A smoothing method for acoustic propagation loss field based on RTS method

Abstract

The present invention discloses a method for smoothing a sound propagation loss field based on an RTS method. The present invention comprises the following steps: first, based on a parabolic equation model, a sound field is constructed using a sound velocity field and a terrain to obtain an original sound propagation loss field; then a physical guidance model for smoothing the sound propagation loss field is determined; a dynamic deduction model is established according to the physical guidance model, and a measurement model is established according to the relationship between the original sound propagation loss field and a true value, and then a discrete dynamic system model is determined; finally, the original sound propagation loss field is smoothed using an RTS filtering method to obtain a smoothed sound propagation loss field. Compared with an unsmoothed sound propagation loss field, the smoothed sound propagation loss field can better meet the requirements for low fluctuations in the sound field in engineering practice; at the same time, the smoothing method guided by the physical model can be closer to the actual sound field.

Description

Technical Field

[0001] The invention relates to a sound propagation loss field smoothing method in the field of underwater sound signal processing, and in particular to a sound propagation loss field smoothing method based on an RTS method. Background Art

[0002] As people pay more and more attention to marine resources, the demand for underwater operations has greatly increased. For example, underwater vehicles are equipped with sonar equipment for underwater communication, underwater navigation or underwater target tracking, and underwater environment monitoring. These marine tasks are inseparable from underwater perception based on sonar. Sonar range refers to the farthest distance of an object that can be detected by sonar, and is an important indicator of sonar in the application of underwater acoustic environment. In the process of predicting the sonar range, propagation loss is a very important indicator. There are two main ways to calculate the sound propagation loss used for sonar range prediction. The first way is to calculate the result through an ideal model. However, the ideal sound field calculated by various models is not exactly the same as the actual sound field. Under ideal conditions, the wave source radiates continuous waves; while the wave source in the actual sound field radiates pulse waves. A series of sound pressure maximum and minimum areas will appear near the wave source in the sound field, which is called the near field area of ​​the sound field. The uneven distribution of sound pressure in the near field area is due to the different distances from each point of the wave source to a point on the axis, the difference in wave path, and the phase difference when superimposed on each other, which interferes with each other, so that the sound pressure at some positions is strengthened, and at other positions is weakened, resulting in the appearance of sound pressure maximum and minimum points. Since the ideal sound field is a continuous wave, the sound waves radiated from various points of the wave source will produce complete interference at a certain point in the sound field; while the actual sound field is a pulse wave, the pulse wave lasts for a very short time, and the sound waves radiated from various points of the wave source will only produce incomplete interference or no interference at a certain point, resulting in the sound pressure variation amplitude on the axis of the near field area of ​​the actual sound field being smaller than that of the ideal sound field, the pole value is reduced, and it is relatively smooth.

[0003] According to the Fourier series, the pulse wave can be regarded as the sum of a constant term and an infinite number of harmonics and coharmonics of the fundamental frequency. Each frequency wave determines a sound field, and the total sound field is the superposition of different sound fields. Since the position of the sound pressure extreme point on the wave source axis changes with the wavelength, the extreme points of the sound field at different frequencies are different. After superposition, the total sound pressure tends to be uniform, which improves the uneven distribution of sound pressure in the near field.

[0004] The wave source of the actual sound field of the ideal sound source is non-uniformly excited, with a large amplitude at the center of the wave source and a small amplitude at the edge. Since the path difference caused by the edge of the wave source is large, the influence on interference is also relatively large, so the interference generated by this non-uniformly excited actual wave source is smaller than that of the ideal wave source with uniform excitation. When the excitation intensity of the wave source changes according to the Gaussian curve, the sound pressure on the axis of the near field will not have maximum and minimum values.

[0005] It should be noted that the sound pressure distribution of the actual sound field and the ideal sound field on the axis of the far field is basically the same. This is because when the distance to the wave source is far enough, the path difference from each point of the wave source to a point on the axis is significantly reduced, which greatly weakens the interference of the waves, or even eliminates them.

