Information entropy-based deep sea very-close water surface underwater target identification methodTechnical Field
The invention is used in the field of deep sea very close target depth estimation, and particularly relates to a deep sea very close water surface underwater target identification method based on information entropy.
Background
At present, the prior art has a certain achievement for realizing target ranging and depth fixing by utilizing a horizontal array in a deep sea environment. For example, CN115825965A proposes a three-dimensional positioning method of a direct sound area target based on a deep sea submarine double horizontal array, firstly, obtaining a target azimuth estimation through beam forming of the double horizontal array, then, tracking the maximum value of the beam output power of the double horizontal array and tracking azimuth angle, combining the positions of the two arrays to perform target distance estimation, finally, inputting the estimated target distance, a preset sound source depth and marine environment parameters obtained by actual marine experimental field measurement into a ray model, obtaining a time delay difference between a direct path and a sea primary reflection path, constructing a copy sound pressure field, and establishing a depth estimation fusion ambiguity function of the double horizontal array, wherein the maximum value is the depth of the target. Since CN115825965A is a direct-utilized beamforming to estimate the sound source azimuth, for very close range sound sources, the sound waves do not meet the plane wave assumption, and the selection of the reference sound speed for beamforming may cause a larger direction finding error, which affects the subsequent estimation of the sound source depth. In addition, as the sound source considered by the method is a pulse sound source, the actual signal is generally continuous, and the arrival time delay of the direct wave is inconvenient to obtain in the case.
Simulation of the prior art shows that when a sound source approaches the water surface, the phenomenon that the estimation of the depth of the sound source is interfered by high side lobes occurs.
Disclosure of Invention
Aiming at the problem that the depth estimation of a target is fuzzy under the deep sea very close range, the invention provides an information entropy-based deep sea very close range water surface underwater target identification method. On the basis, the invention designs a statistic named information entropy, wherein the statistic is obtained by extracting a column corresponding to the position with the maximum intensity in the ambiguity function, designing a unique calculation method to calculate the entropy value of the column, giving an information entropy threshold according to a channel, and finally identifying an underwater target on the water surface by utilizing the size of the information entropy. The invention only needs to use a single horizontal array, and the considered signal is a line spectrum signal, thereby breaking through the restriction that only transient signals are considered before. In addition, the method is used for researching how to roughly estimate the target depth in the deep sea in the very short distance by utilizing the horizontal array, and the target depth interval is roughly estimated in the very short distance.
In order to achieve the above purpose, the present invention provides the following technical solutions:
The method for identifying the deep sea very close water surface underwater target based on the information entropy comprises the following steps:
s1, receiving sound source data of a deep sea very close distance position by utilizing an N-element horizontal receiving array;
s2, setting a measuring field, and calculating sound pressure of each array element of the array by using a virtual source method modelThe measured angle is obtained by a beam forming methodThereby obtaining the sound pressure cross spectrum matrix;
S3, assuming a sound source depth, calculating sound wave arrival angles under different sound source distances and directions, and carrying out the measurement anglesProcessing the wave arrival angle matching field to obtain a sound source distance-azimuth ambiguity function;
S4, extracting points with the intensity larger than b in the sound source distance-azimuth ambiguity function, and performing fitting treatment and translation treatment on the points to obtain a sound source distance-azimuth distribution curve cluster;
s5, enabling the distance and the azimuth of the sound source to be distributed on the sound source distance-azimuth distribution curve cluster, setting the scanning range of the sound source depth and configuring a copying field;
S6, calculating complex sound pressure received by the array under the copy field, and performing matching field processing based on sound pressure signals to obtain a sound source depth-distance ambiguity function;
and S7, finding out a column corresponding to the maximum value point of the sound source depth-distance ambiguity function, calculating the information entropy of the column, and judging whether the target is a water surface target or an underwater target according to the information entropy.
Further, in step S1, the sound field of the sound source data is composed of a direct wave of the sound wave and a sea surface reflected wave.
