, in the one described in any one of claims 1 to 3, wherein the center frequency calculation unit, the spectral value becomes a value more than the upper threshold value S _H in the identification data For the identification data (f _i , S _i −),

To obtain the wave frequency f _wn .

According to a fifth aspect of the invention, there is provided a connector described in any one of claims 1 to 4, wherein the attenuation coefficient calculation unit, the spectrum value is greater than or equal to a value lower threshold S _L in the identification data For the identification data (f _i , S _i −), the square sum J _ζ of the deviation between the spectrum value and the spectrum value S of the spectrum of the wave model corresponding to the frequency is set to the minimum value.

The wave attenuation coefficient ζ of the wave model satisfying the above is set as the wave attenuation coefficient ζ _wn .

According to a sixth aspect of the invention, a wave identifier for identifying the wave frequency and wave attenuation coefficient from biopsy attitude position,
Detection direction
And the command rudder angle δc, and using the wave frequency and the wave attenuation coefficient identified by the wave identifier,
And estimated angular velocity
An estimator that outputs and
A feedback gain loop for applying the feedback gain to the deviation between the set course and the estimated azimuth and the estimated angular velocity to output the command steering angle δc;
In the marine vessel automatic steering device having a steering machine that changes the steering angle according to the command steering angle δc output from the feedback gain loop , the wave identifier includes:
A wave identification unit for identifying the wave frequency and the wave attenuation coefficient from the spectral data of the time-series data of the detected direction signal;
The wave state is determined by using the time-series data, the wave frequency identified by the wave identification unit f _wn = ω _wn / 2π, and the wave attenuation coefficient ζ _wn, and correcting the wave attenuation coefficient suitable for the wave state And a section.

The invention described in claim 7 is the one described in claim 6, wherein the wave determination unit is a transfer function G _e from the wave to the command steering angle.

In contrast, G _e when the wave attenuation coefficient identified by the wave identification unit is substituted for ζ _wn.
And G _e when 1 / √2 is substituted for ζ _wn.
A constant setting unit for obtaining each constant of
Obtained by the constant setting unit
Against

An evaluation amount calculation unit for _obtaining respective evaluation values J _on and J _off represented by:
Wave attenuation coefficient obtained by correcting the value of ζ _wn used for the smaller one of the evaluation values J _on and J _off
Adopted as

An attenuation coefficient determining unit,
It is characterized by providing.

  According to the present invention, the time-series data of the detected azimuth signal that should reflect the wave model is converted into spectral data. In the spectrum of the detected azimuth signal, the low frequency is a steering component and the high frequency is a wave component. Therefore, the wave frequency and the wave attenuation coefficient, which are wave parameters, can be obtained by obtaining spectrum data of a predetermined region and performing spectrum analysis of the spectrum data reflected by the wave component.

  According to the invention of claim 2, by normalizing the spectrum data obtained from the time-series data of the detected azimuth signal, it becomes possible to compare the degree of coincidence with the theoretical spectrum of the wave model, The wave attenuation coefficient value corresponding to the spectrum can be determined from the degree of coincidence with the spectrum of the wave model.

  According to the third aspect of the present invention, an envelope point group consisting of a set of envelope points at which the spectrum value increases or becomes equal to a frequency having the maximum spectrum value in a predetermined region from an arbitrary start frequency is obtained. Thus, enveloped data can be extracted from uneven spectral data. Further, among the plurality of envelope points, among the envelope points where the number of envelope points is the maximum, the envelope point group closest to the frequency at which the start frequency takes the maximum spectral value is used as identification data. The number of data can be secured and the number of data / data frequency range can be increased, and the accuracy of identification data can be increased.

  According to the invention described in claim 4, the theoretical spectrum of the wave model is not symmetrical with respect to the center frequency, but in the region where the spectrum value close to the center frequency is high, the asymmetry is small. The influence of asymmetry can be reduced by obtaining the center frequency using only the identification data whose value is equal to or higher than the upper threshold value and using it as the wave frequency.

