JPH03282287A - Depth detection method and depth detection device for underwater objects - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method and apparatus for detecting and positioning underwater targets using ultrasonic waves.

[Prior Art] As described in Japanese Unexamined Patent Application Publication No. 61-110074, a conventional device provides directivity in both horizontal and vertical directions using a receiver arrangement and a phasing means, and determines the azimuth and depression angle of a target. It is used to measure. Further, as described in Japanese Patent Application Laid-Open No. 62-217173, a plurality of receivers are arranged vertically and a sound ray diagram is also used to determine the depth and horizontal distance of an unknown sound source.

[Problems to be Solved by the Invention] In the above-mentioned conventional techniques, in order to obtain the depth of the target, the vertical dimension of the receiver must be increased or a plurality of receivers must be installed in the vertical direction. It is necessary to sharpen the directivity in the vertical direction by arranging them. In other words,
Conventionally, devices that attempt to detect the depth of an underwater object in addition to its direction and distance have had to use larger directivity in both horizontal and vertical directions.

SUMMARY OF THE INVENTION The present invention aims to provide a method and apparatus for determining the depth and depression angle of an underwater object without particularly requiring vertical directivity for depth detection during position detection of the underwater object. do. The purpose of this is to avoid increasing the size of the device.

[Means for Solving the Problems] In order to achieve the second objective, the depth detection method of the present invention has the following features:
The orientation of the underwater object on the horizontal plane and the direct distance to the object are measured many times, and using these measurement data and the position data of the mother ship at each measurement, the observation matrix is calculated, the Kalman gain is calculated, It is characterized by performing arithmetic processing for updating the state vector and updating the covariance matrix based on this.

In order to achieve the above object, the depth detection device of the present invention includes a means for power amplifying a pulse-modulated transmitter signal, converting it into an acoustic signal via a transducer, and emitting it into the water, and a means for emitting the signal into the water from an underwater object. means for receiving the reflected signal by the transducer, means for amplifying and phasing the received signal and converting it into azimuth data by the azimuth calculator, and a timing signal obtained from the pulse modulation output and the reception output. means for converting the data into direct distance data using a distance calculator; means for calculating the mother ship position using the mother ship position calculator using the speed and heading data of the mother ship; The present invention is characterized by comprising an arithmetic processing means for obtaining an output of a position component of an underwater object by applying a Kalman filter.

[Function] In the following, it is believed that the function will be effectively clarified by explaining the embodiments including mathematical treatment, but here, the function will be explained in particular while outlining the depth detection method of the present invention. do.

Position coordinates X1 of the desired underwater object (hereinafter referred to as target object)
, yll, and Zll are elements of the state vector. on the other hand,
What is observed is the orientation θ of the target object on the horizontal plane and the direct distance r to the target object, which are elements of the observation vector. Furthermore, there is an observation matrix that provides a relational expression between observation vectors and state vectors. Therefore, the depth detection method of the present invention is to go to the bottom, observe the observation vector, and estimate the state vector using the observation matrix.

If there are no errors in the observed data, and even if there are errors, if the results obtained using the data are within the required precision, the above can be achieved using the results of a small number of observations depending on the number of unknowns. It becomes possible to find the state vector from the relationship.

The present invention does not directly measure the angle of depression toward the target object.
Furthermore, since we are trying to find the positional components of X1, yl, and Zl above on the premise that the observation of the direction and direct distance of the target object inevitably includes errors, we do not accept large errors in the initial state vector. There must be. In other words, in this state, the degree of dispersion of errors in the state vector, so to speak, the covariance value is large. Therefore, the state vector is updated one by one by measuring the direction and direct distance of the target object many times. In this process, the covariance value decreases successively, and ultimately the state vector is brought closer to a certain convergence value.

To this end, first assume the initial value of the state vector, and also determine the covariance matrix that provides the covariance of the state vector. Using these values as initial values, the assumed state vector and covariance matrix are modified in accordance with the difference between the observed vector derived from the assumed state vector and the actually observed data. This correction value is a value proportional to the Kalman gain, and therefore, the state vector and covariance matrix are successively corrected and updated using the Kalman gain. Then, such updates are performed many times, that is, each time observation data is obtained. Here, correction using the Kalman gain allows the state vector and covariance matrix to be corrected to finally converge to a certain value if the error in the observed data is within a certain range.

In the above explanation, the position data of the mother ship has been omitted for convenience of explanation, but it goes without saying that the position data of the mother ship is also used in the correction process that is performed every time observation data is obtained.

The above method makes it possible to estimate the position of the target object. In other words, it becomes possible to obtain the depth of the target object and the depression angle without particularly requiring directivity in the vertical direction.

