Information processing device, information processing method, and information processing program

Abstract

The objective is to provide an information processing device, an information processing method, and an information processing program that can reduce the number of times propagation loss is calculated. [Solution] The information processing device comprises a group division unit that divides multiple search points within a target sea area into multiple groups containing similar sound velocity distributions and calculates a representative sound velocity distribution for each group, and a propagation loss calculation unit that calculates the propagation loss for each group based on the representative sound velocity distribution.

Description

[Technical Field]

[0001] The present invention relates to an information processing device, an information processing method, and an information processing program for calculating the propagation loss of sound waves from an object in water to a receiver. [Background technology]

[0002] It is known that the presence of marine organisms and other objects can be detected by receiving sound waves emitted from these objects. Sound waves originating from an object propagate through the ocean, undergoing repeated reflections and refractions, before reaching the receiver. Since the propagation path depends on the sound velocity distribution in the ocean (the relationship between depth and sound velocity), the optimal depth of the receiver for detecting the object at the longest possible distance (hereinafter referred to as "optimal depth") also depends on the sound velocity distribution.

[0003] When searching for an object while moving a receiver, the sound velocity distribution differs at each search point in the target sea area, and therefore the optimal depth also differs at each point. Consequently, in order to efficiently carry out the search, it is necessary to plan how to set the receiver depth at each point prior to the search. Conventional techniques include, for example, calculating the optimal depth using propagation loss calculated using a sound wave propagation model described in Non-Patent Document 1. In conventional techniques, by determining the optimal depth for the sound velocity distribution at multiple search points, a plan for searching for an object can be formulated. [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Richard P. Hodges, "6.6 Propagation Loss Model Descriptions," in Underwater Acoustics (2010) [Overview of the project] [Problems that the invention aims to solve]

[0005] However, with conventional technology, it is necessary to calculate propagation loss as many times as there are search points. When the target area is large, the number of search points inevitably becomes very large, but since calculating propagation loss using sound wave propagation models is generally computationally expensive, calculating propagation loss as many times as there are search points results in an enormous computational burden. As a result, problems such as the fact that a lot of time is required to plan the setting of receiver depth have arisen.

[0006] The present invention is based on the above-mentioned problems and aims to provide an information processing device, an information processing method, and an information processing program that can reduce the number of times propagation loss is calculated. [Means for solving the problem]

[0007] The information processing device according to the present invention comprises a group division unit that divides multiple search points within a target sea area into multiple groups containing similar sound velocity distributions and calculates a representative sound velocity distribution, which is a typical sound velocity distribution for each group, and a propagation loss calculation unit that calculates the propagation loss for each group based on the representative sound velocity distribution.

[0008] Furthermore, the information processing method according to the present invention comprises the steps of: dividing a plurality of search points within a target sea area into a plurality of groups containing similar sound velocity distributions; calculating a representative sound velocity distribution, which is a typical sound velocity distribution for each group; and calculating the propagation loss for each group based on the representative sound velocity distribution.

[0009] Furthermore, the information processing program according to the present invention causes a computer to perform the following steps: divide multiple search points within a target sea area into multiple groups containing similar sound velocity distributions; calculate a representative sound velocity distribution, which is a typical sound velocity distribution for each group; and calculate the propagation loss for each group based on the representative sound velocity distribution. [Effects of the Invention]

[0010] According to the information processing apparatus, information processing method, and information processing program of the present invention, the number of propagation loss calculations can be reduced by calculating the propagation loss based on representative sound velocity distributions for each group divided to include similar sound velocity distributions. [Brief explanation of the drawing]

[0011] [Figure 1] This diagram shows the types of sound channels. [Figure 2] This figure shows the upper depth of the sound channel. [Figure 3] This diagram shows the lower end depth of the sound channel. [Figure 4] This figure shows examples of sound channels included in the sound velocity distribution. [Figure 5] This is a schematic diagram of the information processing device according to Embodiment 1. [Figure 6] This figure shows an example of a group of search locations for the same sound channel sequence within the target sea area. [Figure 7] This is a flowchart showing the operation flow of the information processing device according to Embodiment 1. [Figure 8] This is a flowchart showing the flow of sound channel sequence calculation in Embodiment 1. [Figure 9] This diagram shows the relationship between the upper limit candidate depth, lower limit candidate depth, upper limit depth, and lower limit depth. [Figure 10] This diagram illustrates the direction in which to search for the maximum point of sound speed. [Figure 11] This figure shows an example of finding a sound channel when the upper limit candidate depth z(a) before the update is the sea surface. [Figure 12] This figure shows an example of finding a sound channel when the lower limit candidate depth z(b) before the update is the seabed. [Figure 13] This is a schematic diagram of the information processing device according to Embodiment 2. [Figure 14] This is a flowchart showing the operation flow of the information processing device according to Embodiment 2. [Figure 15]This is a schematic diagram of an information processing device relating to the conventional technology. [Figure 16] This is a flowchart showing the operation flow of an information processing device related to conventional technology. [Modes for carrying out the invention]

[0012] First, we will explain how to calculate propagation loss for multiple search points, using the example of applying the calculation of sound wave propagation loss from the target object to the receiver to the calculation of the receiver's optimal depth. Receiver depth z (s) The maximum distance at which an object can be detected (hereinafter referred to as the "detection distance") can be calculated using the inequality in equation (1).

number

[0013] Here TL(r,z (s) ) is the horizontal distance r from the object to the receiver, and the receiver depth z (s) The propagation loss is given by SL, the source level of the sound waves radiated by the object, NL, the noise level, DI, the directivity gain of the receiver, PG, the processing gain, and DT, the signal-to-noise ratio required for detection (hereinafter referred to as the "detection threshold"). The detection distance can be defined as the maximum value of the horizontal distance r for which the inequality in equation (1) holds. In general, the propagation loss is given by the horizontal distance r and the receiver depth z. (s) In addition to the function that takes the depth of the object (hereinafter referred to as "sound source depth") and frequency as variables, here we will use the horizontal distance r and receiver depth z, which are necessary for explaining the present invention. (s) It is written as a function with only these two variables, and all other variables are omitted.

[0014] In the following explanation, the sound source depth and frequency are assumed to be values ​​based on the characteristics of the object. The left side of equation (1) represents the signal-to-noise ratio of the processor output. Therefore, equation (1) indicates that if the signal-to-noise ratio exceeds the detection threshold DT, the sound waves emitted by the object can be detected. In the following explanation, the sound source level SL, noise level NL, directional gain DI, processing gain PG, and detection threshold DT are assumed to be values ​​based on the object.

[0015] When searching for an object while moving a receiver, the sound velocity distribution differs for each search point in the target sea area, so the optimum depth also differs for each point. Therefore, in order to efficiently carry out a search, it is necessary to plan how to set the receiver depth at each point prior to the search. Conventional techniques for calculating the optimum depth for each point within the target sea area are described below.

[0016] For a point number m (m=1,2,…,M) within the target sea area, the sound velocity distribution is expressed by the depth z n (c) and the sound velocity c mn combination (z n (c) ,c mn ). It is assumed here that n (n=1,2,…,N) is an index of depth for the sound velocity distribution, and z1 (c) <z2 (c) <…<z N (c) holds. z n (c) is the n-th depth of the sound velocity distribution, and does not depend on the point m (m=1,2,…,M). Further, c mn represents the sound velocity at the n-th depth at point m. Furthermore, it is assumed that the points m (m=1,2,…,M) are densely set within the target sea area to such an extent that the difference in sound velocity distribution between adjacent points can be considered sufficiently small.