According to the invention of claim 5, wherein, when the spectral data obtained by converting the series data, since the noise component is the main in the foot of the spectrum, identification data (f where spectral values is greater than or equal to a value lower threshold S _L By performing processing only on _i , S _i- ), the influence of noise components can be reduced. Further, the wave attenuation coefficient can be obtained from the attenuation coefficient of the spectrum of the wave model that minimizes the sum of squares of the deviation between the spectrum value of the identification data and the spectrum value of the wave model.

  According to the sixth aspect of the present invention, in the wave state where the sea state is the dredged state, the S / N ratio of the spectrum of the time series data of the detection direction signal in the wave identification unit is lowered, so that the wave frequency and the wave attenuation coefficient are Although the identification error increases, the wave state such as the drought state is judged and corrected to the wave attenuation coefficient suitable for the wave state, thereby preventing the deterioration of the disturbance removal performance by the feedback loop and the estimator and amplifying the disturbance. It becomes possible to avoid the situation of being done.

According to the seventh aspect of the present invention, the transfer function G _e from the wave to the command rudder angle has the notch filter characteristic and the case where the notch filter characteristic is not given (ζ _wn = 1 / √2). Each transfer function
For each transfer function
For each, the sum of squares with the detected azimuth signal is used as the evaluation value, and the value of ζ _wn used for the smaller evaluation value is the wave attenuation coefficient.
By employing as, the transfer function G _e is increased disturbance from waves to the instruction steering angle can be prevented from being amplified.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
Control system for marine autopilot 10 of interest in one overall schematic configuration the present invention, as shown in FIG. 1, the automatic steering device for outputting an instruction steering angle [delta] _c on the basis of the detected orientation as the preset course The control system includes a feedback loop 15 that stabilizes the closed loop, and an estimator 20 that removes a wave component included in the detected direction.
And the command rudder angle δ _c as inputs, and the estimated direction that is the estimated value of the detected direction
And the estimated angular velocity, which is an estimate of the turning angular velocity
And the feedback loop 15 outputs the output of the estimator 20
Is fed back to the set course. At this time, the estimator 20 obtains the estimated azimuth and the estimated angular velocity using the wave frequency ω _wn and the wave attenuation coefficient ζ _wn identified by the wave identifier 22.

  In order to understand the configuration of the estimator 20, first, an estimator without a wave model (ie, a secondary estimator) proposed by the present inventors in Japanese Patent Application No. 2004-226670 and an estimator with a wave model (ie, an estimator with a wave model) (4th order estimator) will be outlined.

  As shown in Fig. 2, the secondary estimator is the difference between the nominal value hull model (devised based on the Nomoto model) and the detected and estimated directions.

Where

age,

Satisfied. Here, k ₁ and k ₂ indicate estimated gains. The above equation and actual hull model relationship

And the transfer function of the second order estimator is

As obtained. here,
Is the transfer function from azimuth and wave to estimated azimuth,
Is the transfer function from azimuth and wave to the estimated angular velocity,
Means a second order estimator,

Respectively. λ _e2 is the characteristic polynomial of the second order estimator,

Δ ₂ and Δ ₁ are the parameter uncertainty of the nominal value of the hull parameter.

Respectively.

The influence of Δ ₂ and Δ ₁ representing the parameter uncertainty on the transfer function is examined, and the attenuation coefficient ζ _e, which is a design parameter in the characteristic polynomial of the second-order estimator of equation (1), so as not to deteriorate the closed-loop stability. And the natural frequency ω _e [rad / s].

As a result of the study, for natural frequency ω _e , if ω _e is increased,
Therefore, using the natural frequency ω _n of the steering system and the estimation coefficient ρ> 1, the natural frequency ω _e is

And set. Here, the setting of ρ is selected to an appropriate value that satisfies the closed loop stability. If the damping coefficient ζ _e is 1 / √2, the transfer function by Δ ₁
The parameter uncertainty sensitivity can be minimized.