The Kalman filter in the depth detection device of the present invention calculates the observation matrix and Kalman gain by inputting the azimuth data, direct distance data, and mothership position data, and based on this, when updating the state vector, the covariance matrix This is an arithmetic processing means that updates the target object and outputs the position component of the target object. Therefore, according to the depth detection device of the present invention, by using the depth detection method of the present invention, it is possible to detect the depth of a target object without requiring directivity in the vertical direction. Therefore, in the device of the present invention, only the horizontal directivity performance of the transducer is required, so the device is simple and it is possible to avoid increasing the size of the device despite depth detection.

〔Example〕

FIG. 1 shows an embodiment of the hardware configuration of a depth detection device according to the present invention. In FIG. 1, the signal from a transmitter 1 is pulse-modulated by a tone burst signal generator 2, power amplified by a transmitter 3, and sent to a transducer 5 via a transmitter/receiver switch 4, where it is converted into an acoustic signal and transmitted underwater. is radiated to. This sound is reflected by the target object and is received by the transducer 5 with a delay of a time corresponding to the direct distance to the target object. Here, the signal is again converted into an electric signal, and this signal is amplified by a receiver 6 via a transmitter/receiver switch 4 and sent to a phaser 7.

The phased signal is converted into orientation data by the orientation calculator 8. On the other hand, timing signals obtained from the tone burst generator 2 and receiver 6 are sent to a direct distance calculator 9, where they are converted into direct distance data. Further, data on the speed and traveling direction of the mother ship are sent from the speed detector 10 and gyro 11 of the mother ship to the mother ship position calculator 12, where the mother ship position is calculated every moment. The above three types of data (azimuth, direct distance, and mothership position) are input to the Kalman filter 13, which uses the analysis values (target position x,
y and Z components).

FIG. 2 shows a flowchart for calculating an analytical value using a Kalman filter, and also shows arithmetic processing procedures in the depth detection method of the present invention. As shown in FIG. 2, Kalman filter processing is divided into two parts: initial processing and normal processing. First, in initial processing, the state vector X and covariance matrix P are initialized. In the present invention, the state vector X represents the x, y, and Z components of the 1'' target position, and consists of elements Xm, ym, and Z, respectively. Further, the covariance matrix P represents the distribution of each element of the state vector and is a 3×3 matrix. Normal processing is performed when observation data is obtained. First, an observation matrix H is calculated based on observation data, and then a Kalman-Keyne calculation is performed. The state vector X and covariance matrix P are updated based on H and K.

Every time observation data is obtained, the above normal processing is performed,
X and P are improved one after another, and finally a sufficiently accurate state vector X, that is, the target position, is obtained. Generally, a Kalman filter also requires a state transition matrix F, but in the present invention, since the target is stationary, the state does not change. Therefore, F becomes a unit matrix and is unnecessary.

FIG. 3 is a diagram showing the positional relationship between the mother ship and the target object. A mathematical model is constructed using FIG. 3, and Kalman filter processing will be explained below. In the coordinate system in FIG. 3, the X-axis indicates the east-west direction, the y-axis indicates the north-south direction, and the Z-axis indicates the depth direction. The direction O is measured clockwise with north as the reference. In the figure, the target position is (xm+y□zz), which is defined as the state vector X.

It is. Equations (3) and (4) are the state vector
and the observation vector Y.

Since the observation matrix H is x, what can be observed is the target's orientation 0 and direct distance r, so this is set as the observation vector Y.

Then h hll, hl, , h, 1, hl, and 2. Since the mother ship moves freely on the water surface, the position of the mother ship is (Xe
+'l-+O). Also, since the coordinates of the point directly above the target are (xm, ym+o), the orientation θ of the target as seen from the mother ship is -+XsX.

θ = begging
... (3) ym-y・ Also, the direct distance r between the mother ship and the target object is 0 + y, -y.

+ z m (4) x m (7) (8) Since the target is immobile, the state transition matrix F is a unit matrix and does not need to be taken into consideration and is 1°. Here, σ ″ is the estimated value of the variance of the observation direction O, and σ33′ is the estimated value of the variance of the direct distance r.

Now, when the observation data is input, the observation matrix H is calculated using equations (7) to (12). Next, Kalman gain K is calculated using equation (16).

K=PH” [HPHT+R] ...(1
6) Here, I(T represents the transposed matrix of H. Using this Kalman gain K, the state vector X is updated according to equation (17).

y, and Z, are shown. Other elements are Xwa, '1
wa and 2. It shows the covariance between each other. Therefore p,
This matrix is a symmetric matrix.

The observation error estimation matrix R is 22, xn is the state vector before updating, and Xn+ is the state vector after updating. Also θ(X,) and r(
Xn) is the estimated direction and estimated direct distance obtained using the state vector xn before updating.

The covariance matrix P is updated according to equation (18).

Pn+, =Pn-KHPn =・(18)
Here, Pn is the covariance matrix before updating, and p n+ is the covariance matrix after updating.