[0017] FIG. 15 is a schematic configuration diagram of an information processing apparatus 500 according to a conventional technique. The information processing apparatus 500 includes a propagation loss calculation unit 501, a detection distance calculation unit 502, and an optimum depth calculation unit 503. The propagation loss calculation unit 501 operates once for each point m (m=1,2,…,M), and uses the sound velocity distribution (z n (c) ,c mn ) (n=1,2,…,N), receiver depth z k (s) (k=1,2,…,K), horizontal distance from the sound source to the receiver (hereinafter referred to as "horizontal distance") r l (l=1,2,…,L) to calculate the propagation loss TL m (r l ,zk (s) ) is calculated. Here, the horizontal distance r l (l=1,2,…,L) is the distance with point m as 0. Below, for the sake of simplification of notation, the propagation loss TL is used. m (r l ,z k (s) ) to TL mkl This is denoted as follows: Here, k (k=1,2,…,K) is the index of the receiver depth, and there are K receiver depths z k (s) (k=1,2,…,K) are candidate depths from which the optimal depth can be selected.

[0018] The detection distance calculation unit 502 operates once for each point m (m=1,2,...,M), and calculates the horizontal distance r l (l=1,2,…,L), propagation loss TL mkl (k=1,2,…,K, l=1,2,…,L), source level SL, noise level NL, directional gain DI, processing gain PG, and detected threshold DT are used to determine each receiver depth z k (s) Detection distance R for (k=1,2,…,K) mk The optimal depth calculation unit 503 operates once for each point m (m=1,2,…,M) and calculates the detected distance R. mk The optimal depth z is found from (k=1,2,…,K) m (o) Calculate.

[0019] Figure 16 is a flowchart showing the operation flow of the information processing device 500 according to the conventional technology. The information processing device 500 performs the following processing for each point m=1,2,...,M within the search point loop. First, the propagation loss calculation unit 501 calculates the propagation loss TL mkl (k=1,2,…,K, l=1,2,…,L) are calculated (S501). Propagation loss can be calculated using various sound wave propagation models. For example, sound wave propagation models using the ray method, normal mode method, and PE method can be used. Here, the sound velocity distribution used to calculate propagation loss is the sound velocity distribution with respect to point m (z n (c) ,cmn ) and assume that the sound velocity distribution is constant around point m.

[0020] Next, the detection distance calculation unit 502 calculates the detection distance R mk (k=1,2,…,K) are calculated (S502). Specifically, for each receiver depth index k (k=1,2,…,K), the maximum distance index l that satisfies the inequality in equation (1) is calculated, and R mk =r l Let's assume that here the propagation loss TL(r,z) in equation (1) is used. (s) ) section mkl We evaluate the inequality in equation (1) by applying the formula. Also, although the value of the noise level NL in equation (1) may differ depending on the search location m, we will assume here that it is the same value at all search locations.

[0021] Finally, the optimal depth calculation unit 503 calculates the optimal depth z m (o) The following is calculated (S503). The optimal depth calculation unit 503 calculates the detection distance R mk For the k that maximizes (k=1,2,…,K), z m (o) =z k (s) This is calculated as the optimal depth at search point m. The above is performed in a loop of search points, and conventional techniques use the sound velocity distribution of each search point m=1,2,...,M to calculate the optimal depth z for each. m (o) This allows for the determination of the target object, which in turn enables the development of a search plan.

[0022] Embodiment 1. Next, Embodiment 1 of the present invention will be described. First, the concept in Embodiment 1 will be explained. Since the propagation loss that forms the basis for calculating the optimal depth depends on the sound velocity distribution, the propagation losses for similar sound velocity distributions are similar, and the optimal depth calculated based on the propagation loss will also be close in value. Therefore, in Embodiment 1, the sound velocity distribution at each search point m=1,2,...,M (z n (c) ,c mnThe number of propagation loss calculations using the sound wave propagation model is reduced by dividing the data (n=1,2,…,N) into multiple groups containing similar sound velocity distributions and performing propagation loss calculations using the sound wave propagation model only for the representative sound velocity distribution of each group. Here, for group numbers j=1,2,…,J, the j-th group is defined as the set of search points belonging to that group G j ={m ji i=1,2,…,I j This is expressed as}. Here, m ji is the number of the search location, I j This is the number of search locations belonging to the j-th group.

[0023] In Embodiment 1, whether or not the sound velocity distributions are similar is determined by the sound channel. A sound channel refers to the depth range in which a sound wave is confined when its propagation direction changes from upward to downward or downward to upward due to refraction or reflection (hereinafter referred to as "reversal").

[0024] Figure 1 shows the types of sound channels. As shown in Figure 1, there are four types of sound channels. Here, Figure 1 shows the propagation paths of sound waves confined in each sound channel (hereinafter referred to as "sound lines"), illustrating how sound waves propagate from an object at distance 0. (1) Sound channels that are trapped by refraction on both the surface and seabed sides. (2) Sound channels trapped by refraction on the sea surface and reflection from the seabed (3) Sound channels trapped by reflection from the sea surface and refraction from the seabed (4) Sound channels trapped by surface and seabed reflections

[0025] Figure 2 shows the upper end depth of a sound channel. As shown in Figure 2, there are three types of depths that can be the upper end of the depth range of a sound channel. The first is the depth where the speed of sound is maximum, as shown in Figure 2(1) (hereinafter referred to as the "maximum depth"), the second is the depth where the speed of sound is the same as the lower end, as shown in Figure 2(2), and the third is the sea surface, as shown in Figures 2(3) and 2(4). At the maximum depth and the depth where the speed of sound is the same as the lower end, sound waves are reversed by refraction, and at the sea surface, they can be reversed by refraction (Figure 2(3)) or by reflection (Figure 2(4)).

[0026] Figure 3 shows the lower end depth of a sound channel. As shown in Figure 3, there are three types of depths that can be the lower end of the depth range of a sound channel. The first is the depth where the speed of sound is maximum (maximum depth), shown in Figure 3(1); the second is the depth where the speed of sound is the same as the speed of sound at the upper end, shown in Figure 3(2); and the third is the seabed, shown in Figures 3(3) and 3(4). At the maximum depth and the depth where the speed of sound is the same as the speed of sound at the upper end, sound waves are reversed by refraction, and at the seabed, they can be reversed by refraction (Figure 3(3)) or by reflection (Figure 3(4)).

[0027] A single sound velocity distribution at a given search site contains one or more sound channels. Figure 4 shows an example of sound channels included in a sound velocity distribution. As shown in the left column of Figure 4, a typical temperature distribution in deep-sea areas with depths of several thousand meters or more contains three types of sound channels from the four types mentioned above, in order from those with the narrowest depth range: (1), (3), and (4).

[0028] The right column of Figure 4, like Figure 1, shows sound rays propagating from an object at distance 0. Here, the upper end of sound channel (1) is the maximum depth, and the lower end is the depth at which the sound speed is the same as at the upper end. At both the upper and lower ends, the direction of sound wave propagation is reversed due to refraction. The upper end of sound channel (3) is the sea surface and the lower end is the seabed. At the upper end, the direction of sound wave propagation is reversed due to reflection, and at the lower end, it is reversed due to refraction. Similarly, the upper end of sound channel (4) is the sea surface and the lower end is the seabed, but the way it reverses at the lower end is different; in sound channel (4), the direction of sound wave propagation is also reversed at the lower end due to reflection.