Next, consider an estimator with a wave model. An estimator with a wave model (that is, a fourth-order estimator) can separate and extract a wave component included in the detection direction as an estimated wave by incorporating the wave model into the second-order estimator. Therefore, the wave component contained in the estimated azimuth and the estimated angular velocity is removed, and useless steering can be prevented. However, the situation where the introduction of the wave model increases the estimation error must be avoided. Therefore, it is necessary to design a fourth-order estimator that maintains an estimation error of the second-order estimator and realizes effective disturbance removal performance. For example, as shown in FIG. 3, the fourth-order estimator includes a nominal hull model (devised based on the Nomoto model) and a detection direction.
Estimated from
And estimated waves
And minus

And

It can be expressed as. Here, k _{i, i = 1 to 4} are estimated gains,
Indicates a state quantity, respectively. The natural frequency ω _wn and the attenuation coefficient ζ _wn of the wave model are detected by the wave identifier 22 using the detection direction.
Identified from Since the unknown input ν of the wave model assumes white noise, the average value becomes zero. Thus corresponding to ψ _w and ψ _w
Since the average value is also zero, it is not necessary to input ν in the above equation. By rearranging the above equation and the hull model relational equation for δ _c , the transfer function of the fourth-order estimator is

As obtained. here,
Is the transfer function from direction and wave to estimated wave,
Means a fourth order estimator,

Respectively. λ _e4 is the characteristic polynomial of the fourth-order estimator

Indicates. λ _w and c _ew are the wave model polynomial and gain polynomial respectively.

Indicates.

The characteristic polynomial λ _e4 of the fourth order estimator is _expressed using two second order standard systems.

Stipulated in Here, it is assumed that λ _e corresponds to the state estimation of the hull model, and λ _ew corresponds to that of the wave model. ζ _e and ω _e are designed by a secondary estimator. ζ _ew and ω _ew are design parameters of the fourth-order estimator and indicate an attenuation coefficient and a natural frequency [rad / s], respectively. So the estimated gain is

become.

As a result of examination, transfer function corresponding to wave removal
In order to provide a notch filter characteristic at the wave frequency ω _wn and to prevent deterioration of the estimation error, an estimator when there is no wave model (λ _ew = 1) so as to satisfy ω _ew ≧ ω _e Satisfies ω _e = ω _wn (where ω _wn is the natural frequency of the wave model) when the sensitivity to the hull parameter is approximately equal to the sensitivity of the estimator to the hull parameter when the wave model is present the value of omega _e at the time of omega _x, the value of omega _e when not satisfied ω _e = ω _wn when the omega _y, when wave frequency omega _wn is greater than omega _x is the natural frequency omega _e set omega _y, the natural frequency omega _ew set equal to the wave frequency omega _wn, when wave frequency omega _wn is less omega _x, set the natural frequency omega _e to omega _x, waves of modifying the wave frequency omega _wn The frequency ω _wn ^* = ω _x and the natural frequency ω _ew is corrected. It has been found that it should be set equal to the wave number ω _wn ^* .

The configuration of the estimator is not limited to the one outlined above, and may be any configuration. However, when the estimator is to introduce a wave model, the wave frequency ω _wn and the wave It is necessary to specify the damping coefficient ζ _wn .

The configuration of the wave identifier 22 for identifying the wave frequency ω _wn and the wave attenuation coefficient ζ _wn of the wave model is shown in FIG. In this embodiment, the wave parameter is identified by a batch method, and the wave identifier 22 includes a data store unit 24, a spectrum calculation unit 26, a wave identification unit 28, and a wave determination unit 30.

  The data store unit 24 stores the A / D converted detection direction signal as time series data, and the spectrum calculation unit 26 converts the time direction data of the detection direction signal into spectrum data. The identification unit 28 calculates the center frequency and the attenuation coefficient, which are wave parameters, from the spectrum data from the spectrum calculation unit 26, and the wave discrimination unit 30 determines the wave state from the time-series data of the detected azimuth signal and the wave parameters. It is corrected to the optimum value of the wave parameter suitable for the above.