Further, step S2 is to calculate the sound pressure of each array element of the array by using the virtual source method modelThe method specifically comprises the following steps:
S201, obtaining complex sound pressure at a horizontal distance r and a depth z through a virtual source method model, wherein the expression is as follows:
,
In which the time factor is ignoredThe negative sign is to meet the sea surface pressure release boundary condition:, Taking sound velocity at a sound source as a medium wave number, wherein f is the frequency of the sound source, and r and z respectively represent the horizontal distance and depth from an observed point to the sound source;
s202, according to the geometric relation of the virtual source method model:
,,
When (when)In the time-course of which the first and second contact surfaces,AndIs approximately equal toThen:
,,
S203, supposing denominator in sound pressure expressionAndCan be replaced by R, and the exponential function is developed into trigonometric function, sound pressureThe simplification is as follows:
;
S204, if the sound pressure formula is applied to the whole sea area, a virtual source method sound field can be obtained;
s205, obtaining complex sound pressure of an nth array element according to a virtual source method sound field:
,
s206, the array received complex sound pressure vector may be expressed as:
。
further, the measured angle is obtained by a beam forming methodThe method specifically comprises the following steps:
s211, array receiving signal expression:
,
Wherein, theIs the sound source signal spectrum;
S212, complex sound pressure cross spectrum density matrix of arrayThe expression is:
,
S213, representing the output power expression of the beamformer in the frequency domain:
,
Wherein, theIs a beamforming weight vector;
S214, calculateThe angle corresponding to the maximum value of (2) is。
Further, the step S3 specifically includes the following steps:
s301, the arrival angle calculation formula is as follows:
,
Wherein, the、The pitch angle and glancing angle of the sound wave under the horizontal distance r and depth z of the corresponding sound source are respectively;
S302, obtaining a glancing angle calculation formula according to the virtual source method model:
;
S303, assuming that the sound source depth for calculating the angle of arrival is 0 toAny one of the values of r and azimuthRespectively between 0 and 0And-90 DEG to 90 DEG, and obtaining the sound wave arrival angles of the sound sources on the array in the distance azimuth according to the arrival angle and glancing angle formulas;
S304, distance-azimuth ambiguity functionThe calculation formula is as follows:
。
further, the method for obtaining the sound source distance-azimuth distribution curve cluster comprises the following steps:
Extracting points with intensity greater than b from the distance-azimuth ambiguity function, and fitting the points to obtain a sound source distance-azimuth distribution curve;
For the sound source distance-azimuth distribution curvePerforming translation treatment to obtain a sound source distance-azimuth distribution curve cluster:
,
Wherein, theIs the translation amount.
Further, the method for obtaining the sound source depth-distance ambiguity function in step S6 includes the following steps:
S601, obtaining the depth range of the sound source from 0 to 0 through a virtual source method modelDistance/bearing complianceDistributed array receiving complex sound pressure;
S602, a depth-distance or depth-azimuth ambiguity function expression based on a minimum variance processor is as follows:
;
s603, carrying out normalization and logarithmic processing to obtain:
;
S604、 and the abscissa corresponding to the maximum point of the medium intensity is the estimated distance of the sound source.
Further, the calculation method of the information entropy in step S7 includes:
take out S603And (3) a column vector corresponding to the maximum point of the medium intensity, and according to the minimum value of a given vector element of the deep sea channel, setting the value of a point smaller than or equal to a as a, wherein the information entropy has the following calculation formula:
,
wherein the total number of elements of the vector is I, and the probability of the ith element is;
The entropy will be used to identify the surface underwater target.
Further, according to the change rule of the information entropy when the targets are respectively positioned at different directions, different distances and different depths, a depth-distance two-dimensional pseudo color map and a depth-direction two-dimensional pseudo color map of the related information entropy are to be given;
According to the actual channel, giving out the threshold value of the information entropy of the target for identifying the underwater target on the water surface at different directions and different distances;
And when the information entropy is larger than the threshold value, the target is a water surface target, and otherwise, the target is an underwater target.