[0029] The sound rays for each of the three sound channels shown in Figure 1 have different elevation angles from which sound waves are emitted. As the depth of the sound channels increases in the order of (1), (3), and (4), the elevation angle from which the sound waves confined to those sound channels are emitted increases. Here, for example, the type and order of sound channels determined from the sound velocity distribution shown in the left column of Figure 4 are denoted as [1,3,4], and this sequence will be referred to as the "sound channel sequence" from now on.

[0030] Since sound channels are characterized by the path of sound waves, if the sequence of sound channels determined by the sound velocity distribution is the same, the paths of the sound waves will be similar, and the propagation losses will also be considered similar. Therefore, in Embodiment 1, if the sequence of sound channels determined by the sound velocity distribution is the same, it is determined that the sound velocity distribution is similar, and multiple search points with similar sound velocity distributions are grouped together. Then, assuming that similar propagation losses can be calculated from sound velocity distributions that are similar, that is, have the same sound channels, the propagation loss is calculated for a representative sound velocity distribution of each group (hereinafter referred to as the "representative sound velocity distribution"), and the optimal depth is calculated based on the results.

[0031] FIG. 5 is a schematic configuration diagram of an information processing apparatus 100 according to the first embodiment. The information processing apparatus 100 is an apparatus for calculating an optimum depth when detecting the presence of an underwater object in a sonar or the like by receiving sound waves radiated from the object. The information processing apparatus 100 includes a computer having a processor such as a CPU and a memory, a dedicated processing circuit such as an ASIC or an FPGA, or both. As shown in FIG. 5, the information processing apparatus 100 includes, as functional units, a propagation loss calculation unit 1, a detection distance calculation unit 2, an optimum depth calculation unit 3, a sound channel sequence calculation unit 4, a group division unit 5, and an optimum depth setting unit 6. Each functional unit of the information processing apparatus 100 is realized by reading and executing an information processing program stored in a memory, or is realized by a processing circuit corresponding to each functional unit.

[0032] The sound channel sequence calculation unit 4 operates once for each search point m (m=1,2,...,M), based on the sound velocity distribution z n (c) ,c mn ) (n=1,2,...,N) and the sound source depth z (t) to calculate the sound channel sequence S m =[s m1 ,s m2 ,...,s mHm . Here, s mh (m=1,2,...,M, h=1,2,...,H m ) is the type number of the H m sound channels included in the sound channel sequence of search point m (an integer satisfying 1≦s mh ≦4). For example, when the example of FIG. 4 corresponds to search point m, S m =[s m1 ,s m2 ,s m3 =[1,3,4], and H m =3.

[0033] The group division unit 5 classifies identical sound channel sequences S m for each search point m=1,2,...,M into the same group, and divides the search points m=1,2,...,M into groups G j ={m ji,i=1,2,…,I j}(j=1,2,…,J). Here, J is the number of groups created by division, m ji (i=1,2,…,I j ) is group G j (j=1,2,…,J) belonging to the I j th search point number (integer satisfying 1≦m ji ≦M).

[0034] Figure 6 is a diagram showing an example of groups of search points of the same sound channel sequence in a target sea area. In Figure 6, each search point indicated by a black circle is numbered with m=1,2,…,M. Group G j consists of I j search points m j1 , m j2 ,…,m jIj . And for these search points, the sound channel sequence S mj1 , S mj2 ,…,S mjIj are all the same. In addition, the group dividing unit 5 calculates the representative sound speed distribution (z j (j=1,2,…,J) n (c) , c jn (G) ).

[0035] The propagation loss calculation unit 1 is functionally the same as the propagation loss calculation unit 501 of the prior art, but the input sound speed distribution is not the sound speed distribution (z n (c) , c mn ) (n=1,2,…,N) at each search point m=1,2,…,M, but the representative sound speed distribution (z j for each group G (j=1,2,…,J) n (c) , c jn (G) ) (n=1,2,…,N). The propagation loss calculation unit 1 operates once for each group G j (j=1,2,…,J), and based on the representative sound speed distribution (z n (c) , c jn (G))(n=1,2,…,N), horizontal distance r l (l=1,2,…,L) and receiver depth z k (s) Propagation loss TL for (k=1,2,…K) j (G) (r l ,z k (s) Calculate the propagation loss TL (k=1,2,…,K, l=1,2,…,L). For the sake of notation, see below. j (G) (r l ,z k (s) ) to TL jkl (G) This is how it is written.

[0036] The detection distance calculation unit 2 is functionally the same as the conventional detection distance calculation unit 502, but the input propagation loss is the propagation loss TL at each search point m=1,2,...,M mkl (k=1,2,…,K, l=1,2,…,L) rather than each group G j Propagation loss TL for (j=1,2,…,J) jkl (G) (k=1,2,…,K, l=1,2,…,L). The detection distance calculation unit 2 is in group G. j The operation is performed once for each of (j=1,2,…,J), and the propagation loss TL jkl (G) (k=1,2,…,K, l=1,2,…,L), source level SL, noise level NL, directional gain DI, processing gain PG, and detected threshold DT are used to determine each receiver depth z k (s) Detection distance R for (k=1,2,…,K) jk (G) Calculate (k=1,2,…,K).

[0037] The optimal depth calculation unit 3 is functionally the same as the conventional optimal depth calculation unit 503, but the input detection distance is the detection distance R at each search point m=1,2,...,M. mk Rather, each group G j Detection distance R for (j=1,2,…,J) jk (G)(k=1,2,…,K). The optimal depth calculation unit 3 is in group G. j The operation is performed once for each of (j=1,2,…,J), and the detection distance R jk (G) For the k that maximizes (k=1,2,...K), z j (G) =z k (s) Group G j The optimal depth is calculated for (j=1,2,…,J).

[0038] The optimal depth setting unit 6 is for each group G j Optimal depth z for (j=1,2,…,J) j (G) Based on this, the optimal depth z for each search point m=1,2,...,M m (o) Set it.

[0039] Figure 7 is a flowchart showing the operation flow of the information processing device 100 according to Embodiment 1. As shown in Figure 7, first, the sound channel sequence calculation unit 4 calculates the sound velocity distribution (z) for each of the search points m=1,2,...,M within the loop of search points. n (c) ,c mn Sound channel sequence S for (n=1,2,…,N) m =[s m1 ,s m2 ,…,s mHm ] is calculated (S1). Here H m The sound channel sequence S at search point m. m This is the number of elements.

[0040] Figure 8 is a flowchart showing the flow of sound channel sequence calculation in Embodiment 1. The calculation of the sound channel sequence in step S1 of Figure 7 will be explained with reference to Figure 8. As shown in Figure 8, first, the sound velocity distribution (z n (c) ,c mn From (n=1,2,…,N), the sound source depth z (t) The shallowest and deepest depth z n (c)The depth index of z n^ (c) ≤z (t) <z n^+1 (c) Search for the depth index n^ that gives (S101). Note that the symbol "^" written next to n is a hat indicating an estimated value, and is written above n as shown in step S101 in Figure 8. For the searched depth index n^, z n^ (c) The upper side and z n^+1 (c) By determining the upper and lower limits of the sound channel on the lower side, it is possible to obtain a sound channel whose depth range includes the sound source depth.