[0006] Due to the dramatic fluctuations in the ideal sound field calculation results, the sonar range obtained in this way also fluctuates greatly and does not conform to the actual situation. In practical applications, the dramatic predicted fluctuations will affect engineering applications. Therefore, it is necessary to smooth the sound field that stores the sound propagation loss. Summary of the invention

[0007] Aiming at the deficiencies of the existing methods, the present invention provides a sound propagation loss field smoothing method based on the RTS method.

[0008] The objective of the present invention is achieved through the following technical solutions:

[0009] 1. A method for smoothing the acoustic propagation loss field based on the RTS method

[0010] 1) Construct the sound field based on the sound velocity field and terrain to obtain the original sound propagation loss field;

[0011] 2) Determine the smooth guidance model of the sound propagation loss field;

[0012] 3) Constructing a discrete dynamic system model based on the original sound propagation loss field and the sound propagation loss field smoothing guidance model;

[0013] 4) Based on the discrete dynamic system model, the original sound propagation loss field is smoothed using the RTS filtering method to obtain a smoothed sound propagation loss field.

[0014] The sound propagation loss field smoothing guidance model is a simple sound propagation model in shallow waters.

[0015] The specific aspects of 3) are:

[0016] 3.1) A dynamic deduction model is constructed based on the smooth guidance model of the sound propagation loss field. The formula is as follows:

[0017] x k =P L +Q k

[0018] Among them, x k is the real state of the system at time k, Q k is the model error at time k; P L Provide guidance models for smoothing the acoustic propagation loss field;

[0019] 3.2) Construct a measurement model based on the original sound propagation loss field, the formula is as follows:

[0020] y k =x k +R k

[0021] Among them, y k is the measurement result at time k, R k is the measurement error at time k;

[0022] 3.3) The discrete dynamic system model is constructed using the dynamic deduction model and the measurement model. The formula is as follows:

[0023] x k ~p(x k |x k-1 )=N(x k |A k x k-1 ,Q k )

[0024] y k ~p(y k |x k )=N(y k |H k x k ,R k )

[0025] A k =P L (r k ) / P L (r k-1 )

[0026] H k =I

[0027] Q k ~N(0,Q)

[0028] R k ~N(0,R)

[0029] Among them, x k ~p(x k |x k-1 ) represents the conditional probability distribution satisfied by the real state of the system at time k, N() represents the multidimensional Gaussian distribution; P L (r k ) and P L (r k-1 ) are the sound propagation losses calculated by the simple sound propagation model at time k and time k-1, respectively. k is the state transfer matrix; y k ~p(y k |x k ) represents the conditional probability distribution satisfied by the measurement results of the system at time k; Hk is the observation matrix, I is the unit matrix, N(0,Q) is a Gaussian distribution with a mean of 0 and a variance of Q; N(0,R) is a Gaussian distribution with a mean of 0 and a variance of R.

[0030] 2. A computer device

[0031] The computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method when executing the computer program.

[0032] 3. A computer-readable storage medium

[0033] The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method described are implemented.

[0034] The beneficial effects of the present invention are:

[0035] 1) Compared with the unsmoothed sound propagation loss field, the smoothed sound propagation loss field can better meet the requirements of low fluctuation of the sound field in engineering practice;

[0036] 2) The smoothing method guided by the physical model has the potential to be closer to the actual sound field. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a flow chart of the sound propagation loss field smoothing method based on the RTS method proposed in the present invention.

[0038] Figure 2 It is the original sound field of the present invention and the smoothed result when the selected parameter Q has different values, R=0.02.

[0039] Figure 3 It is the result of smoothing the original sound field of the present invention and the selected parameter R when they have different values, Q=0.02.