  The wave identification unit 28 further includes a normalization unit 28A, an identification data extraction unit 28B, a center frequency calculation unit 28C, and an attenuation coefficient calculation unit 28D. The wave determination unit 30 further includes a constant setting unit 30A, an evaluation amount calculation unit 30B, It comprises an attenuation coefficient determination unit 30C and an attenuation coefficient calculation unit 30D. The wave identification unit 28 and the wave determination unit 30 can be configured by computer software. Detailed configurations and operations of the wave identification unit 28 and the wave discrimination unit 30 will be described below.

2 Wave identification unit 2.1 Normalization unit The wave model transfer function G _w is

Is used. Since the wave model spectrum S is the square of the gain of the transfer function,

become. Here, ω represents a frequency. From equation (4), when ω = ω _wn , S (ω) takes the maximum value, and its magnitude is 1.

On the other hand, detection direction
Wave component contained in
The time series of

Given in. Here, σ _w is a constant representing the intensity of waves, ν is random noise with an average value of zero, and t is time. Maximum value for the conversion from the time series data of the detected azimuth spectrum S _h is
Therefore, _Sh is the maximum value
Divided by
Is equivalent to the spectrum S of the wave model. That is,

become. The spectrum normalized to the maximum value of 1 obtained from time series data
Is called a real spectrum.

The normalization unit 28 </ b> A is detected by the spectrum calculation unit 26.
The spectrum obtained by transforming the time series data using Fast Fourier Transform (FFT) is normalized to obtain data on the actual spectrum. Since the FFT-converted real spectrum does not cover the band frequency, it is not a smooth mountain shape but an uneven shape. In order to reduce parameter identification errors due to unevenness, it is preferable to perform smoothing as appropriate. As a countermeasure against aliasing, it is possible to take a moving average of time series data.

2.2 Identification Data Extraction Unit The identification data extraction unit 28B uses identification data used for identification from the actual spectrum normalized by the normalization unit 28A.
To extract.

  Referring to the actual spectrum example in FIG. 5, in the actual spectrum, the low frequency band is a steering component, and the high frequency band is a wave component. Separating the low frequency side and the high frequency side from each other, envelope data is extracted from each.

That is, an envelope point at which the spectrum value becomes larger or equal from the arbitrary starting frequency toward the frequency that takes the maximum spectrum value.
Find a set of (envelope points). In the figure, “x” and “◯” indicate envelope points.

  Next, among the plurality of envelope point groups, the envelope point group closest to the frequency having the maximum start frequency among the envelope point groups having the maximum number of envelope points is used as identification data. The reason for selecting an envelope point group that maximizes the number of envelope points is to increase the number of data, and the reason for selecting an envelope point group whose start frequency is close to the frequency of the maximum spectral value is to increase (data number ÷ data frequency range). It is to do. In the figure, the numbers of x and ◯ are the same on the low frequency side and high frequency side, respectively, but the start frequency of ◯ is closer to the maximum frequency than that of ×. Therefore, the envelope point group of ○ becomes the identification data.

2.3 Center Frequency Calculation Unit The center frequency calculation unit 28C calculates the center frequency from the identification data.

The spectrum of the wave model of equation (4) is shown in FIG. In a spectrum having a certain attenuation coefficient, a frequency range that takes a certain spectrum value is [ω ₁ , ω ₂ ] (ω ₂ > ω ₁ ). From the figure, it is understood that the spectrum shape is determined by the attenuation coefficient, and the spectrum is not symmetrical with respect to ω / ω _wn = 1. That is,

It is. This asymmetry is proportional to the attenuation coefficient.

FIG. 7 shows the degree of asymmetry with respect to the spectrum. From the figure, it can be seen that when the attenuation coefficient is 0.3, the upper threshold value S _H = 0.8 should be set in order to make the degree of asymmetry within 3%. This upper threshold value _SH is used as a criterion for identifying identification data above which threshold value should be targeted when the real spectrum stands. Because if the frequency taking the maximum spectral value not necessarily the center frequency from the relationship of the resolution occurs, the center frequency f _wn = ω _wn / 2π upper threshold S _H or more spectral values
And its frequency f _i [Hz]

Ask for. Here sigma _i means that summing by the number of upper threshold S _H or identification data.