[0041] Next, the upper limit candidates are set (S102). Here, starting from the depth index n^, c is set in order. mn^ ,c m,n^-1 ...and we look at the value of the speed of sound in the direction of decreasing depth, and first the speed of sound is at its maximum, that is, c m,n~-1 <c mn~ and c mn~ >c mn~+1 For n such that z (a) =z n~ (c) , c (a) =c mn~ , let's assume. Also, flg (a) Let z = 0. (a) c is the depth of the candidate that will be the upper end depth of the sound channel (hereinafter referred to as "upper end candidate depth"), c (a) This is the speed of sound at the upper end candidate depth (hereinafter referred to as "upper end candidate speed of sound"), flg (a) The upper limit candidate depth z (a) This is a flag (hereinafter referred to as the "upper end reflection flag") that indicates whether the path of the sound wave that turns is refraction or reflection, and is set to 0 for refraction and 1 for reflection. The symbol "~" written next to n is a tilde that indicates the index of the depth at which the speed of sound is maximum, and is written above n.

[0042] Figure 9 shows the relationship between the upper limit candidate depth, lower limit candidate depth, upper limit depth, and lower limit depth. As shown in Figure 9, the upper limit candidate sound velocity c (a)The lower limit candidate sound speed c will be explained below. (b) Depending on the relative magnitudes, the upper limit candidate depth z (a) The upper limit depth may or may not be. Upper limit candidate sound velocity c (a) is the lower end candidate sound speed c (b) In the following case, as shown in Figure 9(A), the upper end candidate depth z (a) This represents the upper end depth of the sound channel. Upper end candidate sound velocity c (a) is the lower end candidate sound speed c (b) If it is greater than the upper end depth of the sound channel is the upper end candidate depth z, as shown in Figure 9(B). (a) It becomes deeper. In either case, the upper end depth of the sound channel is the upper end candidate depth z (a) That concludes the explanation.

[0043] From here on, the upper limit candidate depth z (a) , upper candidate sound speed c (a) , and upper end reflection flag flg (a) These three are collectively called "upper limit candidates." Note that if there is no depth at which the speed of sound is maximum before reaching the sea surface, z (a) =0, c (a) =c m1 Let's assume that the speed of sound has a non-negative gradient, i.e., c m1 ≤c m2 In this case, sound waves are refracted upwards and turn at the sea surface always results in reflection, therefore flg (a) Let = 1, and the negative gradient, i.e., c m1 >c m2 In this case, sound waves refract downwards and turn at the sea surface due to refraction, therefore flg (a) Set = 0. The above is the procedure for setting the upper limit candidate in step S102.

[0044] Similarly, the lower limit candidates are set (S103). Here, starting from n^+1, c mn^+1 ,c mn^+2 ...and so on, we look at the value of the speed of sound in the direction of increasing depth, and the depth at which the speed of sound first reaches its maximum is the lower limit candidate depth z (b) , the lower end candidate sound speed is c (b) Set the lower end reflection flag to flg (b) Let = 0.

[0045] As shown in Figure 9, candidate lower end sound velocity c (b) and upper limit candidate sound speed c (a) Depending on the relative magnitudes, the lower end candidate depth z (b) The lower end depth may or may not be. Upper end candidate sound velocity c (a) is the lower end candidate sound speed c (b) In the above case, as shown in Figure 9(B), the lower end candidate depth z (b) This represents the lower end depth of the sound channel. The upper end candidate is the sound velocity c. (a) is the lower end candidate sound speed c (b) If it is smaller than the lower end depth of the sound channel is lower end candidate depth z, as shown in Figure 9(A). (b) It becomes shallower. In any case, the lower end depth of the sound channel is the lower end candidate depth z (b) The results are as follows:

[0046] From here on, the lower limit candidate depth z (b) , lower end candidate sound speed c (b) , and the lower end reflection flag flg (b) These three are collectively called "lower end candidates." Note that if there is no depth where the speed of sound is maximum before reaching the seabed, z (b) =z N (c) , c (b) =c mN And the speed of sound has a non-positive gradient, that is, c mN-1 ≤c mN In this case, sound waves are refracted downwards and turn at the seabed always results in reflection, therefore flg (b) Let = 1, and the positive gradient, i.e., c m,N-1 <c mN In this case, sound waves refract upwards and turn around at the seabed due to refraction, therefore flg (b) Set = 0. The above is the procedure for setting the lower end candidate in step S103.

[0047] Up to this point, the upper limit candidate depth z (a) , upper candidate sound speed c (a) , upper end reflection flag flg (a) , lower end candidate depth z (b) , lower end candidate sound speed c (b) , and the lower end reflection flag flg (b)This was set. Next, the index h of the sound channel is initialized to 1 (S104). Then, the upper end reflection flag flg, which was set in steps S102 and S103 respectively, is set. (a) and lower end reflection flag flg (b) Based on this, the sound channel sequence S for the search point m m The h (=1st) element s mh This is determined using equation (2) (S105).

number

[0048] Here, "==" indicates that the values ​​on both sides are equal. At this point, the upper end reflection flag flg (a) and lower end reflection flag flg (b) If both are 1, meaning the sound channels are reflected at the sea surface and the seabed (S106: YES), the depth range will not extend any further and no elements will be added to the sound channel sequence, so the sound channel sequence calculation is terminated. On the other hand, the upper reflection flag flg (a) and lower end reflection flag flg (b) If either or both of the following are not 1 (S106:NO), proceed to the next step.

[0049] Upper end candidate sound speed c (a) lower limit candidate sound speed c (b) Compare (S107), upper limit candidate sound speed c (a) The lower limit candidate sound speed c (b) If it is smaller than (S107:YES), the upper limit candidate depth z (a) If z is not 0 (S108: YES), the upper limit candidate is updated (S109). Here, the upper limit candidate depth z (a) , upper candidate sound speed c (a) , and upper end reflection flag flg (a) The update will be performed using the following procedure.

[0050] First, the current upper limit candidate depth z (a)We then look at the speed of sound values ​​as we move towards shallower depths. Figure 10 illustrates the direction in which we search for the maximum point of sound speed. Upper limit candidate sound speed c (a) If the upper limit is smaller, the upper limit candidate depth z (a) The reason for looking at the sound velocity values ​​towards shallower depths is explained using Figure 10(A). In Figure 10(A), the solid arrow represents the sound channel with the widest depth range found so far, and the dashed arrow represents the next sound channel to be found. The next sound channel to be found will have an upper end depth that is the maximum depth of sound, as shown in Figure 10(A1), or an upper end depth that is the lower end candidate sound velocity c, as shown in Figure 10(A2). (b) Either the depth will be the same as the speed of sound, or the upper end depth will be the sea surface, as shown in Figure 10(A3).