[0040] Figure 4 is the smoothing result when the same parameters are selected under different physical guidance models in the present invention; wherein (a) is the smoothing result when the same parameters are selected under a simple sound propagation model, and (b) is the smoothing result when the same parameters are selected under a uniform model;

[0041] Figure 5 1 is a graph showing how the smoothing results vary with distance when the same parameters are selected under different physical guidance models in the present invention; (a) is a graph showing how the smoothing results vary with distance when the same parameters are selected under a simple sound propagation model, and (b) is a graph showing how the smoothing results vary with distance when the same parameters are selected under a uniform model. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the implementation of the present invention clearer, the technical solution in the embodiment of the present invention will be described in more detail below with reference to the accompanying drawings in the embodiment of the present invention.

[0043] The present invention is a sound propagation loss field smoothing method based on the RTS method. First, the sound field is constructed based on the parabolic equation model using the sound velocity field and the terrain to obtain the sound propagation loss field; then the physical guidance model for smoothing the sound propagation loss field is determined; a dynamic deduction model is established based on the physical guidance model, and a measurement model is established based on the relationship between the sound propagation loss field and the true value, and finally the RTS filtering method is used for smoothing to obtain a smoothed sound propagation loss field.

[0044] like Figure 1 As shown, the present invention comprises the following steps:

[0045] 1) Use the parabolic equation model to construct the sound field based on the sound velocity field and terrain to obtain the original sound propagation loss field, such as Figure 2 (a) and Figure 3 As shown in (a);

[0046] 2) Determine the smooth guidance model of the sound propagation loss field;

[0047] The guidance model for smoothing the sound propagation loss field is a simple shallow sea sound propagation model. The simple shallow sea sound propagation model is:

[0048] P L =15lgr+0.06r*10 -3

[0049] Among them, r is the sound propagation distance, that is, the distance from the current position to the sound source.

[0050] 3) Constructing a discrete dynamic system model based on the original sound propagation loss field and the sound propagation loss field smoothing guidance model;

[0051] 3) Specifically:

[0052] 3.1) A dynamic deduction model is constructed based on the smooth guidance model of the sound propagation loss field. The formula is as follows:

[0053] x k =P L +Q k

[0054] Among them, x k is the real state of the system at time k, Q k is the model error at time k; P L Provide guidance models for smoothing the acoustic propagation loss field;

[0055] 3.2) Construct a measurement model based on the original sound propagation loss field, the formula is as follows:

[0056] y k =x k +R k

[0057] Among them, y k is the measurement result at time k, R k is the measurement error at time k; model error Q k and measurement error R k All are Gaussian noise.

[0058] 3.3) The discrete dynamic system model is constructed using the dynamic deduction model and the measurement model. The formula is as follows:

[0059] x k ~p(x k |x k-1 )=N(x k |A k x k-1 ,Q k )

[0060] y k ~p(y k |x k )=N(y k |H k x k ,R k )

[0061] A k =P L (r k ) / P L (r k-1 )

[0062] H k =I

[0063] Q k ~N(0,Q)

[0064] R k ~N(0,R)

[0065] Among them, x k ~p(x k |x k-1 ) represents the conditional probability distribution satisfied by the real state of the system at time k, N() represents the multidimensional Gaussian distribution; P L (r k ) and P L (r k-1 ) are the sound propagation losses calculated by the simple sound propagation model at time k and time k-1, respectively. kis the state transfer matrix, which represents the conversion relationship between the state at the previous moment and the state at the current moment; k ~p(y k |x k ) represents the conditional probability distribution satisfied by the measurement results of the system at time k; H k is the observation matrix, which represents the conversion relationship between the measured value and the actual value; I is the unit matrix, N(0,Q) is a Gaussian distribution with a mean of 0 and a variance of Q; N(0,R) is a Gaussian distribution with a mean of 0 and a variance of R.

[0066] 4) Based on the discrete dynamic system model, the original sound propagation loss field is smoothed using the RTS filtering method to obtain a smoothed sound propagation loss field, which is used to predict the sonar range.