2.4 Attenuation coefficient calculation unit The attenuation coefficient calculation unit 28D calculates an attenuation coefficient from the identification data.

As shown in FIG. 6, the base of the spectrum has a rapidly expanding base near zero. The base region of the real spectrum is excluded from the identification range because the noise component is dominant. The effective area and the ineffective area need to be appropriately separated. For example, the effective area of the spectrum is set to 95% in the upper part and the ineffective area is set to 5% in the lower part. The frequency range [ω ₁ , ω ₂ ] at the lower threshold S _L = 0.05 is obtained by using the equation (4).

2 roots
As obtained. The attenuation coefficient is calculated using the identification data in the frequency range [ω ₁ , ω ₂ ] with S _L = 0.05.

The attenuation coefficient ζ _wn is given by the attenuation coefficient ζ of the spectrum of the wave model that minimizes the evaluation function using the square sum of the deviation between the spectrum value of the identification data and the spectrum value of the wave model corresponding to the frequency. That is,

Ask for. Here, S means the spectral value of the wave model, and Σ _i means the sum of the number of identification data in the identification range. Since the above equation has already obtained ω _wn = 2πf _wn , it becomes a function of ζ _wn and results in a minimum value problem with one variable. The solution should be golden section.

  The wave parameter can be identified from the spectrum data of the time-series data of the detected azimuth signal by the wave identification unit 28 as described above.

3 Wave Discriminating Unit 3.1 Derivation of Transfer Function Next, the transfer function from the wave to the command steering angle is examined. Based on the block diagram of the steering system shown in FIGS. 1 and 3, the steering machine is omitted for the sake of simplicity, the set course (reference course), parameter uncertainty (the error between the nominal value of the hull parameter and the actual value), and Steering angle offset, etc. shall be zero. In the figure, an estimator 20 is a quaternary estimator in which a wave model is incorporated, P is a hull model, K _P and K _D are feedback gains of the feedback loop 15 and represent a proportional gain and a differential gain, respectively. Yes.

  The transfer function of the fourth-order estimator shown in FIG.

As obtained. (2) Organizing the above formula and the above formula,
Is

Can be expressed as here

_Where K _sn , T _sn , and T _s3n are the nominal values of the hull parameters, respectively, the turning force gain and two hull time constants, and ω _wn and ζ _wn are the nominal values of the wave parameters, the center frequency and the attenuation frequency, respectively. Each is shown.
Also, the related formula of the command rudder angle is

Therefore, the transfer function from the wave to the command rudder angle is

As obtained. Here, G _p and G _e indicate transfer functions related to the steering system and the estimator, respectively.

The transfer function G _p has a characteristic polynomial of the secondary steering system.
become. Here, ζ _n and ω _n indicate the damping coefficient and the natural frequency of the steering system, respectively. The cut-off frequency ω _c of G _p is

Given in. Since wave frequency can be assumed to be a higher frequency side than omega _c, the gain of the high frequency than omega _c of G _p is

It becomes a constant value. However, K _D > 0.
On the other hand, the transfer function G _e is composed of a characteristic polynomial of a fourth-order estimator and a feedback gain.

become.

3.2 Transfer Function Characteristics FIG. 8 shows a numerical example of the gain characteristics of the transfer functions G _p and G _e . The figure shows K _sn = 0.047, T _sn = 61, T _s3n = 8.2, K _P = 1, ζ _n = 0.9, ω _n = 0.0246, K _D = 43.8, ω _e = Ρω _n , ρ = 3, ω _ew = ω _wn , ζ _e = ζ _ew = 1 / √2 and k _i , i = 1 to 4 are used to change the attenuation coefficient ζ _wn and the wave frequency ω _wn . .