[0051] In any case, the next lower end depth of the sound channel found will be the current candidate lower end depth z. (b) It will not go any deeper than this, and the upper limit depth will be the current upper limit candidate depth z (a) It becomes shallower than this. This is the upper limit candidate for sound velocity c. (a) If the upper limit is smaller, the upper limit candidate depth z (a) This is why we look at the speed of sound values ​​as we move towards shallower depths. The first point is where the speed of sound is maximum, and that maximum value is the upper limit candidate for the speed of sound c. (a) Search for a depth greater than this, and select a candidate upper limit depth z (a) The maximum depth is updated, and the upper limit candidate sound velocity c (a) Updated upper limit candidate depth z (a) The upper limit candidate depth z is updated at the speed of sound. (a) Figures 10(A1) and (A2) show how the data is updated.

[0052] However, if such a depth is not found until the sea surface is reached, the upper limit candidate depth z (a) The upper limit candidate sound speed c is updated at sea surface depth (0). (a) This is updated at the speed of sound at sea level. In this way, the upper limit candidate depth z (a) Figure 10(A3) shows how the update process works. Also, the upper limit candidate sound velocity c before and after the update. (a) When compared to the updated upper limit candidate sound velocity c,(a) If it is smaller, the upper reflection flag is also flg (a) Update to =1. This concludes the procedure for updating the upper limit candidate in step S109.

[0053] Upper limit candidate depth z before update (a) that is the sea surface, that is, z (a) The case where == 0 (S108:NO) will be explained using Figure 11. Figure 11 shows the upper limit candidate depth z before the update. (a) This figure shows an example of finding a sound channel when the depth is at sea level. In this case, the upper limit candidate depth is z. (a) It is not possible to look at the value of the speed of sound at shallower depths than this. At this time, flg (a) If == 0 (S110: YES), only the upper end reflection flag is set to flg (a) Update to =1 (S111).

[0054] This corresponds, for example, to the stage of finding the sound channel in Figure 11(B) after the sound channel in Figure 11(A). In Figure 11(A), the upper inversion is refraction, whereas in Figure 11(B), the upper inversion is reflection. The lower end depth of the sound channel is the lower end candidate depth z in Figure 11(A). (b) It is located at a shallower depth than that, and the speed of sound at that depth is equal to the speed of sound at the upper end depth of the sound channel. On the other hand, in Figure 11(B), the lower end depth of the sound channel is the lower end candidate depth z (b) However, the lower limit candidate depth z (b) Since the flag itself is not updated, the upper end reflection flag flg (a) This will only update the value from 0 to 1.

[0055] Also, upper end candidate depth z (a) that is the sea surface, that is, z (a) == 0 (S108: NO), and also flg (a) ==0 (S110:NO), flg (a) If ==1, the lower limit candidate is updated (S112). Here, the lower limit candidate depth z (b) , lower end candidate sound speed c (b) and lower end reflection flag flg (b)is updated.

[0056] This corresponds to, for example, the step of finding the sound channel of FIG. 11(C) next to the sound channel of FIG. 11(B). The lower end candidate depth z (b) is updated to a deeper position as indicated by the thick arrow in FIG. 11(C). However, in this update, when z (b) == z N (c) and flg (b) == 0, only flg (b) is updated to 1.

[0057] The above is the procedure when the upper end candidate sound speed c (a) is smaller than the lower end candidate sound speed c (b) (S107: YES). When the processing of steps S108 to S112 is completed in this procedure, the sound channel index h is updated by incrementing it by 1 (S118), and the h-th element s m of the sound channel sequence S mh is determined using the right-hand side of equation (2) (S119).

[0058] Next, the case where the lower end candidate sound speed c (b) is equal to or less than the upper end candidate sound speed c (a) (S107: NO) will be described. In the case where the lower end candidate sound speed c (b) is equal to or less than the upper end candidate sound speed c (a) (S107: NO), and the lower end candidate depth z (b) is not the seabed depth z N (c) (S113: YES), the lower end candidate is updated (S114). Here, the lower end candidate depth z (b) , the lower end candidate sound speed c (b) , and the lower end reflection flag flg (b) are updated by the following procedure.

[0059] First, the sound speed values are checked from depth z (b) further toward deeper depths. When the lower end candidate sound speed c (b) is equal to or less than the upper end candidate sound speed c (a) , the lower end candidate depth z(b) The reason why the sound speed value is further checked toward greater depths will be described with reference to FIG. 10(B). In FIG. 10(B), the solid-line arrow indicates the sound channel with the widest depth range that has been found so far, and the dashed-line arrow indicates the sound channel to be found next.

[0060] For the sound channel to be found next, either the lower-end depth becomes the maximum depth of sound speed as shown in FIG. 10(B1), or the lower-end depth has the same sound speed as the candidate upper-end sound speed c (a) or the lower-end depth reaches the seabed as shown in FIG. 10(B3). In any case, the upper-end depth of the sound channel to be found next is never shallower than the current candidate upper-end depth z (a) , and the lower-end depth is deeper than the current candidate lower-end depth z (b) . This is the reason why the sound speed value is further checked toward greater depths from the candidate lower-end depth z (b) when the candidate lower-end sound speed c (b) is smaller.

[0061] Then, a depth at which the sound speed is first at a maximum and the maximum value is larger than c (b) is searched for, the candidate lower-end depth z (b) is updated with the depth at which the maximum sound speed occurs, and the candidate lower-end sound speed c (b) is updated with the updated candidate lower-end sound speed. How the candidate lower-end depth z (b) is updated in this manner is shown in FIG. 10(B1) and (B2). However, when such a depth is not found before reaching the seabed, the candidate lower-end depth z (b) is updated to the seabed depth z N (c) , and the candidate lower-end sound speed c (b) is updated with the sound speed of the seabed. How the candidate lower-end depth z (b) is updated in this manner is shown in FIG. 10(B3). Further, when comparing the candidate lower-end sound speed c (b) before and after updating, if the updated candidate lower-end sound speed c (b) is smaller, the lower-end reflection flag is also updated to flg (b) =1. The above is the procedure for updating the lower-end candidate in step S114.

[0062] Next, the lower limit candidate depth z before the update (b) that is the seabed, that is, z (b) ==z N (c) The case (S113:NO) will be explained using Figure 12. Figure 12 shows the candidate lower end depth z before the update. (b) This figure shows an example of finding a sound channel when the bottom is the seabed. In this case, the lower end candidate depth z (b) It is not possible to look at the value of the speed of sound at deeper depths. At this time, flg (b) If == 0 (S115: YES), only the lower end reflection flag is set to flg (b) Update to =1 (S116).

[0063] This corresponds, for example, to the stage of finding the sound channel in Figure 12(B) after the sound channel in Figure 12(A). In Figure 12(A), the lower inversion is refraction, whereas in Figure 12(B), the lower inversion is reflection. The upper end depth of the sound channel is the upper end candidate depth z in Figure 12(A). (a) It is located at a greater depth than that, and the speed of sound at that depth is equal to the speed of sound at the lower end depth of the sound channel. On the other hand, in Figure 12(B), the upper end depth of the sound channel is the upper end candidate depth z (a) However, the upper limit candidate depth z (a) Since the flag itself is not updated, the lower end reflection flag flg (b) This will only update the value from 0 to 1.

[0064] Also, lower end candidate depth z (b) that is the seabed, that is, z (b) ==z N (c) (S113:NO) and flg (b) ==0 (S115:NO), flg (b) If ==1, the upper limit candidate is updated (S117). Here, the upper limit candidate depth z (a) , upper candidate sound speed c (a) and upper-end reflection flag flg (a) An update will be performed.