The above processing is compared with the normal processing shown in the flowchart of FIG. 2 as follows. The calculation of H in the flowchart refers to the second (6)th, or specifically, the (7) to (1st)
2) It refers to the calculation of equations.

Similarly, the calculation of K refers to the calculation of equation (16), and can be obtained by matrix multiplication, addition, and matrix inverse operation. The update of X is based on equation (17). Here, θ and r are the observation data themselves, and θ(Xn) is the (3rd)
The equation, r (X,), is determined by equation (4). Finally, the update of P is expressed by equation (18) and can be calculated by matrix multiplication and subtraction.

If the method described above is implemented, the target depth can be obtained only from the azimuth data obtained from the transducer, the direct distance data, and the mother ship position data obtained from the mother ship. Also, the mother ship only moves on the water surface and does not need to move vertically.

Examples of experiments according to the present invention are shown below in the flowchart in Figure 2 and in Figure 3.
This will be explained using the mathematical model in Figure 4 and the experimental results in Figure 4. In Fig. 3, the mother ship moves in a circle starting from the origin and observes the direction O and the direct distance r every 10 seconds. Note that the observed data includes some errors, and this error is assumed to be normally distributed. θ
The error is the mean value O standard deviation 0.002 radian (approximately 0.
11 degrees) Also, the error of r was set to an average value O standard deviation of 0.02 m. In addition, the initial values of the state vector are X, "r,
Sinθ.

ym=r,,cosθ.

Zm= lom And so. Here Oo and r. is the initial observed value.

Depth 2. Since there is no valid value for this, we simply chose a value that is far from the true value (-50m).

The value given to the observation error estimation matrix R is σ, , =0. 2 radians (approximately 11 degrees) σ, , = 2 m. The initial value of the covariance matrix is 100 for both pl l pl m and p81.
It was set to 0. This means that the error distribution of the target position is initially assumed to be 5 m (approximately 31.6 m). The initial values of other elements of the covariance matrix were O.

When an experiment was conducted under the above conditions, the results shown in FIG. 4 were obtained. The true value of the target position is x m =] 000m,
ym = 500 m, Z, knee 50 m, while for X and y, values close to the correct answer were naturally obtained immediately, and for Z, which cannot be directly observed, the 11th measurement was 100 m. A tendency towards the true value can be seen after about 150 seconds, and after 150 seconds it almost converges to the true value and has remained at this value ever since.

[Effects of the Invention] According to the present invention, the depth of the target can be obtained without directly measuring the vertical angle (inclination angle) of the target. Since the transducer does not need to have vertical directivity, the vertical dimensions of the transducer can be reduced. It is also possible to measure the depth of the target and the angle of depression using a conventional transducer that can only measure the azimuth and direct distance. Originally, the directivity in the vertical direction was intended to be broad in order to be less susceptible to the vibration of the mother ship, but the present invention also meets this requirement.

[Brief explanation of the drawing]

Fig. 1 shows an example of the hardware configuration of the depth detection device of the present invention, and Fig. 2 shows a flowchart for calculating an analytical value using a Kalman filter, and also shows the calculation processing procedure in the depth detection method of the present invention. Figure shown. FIG. 3 is a mathematical model diagram for explaining the present invention, and FIG. 4 is a chart showing experimental results according to an embodiment of the present invention. Code explanation ■... Transmitter 2... Tone burst signal generator 3... Transmitter 4... Transmit/receive switch 5...
Transducer/receiver 6... Receiver 7... Phaser
8... Direction calculator 9... Direct distance calculator
10... Speed detector 11... Gyro 1
2...Mother ship position calculator 13...Kalman filter

Claims (1)

[Claims] 1. In a method for detecting the depth of an underwater object by measuring from an observation point on a mother ship and calculating the measured data, the measurement includes the direction of the underwater object on a horizontal plane and , the direct distance to the above object is measured many times, and using these measurement data and the position data of the mother ship at each measurement, the observation matrix is calculated, the Kalman gain is calculated, and the state vector is updated based on this. A depth detection method for an underwater object characterized by performing calculation processing for updating a covariance matrix. 2. In a device that detects the depth of an underwater object by emitting an acoustic signal to an underwater object and analyzing the reflected signal from the underwater object, the pulse-modulated signal of the transmitter is power amplified and the acoustic signal is transmitted via a transducer. means for converting it into a signal and radiating it into the water; means for receiving the reflected signal from the underwater object by the transducer; and means for amplifying and phasing the received signal and converting it into azimuth data by an azimuth calculator; Means for converting into direct distance data by a direct distance calculator using the timing signal obtained from the above pulse modulation output and the above received output, and calculation of the mother ship position by the mother ship position calculator using the speed and traveling direction data of the mother ship. A depth detection device for an underwater object, comprising: means for inputting the azimuth data, direct distance data, and mother ship position data into a Kalman filter to obtain an output of a position component of the underwater object.