Compared with the prior art, the invention has the beneficial effects that:
The invention provides an information entropy-based deep sea very close water surface underwater target identification method, which comprises the steps of obtaining arrival angles of sound sources at array positions under different distance orientations by assuming a sound source depth and combining a glancing angle calculation formula, carrying out matching output with beam forming measurement angles to obtain sound source distance-orientation distribution curve clusters, then combining a depth scanning range and the curve clusters to design a copy field, and carrying out matching field processing based on sound pressure signals to obtain an ambiguity function. And finally, extracting a column corresponding to the maximum value point of the ambiguity function, calculating the information entropy according to the information entropy calculation flow provided herein, and identifying the underwater target on the water surface according to the value, thereby solving the problem of estimating the ambiguity of the depth of the target in the deep sea in the very close range.
The basic principle and the implementation scheme of the invention are verified by computer numerical simulation, and the result shows that the deep sea very close water surface underwater target identification method based on the information entropy can judge whether the target is a water surface target or an underwater target.
In order to more clearly illustrate the structural features and efficacy of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and examples.
Drawings
FIG. 1 is a flow diagram of a method for identifying a deep sea very close water surface underwater target based on information entropy;
FIG. 2 is a schematic view of a uniform linear array deployed horizontally at the offshore bottom;
FIG. 3 is a cross-sectional view of the sound velocity in deep sea;
FIG. 4 is a deep sea propagation loss map (sound source depth 50m, frequency 100 Hz);
FIG. 5 is a schematic diagram of virtual source method model geometry;
FIG. 6 is a graph of sound field propagation loss (sound source depth 50m, frequency 100 Hz) by virtual source method;
FIG. 7 is a sound source distance-azimuth distribution curve and curve cluster (circle position represents sound source real position; FIG. 7 a shows sound source distance-azimuth distribution curve before translation; FIG. 7 b shows sound source distance-azimuth distribution curve cluster after translation);
fig. 8 shows matching field outputs based on sound pressure signals at different sound source depths ("o" represents a true position, "×" represents an estimated position, and a solid line box represents column vectors required for calculating information entropy; fig. 8 shows matching field outputs of sound source depths 5m, 20, 100, 200, respectively);
Fig. 9 shows the relationship between information entropy and target distance, depth, and azimuth (fig. 9 shows a two-dimensional information entropy depth-distance map and fig. 9 shows b two-dimensional information entropy depth-azimuth map).
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Specific implementations of the invention are described in detail below in connection with specific embodiments.
Example 1
Referring to fig. 1, the invention provides a deep sea very close water surface underwater target identification method based on information entropy, which is characterized by comprising the following steps:
s1, receiving sound source data of a deep sea very close distance position by utilizing an N-element horizontal receiving array;
s2, setting a measuring field, and calculating sound pressure of each array element of the array by using a virtual source method modelThe measured angle is obtained by a beam forming methodThereby obtaining the sound pressure cross spectrum matrix;
S3, assuming a sound source depth, calculating sound wave arrival angles under different sound source distances and directions, and carrying out the measurement anglesProcessing the wave arrival angle matching field to obtain a sound source distance-azimuth ambiguity function;
S4, extracting points with the intensity larger than b in the sound source distance-azimuth ambiguity function, and performing fitting treatment and translation treatment on the points to obtain a sound source distance-azimuth distribution curve cluster;
s5, enabling the distance and the azimuth of the sound source to be distributed on the sound source distance-azimuth distribution curve cluster, setting the scanning range of the sound source depth and configuring a copying field;
S6, calculating complex sound pressure received by the array under the copy field, and performing matching field processing based on sound pressure signals to obtain a sound source depth-distance ambiguity function;
and S7, finding out a column corresponding to the maximum value point of the sound source depth-distance ambiguity function, calculating the information entropy of the column, and judging whether the target is a water surface target or an underwater target according to the information entropy.
The invention only needs to use a single horizontal array, and the considered signal is a line spectrum signal, thereby breaking through the restriction that only transient signals are considered before. In addition, the method is used for researching how to roughly estimate the target depth in the deep sea in the very short distance by utilizing the horizontal array, and the target depth interval is roughly estimated in the very short distance.