[0065] This corresponds, for example, to the stage of finding the sound channel in Figure 12(C) after the sound channel in Figure 12(B). Upper limit candidate depth z in Figure 12(B) (a) However, as shown by the thick arrow in Figure 12(C), it is updated to the shallower side. However, in this update z (a) ==0 and flg (a) When == 0, flg (a) Only update the value to 1.

[0066] The above are candidates for the lower limit of sound velocity c. (b) Upper limit candidate sound speed c (a) The following procedure applies to the case (S107:NO). In this procedure, after the processing in steps S113 to S117 is completed, the sound channel index h is updated by incrementing it by 1 (S118), and the sound channel sequence S is obtained using the right-hand side of equation (2). m The h-th element s mh To decide (S119).

[0067] Upper edge reflection flag flg (a) and lower end reflection flag flg (b) The process in steps S107 to S119 is repeated until both values ​​become 1 (S120: NO), thereby generating the sound channel sequence S for the search point m. m elements s m1 ,s m2 The following are determined in order: upper end reflection flag flg (a) and lower end reflection flag flg (b) When both become 1 (S120:YES), the sound channel calculation is terminated. This is because, for sound channels reflected from the sea surface and seabed, the depth range will not extend any further, and no elements will be added to the sound channel sequence. Therefore, the sound channel sequence S m The last element s mHm According to the right-hand side of equation (2), s must always be s mHm = 4. Here, the sound channel sequence S m Number of elements H m This is the index h of the sound channel at this point.

[0068] Note that in Figure 8, the upper limit candidate sound velocity c (a) lower limit candidate sound speed c (b) If they are the same (c (a) =c (b) ) is to perform the processing in steps S113 to S117, but the upper limit candidate sound velocity c (a) lower limit candidate sound speed c (b) If they are the same (c (a) =c (b) ) The processing in steps S108 to S112 may be performed. That is, the upper limit candidate sound velocity c (a) is the lower end candidate sound speed c (b) In the following case (c (a) ≤c (b) ) The process in steps S108 to S112 is performed, and the lower end candidate sound velocity c (b) Upper limit candidate sound speed c (a) If it is smaller than (c (a) >c (b) ) The processing in steps S113 to S117 may be performed.

[0069] Returning to Figure 7, the sound channel sequence S for each search point m=1,2,...,M m Once the calculation is complete, the group division unit 5 divides the sound channel sequence S m Based on this, search points m=1,2,…,M belong to group G j ={m ji i=1,2,…,I j It is divided into groups (j=1,2,…,J) (S2). Here, J is the number of groups to which it is divided, I j is the number of search locations belonging to group j. The procedure for dividing into groups is as follows:

[0070] First, S1 = S m m that satisfies m 1i Let (i=1,2,…,I1) be the first group G1={m 1i Create {i=1,2,…,I1}. Here the first element is m 11 Let = 1. Next, find the smallest value of m that does not belong to G1. min (1) In contrast, S mmin(1) =S m m that satisfies m 2iLet (i=1,2,…,I2) be the second group G2={m 2i Create {i=1,2,…,I2}. Here the first element is m 21 =m min (1) Let's assume that the minimum value of m does not belong to G1 or G2 is m. min (2) S mmin(2) =S m m that satisfies m 3i Let (i=1,2,…,I3) be the third group G3={m 3i Create {i=1,2,…,I3}. Here the first element is m 31 =m min (3) Repeat the above steps until all search locations belong to one of the Group G groups. j It belongs to (j=1,2,…,J), that is, {1,2,…,M}=G1∪G2∪…∪G J The loop stops when this happens. mmin(1)、 S mmin(2) Subscript mmin(1)、mmin(2) These are m min (1) and m min (2) Let it be so.

[0071] Here, two sound channel sequences S m and S μ (m,μ=1,2,…,M) are equal to (S m =S μ ) means that each contains an equal number of sound channels (H m =H μ ), and the types of sound channels included in each are the same, including their order (s m1 =s μ1 , s m2 =s μ2 ,…,s mHm =s μHμ )

[0072] The group division section 5 finally divides each group G j (j=1,2,…,J) representative sound velocity distribution (z n (c),c jn (G) Calculate the number of search points m belonging to that group, as shown in equation (3) below. ji (i=1,2,…,I j The average value of the sound velocity distribution for ) can be used as the representative sound velocity distribution. In addition, the group division section 5 is group G j The representative sound velocity distribution may be calculated using any method that can be considered representative of the sound velocity distribution.

number

[0073] Once the group division in step S2 is complete, propagation loss calculation (S3), detection distance calculation (S4), and optimal depth calculation (S5) are then performed for j=1,2,...,J in the group loop. In step S3, the propagation loss calculation unit 1 calculates the group G j Representative sound velocity distribution (z n (c) ,c jn (G) )(n=1,2,…,N), receiver depth z k (s) (k=1,2,…,K) and horizontal distance r l The propagation loss is calculated for (l=1,2,…,L). The propagation loss calculation unit 1 uses the sound wave propagation model to calculate the group G j Propagation loss TL for jkl (G) Calculate (k=1,2,…,K, l=1,2,…,L).

[0074] Next, in step S4, the detection distance calculation unit 2 calculates group G j Propagation loss TL for jkl (G) (k=1,2,…,K, l=1,2,…,L), horizontal distance r l The detection distance is calculated for (l=1,2,…,L), sound source level SL, noise level NL, directional gain DI, processing gain PG, and detection threshold DT. The detection distance calculation unit 2 calculates the detection distance R for the largest distance index l that satisfies the inequality in equation (1).jk (G) =r l is given. Here, the propagation loss TL(r,z of formula (1) (s) ), TL is applied to the term of jkl (G) to evaluate the inequality of formula (1).

[0075] Then, in step S5, the optimum depth calculation unit 3 calculates for group G j and receiver depth z k (s) (k=1,2,…,K), detection distance R jk (G) and the optimum depth is calculated for receiver depth z k (s) (k=1,2,…,K). The optimum depth calculation unit 3 calculates for group G j and detection distance R jk (G) (k=1,2,…,K), z is obtained for k that maximizes j (G) =z k (s) is calculated as the optimum depth.

[0076] The above processing is executed in the group loop, and for each group G j (j=1,2,…,J), the group-specific optimum depth z j (G) is calculated. Then, in order to set these results as the optimum depth z for each search point m=1,2,…,M m (o) , optimum depth setting is performed by the optimum depth setting unit 6 (S6). The optimum depth setting unit 6 applies formula (4) for j=1,2,…,J to set the optimum depth z for each search point m=1,2,…,M m (o)

Mathematical Expression

[0077] ​As described above, according to Embodiment 1, the number of times propagation loss is calculated by the propagation loss calculation unit 1 can be reduced from the number of search points M, which is the number of times propagation loss is calculated by the propagation loss calculation unit 501 in the conventional technology, to the number of groups J. This reduces the computational load on the propagation loss calculation unit 1 and shortens the time required for planning the setting of the receiver depth.

[0078] Embodiment 2. In Embodiment 1, groups were created assuming that search points with the same sound channel sequence have similar sound velocity distributions and that similar propagation losses can be calculated. In Embodiment 2, among the search points with the same sound channel sequence belonging to a group, those that are close to each other are considered to have the same sound velocity distribution. The results of this grouping are further divided, and the representative sound velocity distribution and the propagation loss for the representative sound velocity distribution are calculated for each of the subdivided groups in the same manner as in Embodiment 1, thereby improving the accuracy of the propagation loss.

[0079] Figure 13 is a schematic diagram of the information processing device 100A according to Embodiment 2. The information processing device 100A is a device for calculating the optimal depth, similar to Embodiment 1, and includes a computer equipped with a processor such as a CPU and memory, a dedicated processing circuit such as an ASIC or FPGA, or both. As shown in Figure 13, the information processing device 100A includes, as functional units, a propagation loss calculation unit 21, a detection distance calculation unit 22, an optimal depth calculation unit 23, a sound channel sequence calculation unit 4, a group division unit 5, an optimal depth setting unit 26, and a group re-division unit 7. Each functional unit of the information processing device 100A is realized by reading and executing an information processing program stored in memory, or by a processing circuit corresponding to each functional unit.

[0080] The operation of the sound channel sequence calculation unit 4 and the group division unit 5 in this embodiment is the same as in Embodiment 1. The propagation loss calculation unit 21, detection distance calculation unit 22, optimal depth calculation unit 23, and optimal depth setting unit 26 operate in the same way as the propagation loss calculation unit 1, detection distance calculation unit 2, optimal depth calculation unit 3, and optimal depth setting unit 6 in Embodiment 1, respectively, but the input variables and output variables are different.

[0081] The group re-division unit 7 is the group G calculated in the group division unit 5 j Search points m belonging to each of (j=1,2,…,J) j1 ,m j2 ,…,m jIj Groups with short distances between search locations Γ ju (u=1,2,…,U j ) is further divided into. In Embodiment 2, group G divided in group division section 5 j (j=1,2,…,J) is called the "initial group," and the group Γ is a subgroup created by the group subdivision unit 7. ju (u=1,2,…,U j ) is called a "repartition group". Γ ju The subscript j indicates that the group was created by subdividing the j-th group, and the subscript u is the index number of the group created by subdividing the j-th group. j Γ is the number of subdivided groups created by subdividing the j-th group. Below, we refer to the subdivided group Γ ju The search points that are elements of μ jui (i=1,2,…,I ju (r) )

[0082] Here I ju (r) This is the subdivision group Γ ju (j=1,2,…,J, u=1,2,…,U jThis is the number of search locations belonging to ). The sound velocity distribution does not change abruptly with respect to changes in search locations, and can be considered almost identical if the distance between search locations is short. Therefore, by subdividing in this way, it is possible to calculate groups that include only search locations with closer sound velocity distributions. The group subdivision unit 7 calculates the subdivision groups and also calculates the subdivision group Γ ju (j=1,2,…,J, u=1,2,…,U j The sound velocity distribution (z) that represents the search location contained in each of the ) n (c) ,c jun (Γ) Calculate (n=1,2,…,N).

[0083] Figure 14 is a flowchart showing the operation flow of the information processing device 100A according to Embodiment 2. First, within the search point loop, the sound channel sequence calculation unit 4 calculates the sound channel sequence S for each of the search points m=1,2,...,M. m The following is calculated (S21), and the group division unit 5 divides the search points m=1,2,...,M into the initial group G j The signal is divided into (j=1,2,…,J) (S22). Up to this point, it is the same as in Embodiment 1.

[0084] Next, within the loop of the initial groups, each initial group G j For (j=1,2,…,J), the following is performed. First, the group repartitioning unit 7 creates an initial group G j The search locations are further divided into groups based on the proximity of the search locations, and the divided groups are Γ ju (u=1,2,…,U j ) is calculated (S23). A method called cluster analysis can be applied as a way to group search locations that are close to each other. Generally, cluster analysis divides a set of data into subsets called clusters, and makes the characteristics of the data included in each subset similar. Here, cluster analysis is applied by defining the similarity of the characteristics of two search locations as the distance between the search locations. Here, the initial group G j Two search locations m belonging to jp and mjq (p,q=1,2,…,I j The positions of ) are in the xy coordinate system, respectively (x jp ,x jp ) and (x jq ,x jq When expressed as ), the distance between these two search points is d(m jp ,m jq ) is defined by the following equation (5).

number

[0085] There are several methods for cluster analysis, but one example is the k-means method. The k-means method uses U as the number of subdivided groups. j It is necessary to decide this in advance, and as will be described later, it can be used as a variable to balance the trade-off between computational complexity and accuracy. Also, subpartition group U j If the value of cannot be determined in advance, hierarchical clustering can be used as a cluster analysis method.

[0086] In step S23, the group is subdivided into subgroup Γ ju Representative sound velocity distribution for (z n (c) ,c jun (Γ) )(u=1,2,…,U j We also calculate n=1,2,...,N). For example, it can be calculated using equation (6), which is the same as equation (3) for calculating the average speed of sound.

number

[0087] Also, the subdivision group Γ ju A representative sound velocity distribution may be calculated using any method that can be considered representative of the sound velocity distribution. Once the group subdivision is complete, in the next loop of subdivision groups, u=1,2,…,U j Propagation loss calculation (S24), detection distance calculation (S25), and optimal depth calculation (S26) are performed for this.

[0088] In step S24, the propagation loss calculation unit 21 re-divides the group Γ ju Representative sound velocity distribution (z n (c) ,c jun (Γ) ), receiver depth z k (s) (k=1,2,…,K) and horizontal distance r l The propagation loss is calculated for (l=1,2,…,L). The propagation loss calculation unit 21 uses the sound wave propagation model to calculate the subdivision group Γ ju Propagation loss TL for jukl (Γ) Calculate (k=1,2,…,K, l=1,2,…,L).

[0089] In step S25, the detection distance calculation unit 22 subdivides the group Γ ju Propagation loss TL for jukl (Γ) (k=1,2,…,K, l=1,2,…,L), horizontal distance r l The detection distance is calculated for (l=1,2,…,L), sound source level SL, noise level NL, directional gain DI, processing gain PG, and detection threshold DT. The detection distance calculation unit 22 calculates the detection distance R for the largest distance index l that satisfies the inequality in equation (1). juk (Γ) =r l Let's assume that here the propagation loss TL(r,z) in equation (1) is used. (s) ) section jukl (Γ) Apply this to evaluate the inequality in equation (1).

[0090] In step S26, the optimal depth calculation unit 23 subdivides the group Γ ju and receiver depth z k (s) Detection distance R for (k=1,2,…,K) juk (Γ) and receiver depth z k (s) The optimal depth is calculated for (k=1,2,…,K). The optimal depth calculation unit 23 is for the subdivision group Γju For the detection distance R juk (Γ) For the k that maximizes (k=1,2,…,K), z ju (Γ) =z k (s) This is calculated as the optimal depth.

[0091] The above is performed in the loops of the subdivision group and the initial group, and the subdivision group Γ ju (j=1,2,…,J, u=1,2,…,U j ) Optimal depth z for each subdivision group ju (Γ) The following is calculated. Then, the optimal depth setting unit 26 uses these results to determine the optimal depth z for each search point m=1,2,...,M. m (o) The optimal depth is set to achieve this. The optimal depth setting unit 26 sets j=1,2,...,J, u=1,2,...,U j Applying equation (7), similar to equation (4), to the search points m=1,2,...,M, we obtain the optimal depth z m (o) Set it.

number

[0092] As described above, according to Embodiment 2, by subdividing each of the groups divided using the sound channel sequence, the distance between search points belonging to the subdivided groups is restricted to being closer than in the original group, and the sound velocity distribution is limited to only those search points with a more similar sound velocity distribution. As a result, the difference between the sound velocity distribution of search points belonging to the subdivided groups and the representative sound velocity of the subdivided groups becomes smaller than the difference between the sound velocity distribution of search points belonging to the initial group and the representative sound velocity of the initial group. This also reduces the difference between the propagation loss for each search point and the propagation loss with respect to the representative sound velocity, thus improving accuracy compared to Embodiment 1.

[0093] Furthermore, regarding the number of times propagation loss is calculated, the following relationship exists: the number of executions of the propagation loss calculation unit 1 in Embodiment 1, the number of executions of the propagation loss calculation unit 21 in Embodiment 2, and the number of executions of the conventional propagation loss calculation unit 501, in that order, as shown in equation (8).

number

[0094] When using the k-means method as a cluster analysis technique, the number of subdivided groups U is predetermined. j As the value of (j=1,2,…,J) decreases, the accuracy decreases, and the number of propagation loss calculations decreases, approaching that of Embodiment 1. Also, the number of subdivision groups U j Increasing (j=1,2,…,J) improves accuracy and increases the number of propagation loss calculations. Thus, the number of subdivision groups U increases. j (j=1,2,…,J) can be used as a variable to trade off the accuracy of the results obtained from the computational complexity.

[0095] The above describes embodiments of the present invention, but the present invention is not limited to the configuration of the above embodiments, and various modifications or combinations are possible within the scope of its technical idea. For example, Embodiment 1 describes a method for calculating propagation loss for multiple search points, using the case of application to calculating the optimal depth of a receiver as an example. K receivers with depths z are used to calculate the optimal depth. k (s) Although propagation loss was calculated for (k=1,2,…,K), Embodiment 1 can also be applied as a method for calculating propagation loss for multiple search points when K=1, i.e., when the number of receiver depths is 1.

[0096] Furthermore, in Embodiments 1 and 2, the depth c of the sound velocity distribution n (z)The assumption was that (n=1,2,…,N) is common to all search points m=1,2,…,M, but each search point may have a different depth. In that case, when calculating the representative sound velocity distribution, the sound velocity distribution can be interpolated through appropriate interpolation so that it has a common depth, thereby calculating the representative sound velocity distribution.

[0097] Furthermore, in Embodiment 2, when the k-means method is used as the cluster analysis method, the number of subdivision groups U j I explained that it is necessary to predetermine (j=1,2,…,J). This is a fixed value U0 (U0=U1=U2=…=U) that does not depend on j (j=1,2,…,J). J ) can be set, and a trade-off between precision and computational complexity can be made depending on the value of U0. Also, the number of subdivision groups U j (j=1,2,…,J) can also be determined by the following equation (9).

number

[0098] Here I j The initial group G j This is the number of search locations belonging to the subdivision group Γ. ju (j=1,2,...J, u=1,2,...,U j The desired number of search locations belonging to the group is given by int(), and int() is a function that rounds the value in parentheses to an integer. The desired number I corresponds to the area over which the search locations belonging to the group are located, and can reflect prior knowledge about how much area is considered to have the same sound velocity distribution.

[0099] Furthermore, in Embodiment 2, cluster analysis is used to re-divide the groups, so group G j Subdivision group Γ obtained by subdividing (j=1,2,…,J) ju (u=1,2,…,U j Search site μ belonging to ) jui (i=1,2,…,I ju (r) ) center position (x ju(0) ,y ju (0) ) is calculated by formula (10).

number

[0100] Here, (x jui (r) ,y jui (r) ) is Group G j Subdivision group Γ obtained by subdividing (j=1,2,…,J) ju (u=1,2,…,U j Search site μ belonging to ) jui (i=1,2,…,I ju (r) This represents the position of ) in the xy coordinate system. And the subdivision group Γ ju (j=1,2,…,J, u=1,2,…,U j The speed of sound c of the representative sound velocity distribution jun (Γ) The location of the search point to which the group belongs is (x jui (r) ,y jui (r) ) and center position (x ju (0) ,y ju (0) It can also be calculated using ). For example, equation (11) is an equation that calculates the representative sound velocity distribution using a weighted average, where the closer the sound velocity distribution of a search point is to the center of the subdivided group, the greater its contribution to the representative sound velocity distribution, instead of the simple average of equation (6).

number

[0101] Here lol jui The center position of the subdivision group (x ju (0) ,y ju (0) ) and the location of the search points belonging to the subdivision group (x jui (r) ,y jui(r) This is a weight corresponding to the distance between ) and , and can be the reciprocal of the distance, for example, as in equation (12).

number

[0102] Alternatively, instead of the simple average in equation (6) or the weighted average in equation (11), the sound velocity distribution of the search point closest to the center can be used as the representative sound velocity distribution. [Explanation of symbols]

[0103] 1, 21, 501 Propagation loss calculation unit, 2, 22, 502 Detection distance calculation unit, 3, 23, 503 Optimal depth calculation unit, 4 Sound channel sequence calculation unit, 5 Group division unit, 6, 26 Optimal depth setting unit, 7 Group re-division unit, 100, 100A, 500 Information processing device.

Claims

1. A group division unit divides multiple search points within the target sea area into multiple groups containing similar sound velocity distributions, and calculates a representative sound velocity distribution, which is a typical sound velocity distribution, for each of the groups. An information processing apparatus comprising: a propagation loss calculation unit that calculates the propagation loss for each group based on the representative sound velocity distribution.

2. The system further includes a sound channel sequence calculation unit that calculates a sound channel sequence for each of the aforementioned search locations from the sound velocity distribution of the aforementioned search locations, The information processing device according to claim 1, wherein the group division unit classifies search locations with the same sound channel sequence into the same group.

3. A detection distance calculation unit calculates the detection distance for each group based on the propagation loss for each group, An optimal depth calculation unit calculates the optimal depth for each group based on the detection distance for each group, The information processing apparatus according to claim 1 or 2, further comprising: an optimal depth setting unit that sets the optimal depth for each group to the optimal depth of the search point belonging to the group.

4. The system further comprises a group subdivision unit that divides the multiple search points included in the group into subdivision groups based on the proximity of the search points, and calculates a representative sound velocity distribution, which is a typical sound velocity distribution, for each of the subdivision groups. The information processing apparatus according to claim 1 or 2, wherein the propagation loss calculation unit calculates the propagation loss for each of the re-divided groups based on the representative sound velocity distribution calculated by the group re-divided unit.

5. The steps include dividing multiple search points within the target sea area into multiple groups containing similar sound velocity distributions, The steps include: calculating a representative sound velocity distribution, which is a typical sound velocity distribution for each of the aforementioned groups; The steps include: calculating the propagation loss for each group based on the representative sound velocity distribution; An information processing method comprising the following:

6. The steps include dividing multiple search points within the target sea area into multiple groups containing similar sound velocity distributions, The steps include: calculating a representative sound velocity distribution, which is a typical sound velocity distribution for each of the aforementioned groups; The steps include: calculating the propagation loss for each group based on the representative sound velocity distribution; An information processing program that causes a computer to execute something.

Images