It can be seen from FIG. 6 that G _P has a substantially constant value when ω> ω _e , so that the influence of the wave parameter can be determined by examining G _e . When the wave frequency ω _wn becomes small, the gain of G _e becomes 1 or more when ω> ω _e , and a region where disturbance is amplified is generated. That is, depending on the state of the waves, the disturbance component is amplified, so that the effect of disturbance removal performance cannot be expected. This situation is likely to occur in small ships where the steering frequency ω _n and the wave frequency ω _wn are close.

3.3 Wave State When the sea state is in the drought state, the wave parameter identification error increases because the SN ratio of the wave spectrum decreases in the wave identification unit. As a result, there may be a situation in which the effect of the above disturbance removal performance cannot be expected despite the state of wrinkles. In order to avoid this situation, the notch filter characteristic by the wave model is prevented from working in the case of the ridge or the equivalent state. That is, ζ _wn = ζ _ew = 1 / √2. At this time, the term related to the wave model in equation (28) is

The gain becomes −20 [dB / dec] when ω> ω _e regardless of ω _wn (ζ _wn = 1 / √2 in FIG. 8). For this reason, the gain of G _e of ζ _wn = 1 / √2 does not have a disturbance amplification region as in the case of ζ _wn = 0.1 and ω _wn = 0.15, so that the noise component is not amplified. .

Using this property, the wave state can be determined using the wave parameter identified by the wave identifying unit 28 and the transfer function G _e .

  Hereinafter, the operation of the wave determination unit 30 will be described with reference to FIG.

Constant setting section 30A is identified value zeta _wn of formula (28) by substituting the (notch filter there) and 1 / √2 (without notch filter), set parameters G _e, i.e. a constant, respectively. In particular,

Each parameter (constant) is determined.

Next, the evaluation amount calculation unit 30B calculates an evaluation amount for evaluating the wave state. The evaluation value of the wave state uses G _e with and without the notch filter, and the sum of squares corresponding to the command steering angle using the detected azimuth time-series data. The initial value is zero.

Here, J _on and J _off indicate evaluation amounts when the notch filter is present and absent, respectively. Note that the number of data n taking the sum is preferably the number of time-series data during a period of about 10 times the wave period.

Next, the attenuation coefficient determination unit 30C determines that the smaller one of the two evaluation quantities is a higher setting of the disturbance removal effect for the wave state at that time, and determines the attenuation coefficient used for the smaller one as the wave parameter. The center frequency ω _wn remains the identification value. That is, the wave coefficient is

Determined by.

Thus, wave judging unit 30, by performing the correction with respect to wave attenuation coefficient identified by wave identification unit 28, depending on the wave state, the transfer function G _e from wave to the instruction steering angle becomes large, Amplification of disturbance can be avoided.

It is a block diagram showing the whole structure of the automatic steering device for ships by this invention. It is a block diagram showing the structure of a secondary estimator. It is a block diagram showing the structure of a 4th-order estimator. It is a block diagram showing the structure of a wave identifier. It is a graph showing the example of a real spectrum showing extraction of identification data. It is a graph showing the spectrum of a wave model. It is the graph which calculated | required the asymmetry degree with respect to the spectrum of FIG. The transfer function G _n, is a graph showing the gain characteristic of G _e. It is a block diagram showing the whole structure of the ship automatic steering device.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Ship automatic steering apparatus 15 Feedback loop 20 Estimator 22 Wave identifier 26 Spectrum calculation part 28 Wave identification part 28A Normalization part 28B Identification data extraction part 28C Center frequency calculation part 28D Attenuation coefficient calculation part 30 Wave discrimination part 30A Constant setting Unit 30B evaluation amount calculation unit 30C attenuation coefficient determination unit

Claims (7)