[0067] Specifically:

[0068] The Kalman filter is completed by completing T forward recursions from time 1 to T:

[0069] m′ k =A k m k-1

[0070]

[0071]

[0072]

[0073] m k =m′ k +K k (z k -H k m′ k )

[0074] P k =(IK k H k ) k '

[0075] Based on the backward filtering of Kalman filtering, the RTS filtering method adds forward filtering, which is backward recursive from time T through T times:

[0076]

[0077]

[0078]

[0079]

[0080]

[0081] The result obtained at this time This is the result of smoothing the sound propagation loss through RTS. Fixed R = 0.02, different Q values ​​lead to different filtering results as follows Figure 2 (b)- Figure 2 (f) is shown in Figure 2. Fixed Q = 0.02, different filtering results caused by different R are shown in Figure 2. Figure 3 (b)- Figure 3 (f) is shown. According to the modeling, R represents the difference between the sound propagation loss field calculated by the RAM model and the true value (i.e., the smoothed result). Therefore, the larger the R, the smoother the result. Q represents the tolerance of the smoothing model to the difference between the sound propagation loss values ​​at different locations. The larger the Q, the closer the smoothing result is to the result calculated by the RAM model, i.e., the lower the degree of smoothing. The simulation results are consistent with the theoretical situation, verifying the correctness of the model.

[0082] like Figure 4 (a) and Figure 4 (b) and Figure 5 (a) and Figure 5 As shown in (b), when the smoothing degree is set large enough, the smoothing results of the two models are observed. It can be seen that the smoothing results guided by the simple sound propagation model tend to increase with the increase of distance and depth, while the smoothing results guided by the uniform model are basically not affected by distance and depth, which is consistent with the theoretical situation, verifying the guiding effect of the physical guidance model on the smoothing process.

[0083] The contents described in the embodiments of this specification are merely an enumeration of the implementation forms of the inventive concept. The protection scope of the present invention should not be regarded as limited to the specific forms described in the embodiments. The protection scope of the present invention also extends to equivalent technical means that can be conceived by those skilled in the art based on the inventive concept.

Claims

1. A method for smoothing the acoustic propagation loss field based on the RTS method, characterized in that: The steps include: 1) Construct the sound field based on the sound velocity field and terrain to obtain the original sound propagation loss field; 2) Determine the smooth guidance model of the sound propagation loss field; 3) Constructing a discrete dynamic system model based on the original sound propagation loss field and the sound propagation loss field smoothing guidance model; The specific aspects of 3) are: 3.1) A dynamic deduction model is constructed based on the smooth guidance model of the sound propagation loss field. The formula is as follows: x k =P L +Q k Among them, x k is the real state of the system at time k, Q k is the model error at time k; P L Provide guidance models for smoothing the acoustic propagation loss field; 3.2) Construct a measurement model based on the original sound propagation loss field, the formula is as follows: y k =x k +R k Among them, y k is the measurement result at time k, R k is the measurement error at time k; 3.3) The discrete dynamic system model is constructed using the dynamic deduction model and the measurement model. The formula is as follows: x k ~p(x k |x k-1 )=N(x k |A k x k-1 ,Q k ) y k ~p(y k |x k )=N(y k |H k x k ,R k ) A k =P L (r k ) / P L (r k-1 ) H k =I Q k ~N(0,Q) R k ~N(0,R) Among them, x k ~p(x k |x k-1 ) represents the conditional probability distribution satisfied by the real state of the system at time k, N() represents the multidimensional Gaussian distribution; P L (r k ) and P L (r k-1 ) are the sound propagation losses calculated by the simple sound propagation model at time k and time k-1, respectively. k is the state transfer matrix; y k ~p(y k |x k ) represents the conditional probability distribution satisfied by the measurement results of the system at time k; H k is the observation matrix, I is the unit matrix, N(0,Q) is a Gaussian distribution with a mean of 0 and a variance of Q; N(0,R) is a Gaussian distribution with a mean of 0 and a variance of R; 4) Based on the discrete dynamic system model, the original sound propagation loss field is smoothed using the RTS filtering method to obtain a smoothed sound propagation loss field.

2. The method for smoothing the sound propagation loss field based on the RTS method according to claim 1, characterized in that: The sound propagation loss field smoothing guidance model is a simple sound propagation model in shallow waters.

3. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 2 are implemented.

4. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 2 are implemented.

Images