Example 2
Referring to fig. 1-9, the method for identifying the deep sea very close water surface underwater target based on information entropy comprises the following steps:
step one, calculating a sound field
Assuming that the sound source has a true depth ofThe true distance isThe true azimuth isAnd transmitting a narrow-band signal, and setting an N-element horizontal uniform linear array at a deep sea very close distance position. Because the considered environment is very close, the sound field is mainly composed of direct waves of sound waves and sea surface reflected waves, a map of deep sea propagation loss is obtained by using KRAKENC, as shown in fig. 4, a directivity pattern with alternately appearing maxima and minima is shown in the propagation loss near the sound source, which is a typical deep sea Lloyd mirror interference pattern, and each band-shaped stripe corresponding to the minimum value of the propagation loss is called a wave beam. The sound field can be explained by a virtual source method, the sea surface is regarded as a mirror reflection surface by the virtual source method, and the geometric schematic diagram of the model of the virtual source method is shown in figure 5.
Obtaining complex sound pressure at a horizontal distance r and a depth z through a virtual source method model:
,
In which the time factor is ignoredThe negative sign is to meet the sea surface pressure release boundary condition;, taking sound velocity at a sound source for the wave number of a medium, and taking f as the frequency of the sound source according to the geometric relationship in fig. 5:
,;
When (when)In the time-course of which the first and second contact surfaces,AndIs approximately equal toThen:
;
;
further assume that denominator in the sound pressure expressionAndCan be replaced by R, and the exponential function is developed into trigonometric function, sound pressureThe simplification is as follows:
;
If the sound pressure formula is applied to the whole sea area, a new sound field can be obtained, namely the virtual source method sound field. Likewise, the sound field propagation loss by the virtual source method is plotted as shown in fig. 6. As can be readily seen from fig. 4 and fig. 6, in the very close range, the virtual source method sound field and KRAKENC sound fields are almost identical, and from the simulation point of view, it is reasonable to explain the sound field of the area by the virtual source method, and the related theory derivation is performed on the basis of the virtual source method model.
On the basis, the complex sound pressure of the nth array element can be obtained:
,
The array received complex sound pressure vector can be expressed as:
。
Step two, calculating to obtain a measurement angle
Taking a noise-free environment as an example, the array received signal can be written as:
,
Wherein, theThe complex sound pressure cross spectrum density matrix of the array can be obtained as follows:
,
The output power of the beamformer is represented in the frequency domain as:
,
Wherein, theIs a beamforming weight vector. For example, the weight vectors for conventional beamforming are:
,
Wherein, theFor the reference wavelength of beam forming, d is the array element spacing, calculateThe angle corresponding to the maximum value of (2) is。
Step three, calculating the angle of arrival of the sound wave
Firstly, giving a calculation formula of the angle of arrival:
,
Wherein, the、The pitch angle and glancing angle of the sound wave under the horizontal distance r and depth z of the corresponding sound source are respectively;
taking grazing angle as an example, a grazing angle calculation formula is obtained according to a virtual source method model shown in fig. 5:
,
Assume that the sound source depth for calculating the angle of arrival is 0 toAny one of the values of r and azimuthRespectively between 0 and 0And-90 DEG to 90 DEG, obtaining the sound wave arrival angles of the sound sources on the array in the distance azimuth according to the arrival angle and glancing angle formulas, and finally obtaining the distance-azimuth ambiguity function:
。
Step four, handleAnd extracting points with medium intensity larger than b, and fitting the points. By this method, the sound source distance-azimuth distribution curve can be obtained;
In addition, for reducing errorPerforming translation treatment to obtain a sound source distance-azimuth distribution curve cluster:
,
Wherein, theIs the translation amount.
And fifthly, performing matching field processing based on the sound pressure signals. Obtaining the depth range of a sound source from 0 to 0 through a virtual source method modelDistance/bearing complianceDistributed array receiving complex sound pressure. The depth-distance or depth-azimuth ambiguity function based on the minimum variance processor is given below:
,
The method is characterized by comprising the following steps of normalization and logarithmization treatment:
,
and the abscissa corresponding to the maximum point of the medium intensity is the estimated distance of the sound source.
Step six, calculating information entropy
Taking outAnd (3) a column vector corresponding to the maximum point of the medium intensity, setting the value of a point smaller than or equal to a as a according to the minimum value of a given vector element of the deep sea channel, and finally utilizing an entropy calculation formula:
,
wherein the total number of elements of the vector is I, and the probability of the ith element isThe statistics are used to identify the underwater targets on the water surface;
Finally, the change rule of the information entropy when the targets are respectively positioned at different directions, different distances and different depths is given, and especially the relation between the targets and the depth of the sound source is researched. The depth-distance two-dimensional pseudo color map and the depth-azimuth two-dimensional pseudo color map of related information entropy are to be given, and thresholds of the information entropy of the target for identifying the underwater target on the water surface in different azimuth and different distances are given according to actual channels. And when the information entropy is larger than the threshold value, the target is a water surface target, and otherwise, the target is an underwater target.
Simulation example 2
In order to further verify the target recognition method of the present invention, the present invention provides a simulation embodiment, and referring to fig. 1-9 and tables 1-3, the deep sea very close water surface underwater target recognition method based on information entropy of the present invention is mainly used for judging the approximate interval of the sound source depth under the deep sea very close distance, and specifically includes the following steps:
The first step, simulating standard environment parameters are as follows, sound source frequency 100Hz, sound source depth range of 0-200 m, sound source distance range of 0-2 km, sound source azimuth angle range of-70 DEG to-40 DEG for the array. The array element spacing of the uniform linear array is 7.5m, the array element number is 35, the array depth is 4950m, the signal to noise ratio is 30dB, the submarine longitudinal wave sound velocity is 1650m/s, and the submarine density is 1.6Submarine longitudinal wave attenuation 0.25Amount of translationThe deep sea sound velocity profile is shown in fig. 3. Firstly, setting a measuring field according to standard environmental parameters, and calculating sound pressure of each array element of an array by using a virtual source method modelThe measured angle is obtained by a beam forming methodFinally, a sound pressure cross spectrum matrix is obtained。
Step two, assuming that the sound source depth is any value between 0 and 300m, calculating the sound wave arrival angles under different sound source distances and orientations between 0km and 10km and-90 DEG and 90 DEG by taking 100m as a distance scanning interval and 1 DEG as an azimuth scanning intervalThe sound source distance-azimuth ambiguity function is obtained by:
。
Third step, extractingAnd fitting the points with the medium intensity larger than 0, wherein the fitting order is 20. Thus, the sound source distance-azimuth distribution curve can be obtainedThen carrying out translation treatment to obtain a sound source distance-azimuth distribution curve cluster。
And fourthly, establishing a copying field to finish matching field processing based on the sound pressure signal. The distance and the azimuth of the copying field sound source are distributed inOn the basis of the sound source depth scanning range of 20m and the interval of 1m, the complex sound pressure received by each array element of the array is obtained. The matching field uses a minimum variance processor to obtain a depth-distance ambiguity function by:
,
,
Finally findAnd setting the minimum value of vector elements to be-10 according to the deep sea channel by utilizing the information entropy calculation flow provided herein, enabling all elements smaller than the minimum value of the columns to be equal to the minimum value, and finally solving the information entropy by combining an entropy calculation formula.
Fifthly, respectively discussing the change rule of the information entropy when the targets are positioned in different directions, different distances and different depths, and giving a depth-distance two-dimensional pseudo color map and a depth-direction two-dimensional pseudo color map of the related information entropy. And finally, providing an information entropy threshold value for identifying the underwater targets on the water surface by the method under the corresponding simulation environment. The relevant simulation results are shown in fig. 9.
From the analysis of the simulation results, the information entropy is mainly influenced by the depth of the sound source. With the sound source depth 20m as a boundary, when the sound source depth continues to increase, the information entropy is greatly reduced. This property will be used to identify underwater objects on the surface.
According to the simulation result, in the simulation environment, when the information entropy is not less than 3, the target is a water surface target with depth less than 20m, and conversely, is an underwater target.
Step six, fixing the target azimuth, simulating and analyzing the identification effect of the method on the target under different distances, and fixing the target distance, simulating and analyzing the identification effect of the method on the target under different azimuths. The method is verified to be effective in identifying the very-close water surface underwater targets.
Table 1 relevant symbols herein
TABLE 2 information entropy (non-bolded) and identification result (bolded) of targets at fixed azimuth (-60 °) and different distances
TABLE 3 information entropy (non-bolded) and identification result (bolded) of targets at fixed distance (1.5 km) in different orientations
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.