A wave identifier for identifying the wave frequency and the wave attenuation coefficient from the detected direction ;
Detection direction
And the command rudder angle δc, and using the wave frequency and the wave attenuation coefficient identified by the wave identifier,
And estimated angular velocity
An estimator that outputs and
A feedback gain loop for outputting the command steering angle δc by applying a feedback gain to the deviation between the set course and the estimated direction and the estimated angular velocity,
In the marine vessel automatic steering apparatus having a steering machine that changes the steering angle according to the command steering angle δc output from the feedback gain loop , the wave identifier includes:
A spectrum calculation unit that converts time-series data of the detected orientation signal into spectrum data;
Identification data that is an envelope point cloud from spectral data
An identification data extraction unit that extracts (f _i represents a frequency and S _i − represents a spectrum value);
A center frequency calculator for determining the wave frequency f _wn from the center frequency based on the identification data;
Spectral values identifying data (S _i -) and, to minimize the evaluation value based on the deviation between the spectral values of wave model corresponding to the wave frequency based on the wave frequency determined by said center frequency calculation unit (S _i) An attenuation coefficient calculation unit that _sets the attenuation coefficient of the spectrum S of the wave model to be the wave attenuation coefficient ζ _wn ;
A marine vessel automatic steering apparatus comprising: The wave identifier further includes:
The ship automatic steering apparatus according to claim 1, further comprising a normalization unit that normalizes the spectrum data converted by the spectrum calculation unit.   The identification data extraction unit is an envelope point composed of a collection of envelope points at which a spectrum value increases or becomes equal to a frequency at which a maximum spectrum value is obtained in an area from an arbitrary start frequency in a predetermined area of spectrum data. Among the plurality of envelope points obtained on the lower frequency side and the higher frequency side than the frequency at which the maximum spectrum value is obtained, among the envelope points that have the largest number of envelope points The automatic steering apparatus for a ship according to claim 1 or 2, wherein an envelope point group closest to a frequency at which a start frequency takes a maximum spectral value is extracted as identification data. It said center frequency calculation unit, identification data (f _i, S _i -) to spectral values as the upper threshold value S _H or more values in the identified data to,
The marine automatic steering apparatus according to any one of claims 1 to 3, wherein the wave frequency f _wn is obtained by: For the identification data (f _i , S _i −) in which the spectral value is equal to or higher than the lower threshold value S _L in the identification data, the attenuation coefficient calculation unit calculates a wave corresponding to the spectral value and the frequency. The sum of squares J _ζ of deviation from the spectrum value S of the spectrum of the model is the minimum value.
The marine automatic steering apparatus according to any one of claims 1 to 4, wherein the wave attenuation coefficient ζ of the wave model satisfying the condition is set to the wave attenuation coefficient ζ _wn . A wave identifier for identifying the wave frequency and the wave attenuation coefficient from the detected direction ;
Detection direction
And the command rudder angle δc, and using the wave frequency and the wave attenuation coefficient identified by the wave identifier,
And estimated angular velocity
An estimator that outputs and
A feedback gain loop for outputting the command steering angle δc by applying a feedback gain to the deviation between the set course and the estimated direction and the estimated angular velocity,
In the marine vessel automatic steering apparatus having a steering machine that changes the steering angle according to the command steering angle δc output from the feedback gain loop , the wave identifier includes:
A wave identification unit for identifying the wave frequency and the wave attenuation coefficient from the spectral data of the time-series data of the detected direction signal;
The wave state is determined by using the time-series data, the wave frequency identified by the wave identification unit f _wn = ω _wn / 2π, and the wave attenuation coefficient ζ _wn, and correcting the wave attenuation coefficient suitable for the wave state And
A marine vessel automatic steering apparatus comprising: The wave determination unit is a transfer function G _e from the wave to the command steering angle.
In contrast, G _e when the wave attenuation coefficient identified by the wave identification unit is substituted for ζ _wn.
And G _e when 1 / √2 is substituted into ζ _wn.
A constant setting unit for obtaining each constant of
Obtained by the constant setting unit
Against
An evaluation amount calculation unit for _obtaining respective evaluation values J _on and J _off represented by:
Wave attenuation coefficient obtained by correcting the value of ζ _wn used for the smaller one of the evaluation values J _on and J _off
Adopted as
An attenuation coefficient determining unit,
The marine vessel automatic steering apparatus according to claim 6, further comprising: