WO2022226856A1 - 浅海多层海底地声参数反演方法、装置、计算机设备及存储介质 - Google Patents

浅海多层海底地声参数反演方法、装置、计算机设备及存储介质 Download PDF

Info

Publication number
WO2022226856A1
WO2022226856A1 PCT/CN2021/090789 CN2021090789W WO2022226856A1 WO 2022226856 A1 WO2022226856 A1 WO 2022226856A1 CN 2021090789 W CN2021090789 W CN 2021090789W WO 2022226856 A1 WO2022226856 A1 WO 2022226856A1
Authority
WO
WIPO (PCT)
Prior art keywords
geoacoustic
value
sound pressure
parameter
seabed
Prior art date
Application number
PCT/CN2021/090789
Other languages
English (en)
French (fr)
Inventor
祝捍皓
薛洋洋
崔智强
王其乐
Original Assignee
浙江海洋大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 浙江海洋大学 filed Critical 浙江海洋大学
Priority to PCT/CN2021/090789 priority Critical patent/WO2022226856A1/zh
Priority to CN202180000994.5A priority patent/CN113330439A/zh
Priority to US17/588,224 priority patent/US20220357476A1/en
Publication of WO2022226856A1 publication Critical patent/WO2022226856A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
    • G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
    • G01V1/30—Analysis
    • G01V1/307—Analysis for determining seismic attributes, e.g. amplitude, instantaneous phase or frequency, reflection strength or polarity
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
    • G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
    • G01V1/30—Analysis
    • G01V1/303—Analysis for determining velocity profiles or travel times
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
    • G01V1/16—Receiving elements for seismic signals; Arrangements or adaptations of receiving elements
    • G01V1/18—Receiving elements, e.g. seismometer, geophone or torque detectors, for localised single point measurements
    • G01V1/186—Hydrophones
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
    • G01V1/38—Seismology; Seismic or acoustic prospecting or detecting specially adapted for water-covered areas
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00—Computing arrangements based on specific mathematical models
    • G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V2210/00—Details of seismic processing or analysis
    • G01V2210/60—Analysis
    • G01V2210/63—Seismic attributes, e.g. amplitude, polarity, instant phase
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/10—Numerical modelling

Definitions

  • the present application relates to the field of computer technology, and in particular to a method, device, computer equipment and storage medium for inversion of multi-layer submarine geoacoustic parameters in shallow seas.
  • Seabed geoacoustic parameters are one of the important parameters that make up the marine hydroacoustic environment.
  • the acoustic parameters of the seabed such as sound velocity, density, and sound velocity attenuation, have an important impact on the sound propagation in the marine environment, especially in the shallow sea environment.
  • the degree of mastery of the above-mentioned submarine geoacoustic parameters will directly affect the prediction and evaluation of the performance of underwater acoustic equipment, the numerical prediction of the ocean sound field, and the utilization of the characteristics of the ocean sound field. How to efficiently and accurately obtain the information of submarine geoacoustic parameters has always been a research hotspot in the field of underwater acoustics.
  • the acquisition methods of seabed geoacoustic parameters are currently divided into two categories: direct measurement and indirect measurement.
  • direct measurement method of obtaining seabed sediment samples through drilling sampling and other methods for identification the indirect measurement method of geoacoustic parameters represented by acoustic inversion technology is widely used because of its real-time, fast and efficient technical advantages.
  • the acquisition of seabed geoacoustic parameters Because conventional sonar mostly relies on mid/high frequency sound waves, in the past inversion studies of seabed geoacoustic parameters, most of them only focus on the acoustic properties of the seabed surface, and the seabed is assumed to be a liquid medium.
  • a shallow sea multi-layer seabed geoacoustic parameter inversion method comprising:
  • the geoacoustic parameters in each layer of each seabed model are parameters to be inverted, and the geoacoustic parameters include: density, shear wave speed of sound, longitudinal wave speed of sound, shear wave Attenuation, P-wave attenuation, and seafloor thickness;
  • a preset variation range corresponding to each geoacoustic parameter is obtained respectively, and the value of each geoacoustic parameter is randomly generated based on the preset variation range corresponding to each geoacoustic parameter, and based on each of the geoacoustic parameters
  • the theoretical sound pressure value is obtained by calculating the values of the geoacoustic parameters
  • the BIC value corresponding to each seabed model is obtained by using Bayesian theory
  • the seabed model with the smallest BIC value is used as the target seabed model, and the target geoacoustic parameter value corresponding to the target seabed model is used as the target inversion parameter value.
  • a shallow sea multi-layer submarine geoacoustic parameter inversion device comprising:
  • the establishment module is used to establish a plurality of seabed models, the number of layers corresponding to different seabed models is different, and the geoacoustic parameters in each layer of each seabed model are parameters to be inverted, and the geoacoustic parameters include: density, shear wave sound velocity , longitudinal wave sound velocity, shear wave attenuation, longitudinal wave attenuation and seafloor thickness;
  • a generation module configured to obtain a preset variation range corresponding to each geoacoustic parameter for each seabed model, and randomly generate the value of each of the geoacoustic parameters based on the preset variation range corresponding to each geoacoustic parameter, Calculate the theoretical sound pressure value based on the value of each of the geoacoustic parameters;
  • the acquisition module is used to acquire the actual sound pressure value obtained by the actual measurement
  • an update module configured to compare the theoretical sound pressure value with the actual sound pressure value, adjust and update the value of each geoacoustic parameter according to the comparison result, and return to execute the calculation based on the value of each geoacoustic parameter
  • the step of obtaining the theoretical sound pressure value is until the obtained theoretical sound pressure value matches the actual sound pressure value, and the value of each of the geoacoustic parameters corresponding to the matched theoretical sound pressure value is used as the corresponding value of the parameter to be inverted. target acoustic parameter value;
  • a calculation module used for calculating the BIC value corresponding to each seabed model by adopting Bayesian theory according to the target geoacoustic parameter value corresponding to each seabed model;
  • a determination module takes the seabed model with the smallest BIC value as the target seabed model, and uses the target geoacoustic parameter value corresponding to the target seabed model as the target inversion parameter value.
  • a computer device includes a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the following steps:
  • the geoacoustic parameters in each layer of each seabed model are parameters to be inverted, and the geoacoustic parameters include: density, shear wave speed of sound, longitudinal wave speed of sound, shear wave Attenuation, P-wave attenuation, and seafloor thickness;
  • a preset variation range corresponding to each geoacoustic parameter is obtained respectively, and the value of each geoacoustic parameter is randomly generated based on the preset variation range corresponding to each geoacoustic parameter, and based on each of the geoacoustic parameters
  • the theoretical sound pressure value is obtained by calculating the values of the geoacoustic parameters
  • the BIC value corresponding to each seabed model is obtained by using Bayesian theory
  • the seabed model with the smallest BIC value is used as the target seabed model, and the target geoacoustic parameter value corresponding to the target seabed model is used as the target inversion parameter value.
  • a computer-readable storage medium storing a computer program, when executed by a processor, the computer program causes the processor to perform the following steps:
  • the geoacoustic parameters in each layer of each seabed model are parameters to be inverted, and the geoacoustic parameters include: density, shear wave speed of sound, longitudinal wave speed of sound, shear wave Attenuation, P-wave attenuation, and seafloor thickness;
  • a preset variation range corresponding to each geoacoustic parameter is obtained respectively, and the value of each geoacoustic parameter is randomly generated based on the preset variation range corresponding to each geoacoustic parameter, and based on each of the geoacoustic parameters
  • the theoretical sound pressure value is obtained by calculating the values of the geoacoustic parameters
  • the BIC value corresponding to each seabed model is obtained by using Bayesian theory
  • the seabed model with the smallest BIC value is used as the target seabed model, and the target geoacoustic parameter value corresponding to the target seabed model is used as the target inversion parameter value.
  • the above-mentioned method, device, computer equipment and storage medium for inversion of multi-layer seabed geoacoustic parameters in shallow sea First, multiple seabed models are established, and different seabed models correspond to different layers, and then each seabed model is randomly generated for each geoacoustic sound. The theoretical sound pressure value is calculated based on the value of each geoacoustic parameter. By comparing the theoretical sound pressure value with the actual sound pressure value, the theoretical sound pressure value matching the actual sound pressure value is determined, and then the theoretical sound pressure value is determined. The target geoacoustic parameter values corresponding to each seabed model are finally calculated by Bayesian theory to obtain the BIC value of each seabed model, and the seabed model with the smallest BIC value is used as the target seabed model.
  • the target geoacoustic parameter value is obtained by inversion by comparing the calculated theoretical sound pressure value with the actual sound pressure value, and the Bayesian theory is used for each seabed model to calculate the BIC value, which is determined according to the BIC value.
  • This method not only realizes the efficient and accurate inversion of the geoacoustic parameter values in the target seabed model, but also determines the optimal number of layers of the seabed model.
  • 1 is a flow chart of a method for inversion of multi-layer seabed geoacoustic parameters in shallow seas in one embodiment
  • Fig. 2 is a multi-layer seabed parametric model diagram in one embodiment
  • Fig. 3 is a structural block diagram of a shallow sea multi-layer submarine geoacoustic parameter inversion device in one embodiment
  • Figure 4 is a diagram of the internal structure of a computer device in one embodiment.
  • the shallow sea multi-layer submarine geoacoustic parameter inversion method specifically includes the following steps:
  • Step 102 establish a plurality of seabed models, the number of layers corresponding to different seabed models is different, the geoacoustic parameters in each layer of each seabed model are parameters to be inverted, and the geoacoustic parameters include: density, shear wave speed of sound, longitudinal wave speed of sound, Shear wave attenuation, longitudinal wave attenuation, and seafloor thickness.
  • the establishment of multiple different seabed models is to subsequently find the best seabed model that is consistent with the actual situation.
  • Different seabed models have different layers, that is, different seabed models have different seabed structures.
  • the seabed model is established based on wave theory, and the established seabed model is represented by equations.
  • the equations of the established seabed model involve geoacoustic parameters and sound pressure, that is, geoacoustic parameters and sound pressure are parameters in the seabed model.
  • the geoacoustic parameters in this application are relatively comprehensive, and multiple factors such as density, shear wave sound velocity, longitudinal wave sound velocity, shear wave attenuation, longitudinal wave attenuation, and seabed thickness are simultaneously considered in each layer.
  • each layer contains the corresponding geoacoustic parameters, in the figure, c p , c s , ⁇ b , ⁇ p and ⁇ s represent longitudinal wave sound speed, Shear wave sound velocity, seabed density, longitudinal wave sound velocity attenuation and shear wave sound velocity attenuation, z is the water depth, z s is the sound source water depth, r is the propagation distance, f 0 is the sound source frequency, and the subscripts represent the number of layers.
  • Step 104 For each seabed model, obtain a preset variation range corresponding to each geoacoustic parameter, and randomly generate the value of each geoacoustic parameter based on the preset variation range corresponding to each geoacoustic parameter, and based on the value of each geoacoustic parameter. Calculate the theoretical sound pressure value.
  • the preset variation range of each geoacoustic parameter refers to the preset variation range of the value of the geoacoustic parameter. Since both the geoacoustic parameters and the sound pressure are unknown parameters in the seabed model equation, and there are many geoacoustic parameters, this equation cannot be solved directly.
  • it is a process of continuous optimization by assigning values to each geoacoustic parameter.
  • the assignment method is that each geoacoustic parameter randomly generates the value of the corresponding geoacoustic parameter within the preset variation range, and then the sound pressure value, that is, the theoretical sound pressure value, can be calculated based on the value of each geoacoustic parameter.
  • the variation range of each geoacoustic parameter can be set according to the actual situation. 2000), shear wave sound velocity (900-1100), longitudinal wave attenuation dB/ ⁇ (0.09-0.11), shear wave attenuation dB/ ⁇ (0.09/0.11), thickness ranging from tens of meters (15-25). Values for geoacoustic parameters are then generated based on preset ranges, for example, density: 1.5, longitudinal wave speed of sound 1950, shear wave speed of 900, longitudinal wave attenuation 0.095, shear wave attenuation 0.096, and seafloor thickness 18.
  • step 106 the actual sound pressure value obtained by the actual measurement is obtained.
  • the actual sound pressure value can be measured, and the measurement method can be measured by using a hydrophone.
  • the sound wave emitted by the sound source is monitored through the hydrophone, and then the monitored sound wave is processed to obtain the actual sound pressure value.
  • the result measured by the hydrophone is the audio in wav format, which is imported into matlab and converted into a numerical form, and then the frequency spectrum of this set of data is obtained through Fourier transform, and the amplitude value is the sound pressure value.
  • the actual sound pressure value obtained is composed of a group of sound pressure values, for example, a group of sound pressure values contains 1000 values.
  • Step 108 compare the theoretical sound pressure value with the actual sound pressure value, adjust and update the value of each geoacoustic parameter according to the comparison result, and return to the step of calculating the theoretical sound pressure value based on the value of each geoacoustic parameter, until the obtained theoretical sound pressure value is obtained.
  • the sound pressure value matches the actual sound pressure value, and the value of each geoacoustic parameter corresponding to the matched theoretical sound pressure value is used as the target geoacoustic parameter value corresponding to the parameter to be inverted.
  • the error function to calculate the error value between the theoretical sound pressure value and the actual sound pressure value
  • the theoretical sound pressure value and the actual sound pressure value do not match, it is necessary to update each ground sound within the preset range of each ground sound parameter.
  • the value of the sound parameters, the theoretical sound pressure value is calculated again, and then compared, and the calculation is repeated several times until the calculated theoretical sound pressure value matches the actual sound pressure value.
  • the judgment condition for matching the theoretical sound pressure value and the actual sound pressure value can have various forms. One is to preset the minimum error value, and when the error between the two is smaller than the minimum error value, it is judged that the two match.
  • each geoacoustic parameter is taken as the target geoacoustic parameter value corresponding to the parameter to be inverted.
  • the target geoacoustic parameter value is the value of the geoacoustic parameter obtained through inversion.
  • the theoretical sound pressure value is calculated by assigning values to each geoacoustic parameter, and the inversion of each geoacoustic parameter is realized by comparing the theoretical sound pressure value with the actual sound pressure value.
  • the method of comparing the theoretical sound pressure value and the actual sound pressure value is used to invert each geoacoustic parameter, thereby realizing the efficient and accurate determination of the value of each geoacoustic parameter.
  • Step 110 Calculate the BIC value corresponding to each seabed model by using Bayesian theory according to the target geoacoustic parameter value corresponding to each seabed model.
  • BIC Bayesian Information Criterion, Bayesian Information Criterion
  • step 112 the seabed model with the smallest BIC value is used as the target seabed model, and the target geoacoustic parameter value corresponding to the target seabed model is used as the target inversion parameter value.
  • the target geoacoustic parameter value corresponding to the obtained target seabed model is used as the target inversion parameter value of the geoacoustic parameter (ie, the target inversion result).
  • the above shallow sea multi-layer seabed geoacoustic parameter inversion method first, establish multiple seabed models, different seabed models correspond to different layers, and then randomly generate the value of each geoacoustic parameter for each seabed model, and based on each seabed model.
  • the theoretical sound pressure value is obtained by calculating the value of the sound parameter. By comparing the theoretical sound pressure value with the actual sound pressure value, the theoretical sound pressure value matching the actual sound pressure value is determined, and then the target destination corresponding to each seabed model is determined. Finally, the Bayesian theory is used to calculate the BIC value of each seabed model, and the seabed model with the smallest BIC value is used as the target seabed model.
  • the target geoacoustic parameter value is obtained by inversion by comparing the calculated theoretical sound pressure value with the actual sound pressure value, and the Bayesian theory is used for each seabed model to calculate the BIC value, which is determined according to the BIC value.
  • This method not only realizes the efficient and accurate inversion of the geoacoustic parameter values in the target seabed model, but also determines the optimal number of layers of the seabed model.
  • establishing a plurality of seabed models with different layers corresponding to different seabed models includes: constructing, according to the wave theory, a displacement potential function equation corresponding to each layer in each seabed model; according to the displacement potential function
  • the general solution of each displacement potential function is obtained by equation calculation, and the general solution of each displacement potential function includes a plurality of uncertain coefficients, and the plurality of uncertain coefficients are related to the geoacoustic parameters, and the theoretical sound pressure value is calculated from the displacement potential function.
  • the physical quantities of the seafloor of each layer are represented by the displacement potential function under cylindrical coordinates, and the displacement potential function of each layer is established to satisfy the wave equation equations.
  • the specific The displacement potential function of each layer is shown.
  • each coefficient of the equation system is actually solved to obtain the displacement potential functions of each layer.
  • the sound pressure value at each point in the fluid layer can be obtained.
  • ⁇ (r,z) represents the sound source equation
  • r is the signal propagation distance
  • z is the vertical depth
  • ⁇ is the Laplace operator
  • ⁇ 1 is the displacement potential function in the fluid layer
  • ⁇ p is the elastic seafloor median scalar displacement potential function
  • ⁇ s is Vector displacement potential function
  • the acquiring the actual sound pressure value obtained by the actual measurement includes: using a hydrophone to monitor the sound waves emitted by the sound source, the sound waves are generated by transmitting in water by a transmitting transducer, and the hydrophone The measurement is completed by relative movement between the transmitter and the transmitting transducer; the audio in wav format detected by the hydrophone is imported into matlab and converted into a set of numerical values; Fourier transform is used to measure the set of numerical values Perform processing to obtain a frequency spectrum corresponding to the set of values; calculate the amplitude of the frequency spectrum to obtain the actual sound pressure value, where the actual sound pressure value includes sound pressure values at multiple positions.
  • wav is a sound file format.
  • matlab is a mathematical software used for data analysis etc.
  • the actual marine environment is generally a low-frequency signal, so the propagation distance will be farther, to achieve the purpose of carrying more seabed information, generally around 100HZ, and the launch position can be several meters or tens of meters underwater.
  • a ship is generally used to carry a hydrophone or a sound source for movement.
  • the position of the sound source is fixed, and the experimental boat carries the hydrophone to move to complete the measurement; or the hydrophone is fixed, and the experimental boat carries the sound source to move to complete the measurement;
  • the transmitting transducer is the sound source equipment in the experiment, and the transmitting transducer can Sound waves are launched into the water, and they are described by sound sources in theoretical writing and analysis.
  • the sound-emitting device in the experiment is the transmitting transducer.
  • the transmitting transducer and the hydrophone are generally fixed on the ship with the power supply end, and the transmitting and receiving ends are lowered into the water through the cable, and the depth is determined according to the experimental design. Import the measured audio in wav format into matlab and convert it into a numerical form. After Fourier transform, the spectrum of this set of data is obtained, and the amplitude value is the sound pressure value. When measuring, it will not only measure the sound pressure at one location, but measure the sound pressure at different locations by moving the ship with a hydrophone or a sound source, thereby obtaining a set of sound pressure values. The measurement of the above-mentioned actual sound pressure value is based on the actual marine environment, and the measured value is a set of sound pressure values, which is reliable and accurate.
  • a board made of polyvinyl chloride material is used to simulate the seabed.
  • High-frequency underwater sound is emitted by a sound source in a fixed position.
  • the receiving hydrophone is measured once at a fixed distance.
  • the transmitting transducer is fixed in one end of the water, and the receiving hydrophone is fixed in the mobile micro
  • the worktable moves 2mm each time, and the measurement error is less than 20um.
  • Use the computer to control the mobile workbench, measure and acquire data. After completing the measurement at one position, the table automatically moves to the next position, and a total of 1000 points are measured.
  • the comparing the theoretical sound pressure value with the actual sound pressure value, and adjusting and updating the values of the respective geoacoustic parameters according to the comparison result includes: using an error function to calculate the theoretical sound pressure value and the actual sound pressure value
  • the error value between the pressure values, the formula of the error function is as follows:
  • the error value between the theoretical sound pressure value and the actual sound pressure value is calculated by the error function.
  • the error function is designed by Bayesian theory. Under Bayesian theory, the theoretical sound pressure and the actual sound pressure are established by combining the likelihood function. In this theory, when the error function reaches the minimum value, it indicates that the similarity between the theoretical sound pressure and the actual sound pressure reaches the maximum, that is, the theoretical sound pressure at this time is equal to the actual sound pressure.
  • the error function can accurately reflect the difference between the theoretical sound pressure value and the actual sound pressure value, thereby facilitating better matching to obtain a theoretical sound pressure value matching the actual sound pressure value.
  • a preset variation range corresponding to each geoacoustic parameter is obtained respectively, and each geoacoustic parameter is randomly generated based on the preset variation range corresponding to each geoacoustic parameter
  • Calculate the theoretical sound pressure value based on the value of each of the geoacoustic parameters including: acquiring the initial value of each geoacoustic parameter, the initial value is randomly generated based on the preset variation range;
  • the initial values of the various geoacoustic parameters and the preset variation range are perturbed by the improved simulated annealing method to generate new values of the geoacoustic parameters; according to the new values of the geoacoustic parameters, the corresponding New theoretical sound pressure value;
  • the step of calculating the pressure value until the obtained theoretical sound pressure value matches the actual sound pressure value including: calculating and obtaining a new error value according to the new theoretical sound pressure value and the actual sound pressure value, The error value is compared with the previous error value, the smaller error value and the corresponding geoacoustic parameters are retained, and the execution of the perturbation based on the initial value of each geoacoustic parameter and the preset variation range to generate a new one is returned.
  • the theoretical sound pressure value corresponding to each geoacoustic parameter value retained at the end is used as the sound pressure value matching the actual sound pressure value.
  • the process of determining the theoretical sound pressure value matching the actual sound pressure value is the inversion process of the geoacoustic parameters.
  • the preset variation range of each geoacoustic parameter is set, and then the geoacoustic parameter is initialized.
  • the process of initializing the geoacoustic parameter is to randomly generate the initial value of each geoacoustic parameter within the preset variation range.
  • each initial value is brought into the seabed model to calculate the theoretical sound pressure value, and the theoretical sound pressure value and the actual sound pressure value are brought into the error function to obtain the error value.
  • the error value is used to measure the difference between the theoretical sound pressure value and the actual sound pressure value. The smaller the error value is, the closer the theoretical sound pressure value and the actual sound pressure value are.
  • a perturbation algorithm is used to generate a new value within a preset range by perturbing the initial value as the center to obtain a new value of each of the geoacoustic parameters, and then calculate a new theoretical sound pressure value, and compare the new theoretical sound pressure value with the The actual sound pressure value is calculated by the error function to obtain a new error, the new error value is compared with the initial error value, and the smaller error value and the corresponding geoacoustic parameter value are retained.
  • the above-mentioned perturbation algorithm is used to generate a new value within a preset range through perturbation with the initial value as the center, to obtain a new value of each of the geoacoustic parameters, and to calculate a new error value.
  • the theoretical sound pressure value with smaller error value and the corresponding geoacoustic parameters are retained until convergence is reached, and the value of the geoacoustic parameter corresponding to the last retained theoretical sound pressure value is used as the inversion value.
  • a preset variation range is set for each geoacoustic parameter, thereby ensuring that the randomly generated geoacoustic parameters will not deviate from the actual, and can also ensure randomness, thereby achieving accurate to determine the value of the target geoacoustic parameter.
  • the generating a new value of each of the geoacoustic parameters based on the initial value of each geoacoustic parameter and the preset variation range using an improved simulated annealing method includes: obtaining the current value of the geoacoustic parameter. The number of iterations, the disturbance coefficient is determined according to the current number of iterations; the disturbance condition is obtained, and the disturbance condition is that the lower seabed parameter in the multi-layer seabed model is greater than the upper seabed parameter; according to the preset variation range, the disturbance coefficient and the disturbance condition A new value for each of the geoacoustic parameters is randomly generated.
  • the number of iterations determines the amplitude of the random disturbance.
  • the number of iterations is inversely correlated with the amplitude of the disturbance, and the number of iterations is inversely correlated with the simulated annealing temperature.
  • the lower the simulated annealing temperature the higher the number of iterations, and the smaller the corresponding disturbance amplitude.
  • the disturbance condition is that the parameters of the lower seabed in the multi-layer seabed model are greater than those of the upper seabed.
  • the perturbation process is calculated as follows:
  • Step 1 Set a preset variation range (ie, upper and lower boundaries) for the parameters to be inverted. After the algorithm performs disturbance, the results are kept within this range. The parameter values beyond this range will be eliminated by the out-of-bounds function.
  • the temperature Tmax, the termination temperature Tmin (that is, setting the outer loop termination condition) and the length L of the Markov chain (Markov) are used to indicate the number of populations initially set, which is how many groups to study. For example, the longitudinal wave speed of sound is set to 1000. , that is, each disturbance optimization is 1000 longitudinal wave sound speeds.
  • the second step randomly generate an initial value for each parameter, where m 0 represents the initial value of the parameter to be inverted, S min represents the lower boundary of each parameter interval; SL represents the parameter interval width, that is, the upper boundary minus the lower boundary ; rand(0,1) is a matlab function that can generate random numbers between 0-1.
  • Step 3 Substitute the generated initial value into the seabed model to calculate the error value corresponding to the set of parameters and retain E(m i ).
  • m new represents the new solution after perturbation
  • m now represents the current solution (initial solution in the first cycle)
  • S max represents the upper boundary of the parameter interval
  • a represents the perturbation coefficient.
  • a (1-rand(0,1) ⁇ (1-(t/T) ⁇ b))
  • t represents the current number of iterations
  • T represents the preset total number of iterations
  • b controls the search step
  • the empirical value is generally 2.
  • the t value keeps increasing, that is, the a value maintains a large value when the temperature is high, and maintains a small value when the temperature is low, which can ensure a large amount of disturbance during the initial search.
  • the search interval gradually decreases until the final algorithm converges.
  • Step 6 Judge whether the inner loop termination condition is met (whether the error value reaches convergence), if not, return to step 4, if so, judge whether the outer loop termination condition is met (whether the temperature is less than Tmin), if not, perform cooling, If satisfied, terminate the calculation and output the result.
  • This set of sound pressures is a set of numbers with the same dimension as the actual sound pressure.
  • the sound pressure value of 1 point, then the theoretically calculated sound pressure value is also the sound pressure value of 1000 points.
  • the theoretical sound pressure and the actual sound pressure are brought into the error function to obtain the error value, such as -5, and then after disturbance Then give a set of values, 1.5, 2000, 1000, 0.01, 0.01, 20.
  • calculating the BIC value corresponding to each seabed model according to the target geoacoustic parameter value corresponding to each seabed model using Bayesian theory includes: according to the target corresponding to each seabed model The geoacoustic parameter value and error value are calculated by using the improved Bayesian theory to obtain the BIC value corresponding to each seabed model, and the calculation of the BIC value is realized by the following formula:
  • M is the number of parameters in the model
  • N is the number of data
  • the calculation formula of the BIC value is obtained by deduction, and the size of the BIC value is determined by the error function, the number of model parameters and the number of data, thereby avoiding under-parameterization and over-parameterization of the model, and more effectively select the Optimal seafloor model.
  • the Bayesian theory, the error function, and the BIC formula are derived as follows:
  • the random variables d and m respectively represent the experimental data extracted in the scale-down experiment and the parameters of the seabed model, and N and M respectively represent the number of the vector d and the vector m.
  • the vectors d and m satisfy Bayes' theorem:
  • d) is the posterior probability density (PPD)
  • m) of d is usually expressed by the likelihood function L(m)
  • P(m) is the prior probability density of m function that represents the available model parameter prior information independent of the data
  • P(d) is the probability density function of parameter d. Since P(d) has nothing to do with the parameter m, it can be regarded as a constant, and the above formula can be changed to:
  • the likelihood function is determined by the form of the data and the statistical distribution of errors in the data. Considering that in the actual application process, the statistical characteristics of the error are difficult to obtain independently, the assumption of unbiased Gaussian error is adopted in the processing process, and the form of the likelihood function is:
  • E(m) is the error function, which can be obtained after normalization
  • the integration domain spans the M-dimensional parameter space, and M is the number of parameters to be inverted.
  • the posterior probability density PPD
  • PPD posterior probability density Due to the problem of pair-dimensional parameters in the inversion, in order to interpret the inversion results of the parameters more reasonably, it is necessary to study the correlation characteristics between the model parameters, such as the MAP value, the mean value, and the one-dimensional probability density distribution of the parameters, which are respectively defined as :
  • the likelihood function L(m) needs to be obtained to solve the parameter PPD.
  • the likelihood function is related to the statistical distribution of data errors (including measurement errors and theoretical errors) and is an important factor in quantitatively describing parameter uncertainty. index. This paper assumes that the data error is an independent and identically distributed random variable, then the likelihood function can be expressed as:
  • p f mea represents the measured sound pressure received by a single sensor at position k at frequency f
  • p f pre and C f m represent the model predicted sound pressure and covariance matrix, respectively.
  • the predicted sound pressure p f pre can be expressed by the following formula
  • p f FFM represents the sound pressure calculated by the fast field method (FFM)
  • a f and ⁇ f are the amplitude and phase of the unknown sound source at each frequency.
  • the maximum likelihood estimate is:
  • B f (m) represents the normalized Bartlett mismatcher.
  • Equation (22) into Equation (12) and Equation (20) to get the corresponding error function E(m) that satisfies the estimated value of the maximum likelihood function
  • M is the number of parameters in model I
  • N is the number of data parameters.
  • the model with the smallest BIC value is the optimal model. It can be seen from formula (25) that the size of the BIC value is jointly determined by the error function, the number of model parameters and the number of data, thereby avoiding under-parameterization and over-parameterization models, and more effectively selecting the optimal seabed model.
  • a shallow sea multi-layer submarine geoacoustic parameter inversion device including:
  • the establishment module 302 is used to establish a plurality of seabed models, the number of layers corresponding to different seabed models is different, and the geoacoustic parameters in each layer of each seabed model are parameters to be inverted, and the geoacoustic parameters include: density, shear wave Speed of sound, speed of sound for longitudinal waves, attenuation of shear waves, attenuation of longitudinal waves and seafloor thickness;
  • the generating module 304 is configured to, for each seabed model, obtain a preset variation range corresponding to each geoacoustic parameter, and randomly generate the value of each geoacoustic parameter based on the preset variation range corresponding to each geoacoustic parameter , the theoretical sound pressure value is calculated based on the value of each of the geoacoustic parameters;
  • an acquisition module 306, configured to acquire the actual sound pressure value obtained by actual measurement
  • the updating module 308 is configured to compare the theoretical sound pressure value with the actual sound pressure value, adjust and update the value of each geoacoustic parameter according to the comparison result, and return to execute the value based on the each geoacoustic parameter.
  • the step of calculating the theoretical sound pressure value until the obtained theoretical sound pressure value matches the actual sound pressure value, and using the value of each of the geoacoustic parameters corresponding to the matched theoretical sound pressure value as the corresponding parameter to be inverted The target geoacoustic parameter value;
  • the calculation module 310 is used to calculate the BIC value corresponding to each seabed model by adopting Bayesian theory according to the target geoacoustic parameter value corresponding to each seabed model;
  • the determination module 312 uses the seabed model with the smallest BIC value as the target seabed model, and uses the target geoacoustic parameter value corresponding to the target seabed model as the target inversion parameter value.
  • the establishment module 302 is further configured to construct a displacement potential function equation corresponding to each layer in each seabed model according to the wave theory; calculate the general solution of each displacement potential function according to the displacement potential function equation, and the The general solution of each displacement potential function includes a plurality of uncertainty coefficients, the plurality of uncertainty coefficients are related to the geoacoustic parameters, and the theoretical sound pressure value is calculated according to the displacement potential function.
  • the acquisition module 306 is further configured to use a hydrophone to monitor the sound waves emitted by the sound source, the sound waves are generated by transmitting in water by a transmitting transducer, the hydrophone and the transmitting transducer The measurement is completed by relative movement; the audio in wav format detected by the hydrophone is imported into matlab and converted into a set of values; Fourier transform is used to process the set of values to obtain the set of values.
  • the frequency spectrum corresponding to the numerical value; the actual sound pressure value is obtained by calculating the amplitude of the frequency spectrum, and the actual sound pressure value includes sound pressure values at multiple positions.
  • the update module 308 is further configured to calculate the error value between the theoretical sound pressure value and the actual sound pressure value by using an error function, and the formula of the error function is as follows:
  • the generating module 304 is further configured to acquire initial values of various geoacoustic parameters, where the initial values are randomly generated based on the preset variation range; based on the initial values of the various geoacoustic parameters and the The preset variation range is disturbed by the improved simulated annealing method to generate new values of each of the geoacoustic parameters; corresponding new theoretical sound pressure values are calculated according to the new values of the geoacoustic parameters; the update module 308 further It is used to calculate and obtain a new error value according to the new theoretical sound pressure value and the actual sound pressure value, compare the new error value with the previous error value, and retain the smaller error value and the corresponding ground.
  • Acoustic parameters return to performing the step of performing the perturbation based on the initial values of the respective geoacoustic parameters and the preset variation range to generate new values of the geoacoustic parameters, until the convergence condition is reached, and the last reserved geoacoustic parameters are The theoretical sound pressure value corresponding to the sound parameter value is used as the sound pressure value matching the actual sound pressure value.
  • the generation module 304 is further configured to obtain the current annealing temperature, determine a disturbance coefficient according to the current annealing temperature; determine a disturbance magnitude according to the disturbance coefficient; obtain a disturbance condition, where the disturbance condition is in the multi-layer seabed model The lower-layer seabed parameter is greater than the upper-layer seabed parameter; and a new value of each of the geoacoustic parameters is randomly generated according to the preset variation range, the disturbance amplitude and the disturbance condition.
  • the calculation module 310 is further configured to calculate the BIC value corresponding to each seabed model by using the improved Bayesian theory according to the target geoacoustic parameter value and the error value corresponding to each seabed model.
  • the value is calculated using the following formula:
  • M is the number of parameters in the model
  • N is the number of data
  • Figure 4 shows an internal structure diagram of a computer device in one embodiment.
  • the computer device may be a terminal or a server.
  • the computer device includes a processor, memory, and a network interface connected by a system bus.
  • the memory includes a non-volatile storage medium and an internal memory.
  • the nonvolatile storage medium of the computer device stores an operating system, and also stores a computer program, which, when executed by the processor, enables the processor to implement the above-mentioned method for inversion of multi-layer submarine geoacoustic parameters in shallow seas.
  • a computer program can also be stored in the internal memory.
  • the processor can execute the above-mentioned method for inversion of multi-layer subsea geoacoustic parameters in shallow seas.
  • FIG. 4 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer equipment to which the solution of the present application is applied. Include more or fewer components than shown in the figures, or combine certain components, or have a different arrangement of components.
  • a computer device comprising a memory and a processor, the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to execute the above-mentioned shallow sea multi-layer seabed Steps of the Geoacoustic Parameter Inversion Method.
  • a computer-readable storage medium which stores a computer program, and when the computer program is executed by a processor, causes the processor to execute the steps of the above-mentioned method for inversion of multi-layer submarine geoacoustic parameters in shallow seas .
  • Nonvolatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in various forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Road (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
  • SRAM static RAM
  • DRAM dynamic RAM
  • SDRAM synchronous DRAM
  • DDRSDRAM double data rate SDRAM
  • ESDRAM enhanced SDRAM
  • SLDRAM synchronous chain Road (Synchlink) DRAM
  • SLDRAM synchronous chain Road (Synchlink) DRAM
  • Rambus direct RAM
  • DRAM direct memory bus dynamic RAM
  • RDRAM memory bus dynamic RAM

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Remote Sensing (AREA)
  • General Physics & Mathematics (AREA)
  • Geophysics (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • Geology (AREA)
  • Environmental & Geological Engineering (AREA)
  • Acoustics & Sound (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Oceanography (AREA)
  • Pure & Applied Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Mathematics (AREA)
  • Software Systems (AREA)
  • Algebra (AREA)
  • Probability & Statistics with Applications (AREA)
  • Mathematical Physics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Geophysics And Detection Of Objects (AREA)

Abstract

本申请公开了一种浅海多层海底地声参数反演方法，包括：建立多个海底模型，不同海底模型对应的层数不同，基于每个地声参数对应的预设变化范围随机生成各个地声参数的值，继而计算得到理论声压值，与实际声压值进行比较，根据比较结果调整更新各个地声参数的值，直到得到的理论声压值和实际声压值相匹配，得到目标地声参数值；计算得到每个海底模型对应的BIC值；将BIC值最小的海底模型作为目标海底模型，将目标海底模型对应的目标地声参数值作为目标反演参数值。该方法不仅准确地确定了海底模型的层数，且高效、准确地通过反演得到了目标海底模型中的地声参数值。此外，还提出了一种浅海多层海底地声参数反演装置、计算机设备及存储介质。

Description

浅海多层海底地声参数反演方法、装置、计算机设备及存储介质 技术领域
本申请涉及计算机技术领域，具体涉及一种浅海多层海底地声参数反演方法、装置、计算机设备及存储介质。
背景技术
海底地声参数是组成海洋水声环境的重要参数之一，海底的声速、密度、声速衰减等声学参数对海洋环境中，尤其是浅海环境中的声传播有着重要影响。对上述海底地声参数的掌握程度，将直接影响对水声设备性能的预测与评估、海洋声场的数值预报、海洋声场特征的利用等。如何高效、准确获取海底地声参数信息一直是水声领域的研究热点。
对海底地声参数的获取方法目前分为直接测量与间接测量两类。相比通过钻孔取样等方式获取海底底质样品进行鉴别的直接测量方法，以声学反演技术为代表的地声参数间接测量方法因其具有实时、快速、高效的技术优势，而被广泛应用于海底地声参数的获取。由于常规声呐多依托中/高频段声波开展工作，因而在以往对海底地声参数的反演研究中多只关注海底表层声学特性，且假设海底为液态介质。随着近年来声呐设备向低频/甚低频的发展，以往只对海底表层声学特性的了解和掌握已不能满足对当前声传播问题的解析和验证，开展包含海底结构在内的、深层海底地声参数反演技术的研究愈发迫切。且已有研究成果已经证明，在研究低频/甚低频水声传播问题时，海底横波声速的影响不可忽略，因而在当前海底地声参数反演问题的研究中，将海底视为分层弹性介质，对包含分层结构、横波声速及其衰减在内、深层地声参数进行准确反演是当前海底地声参数的发展目标，相关研究工作亟待开展。
因此，基于上述技术问题需要设计一种新的浅海多层海底地声参数反演方法。
申请内容
基于此，有必要针对上述问题，提出一种高效、准确的浅海多层海底地声参数反演方法、装置、计算机设备及存储介质。
一种浅海多层海底地声参数反演方法，包括：
建立多个海底模型，不同海底模型对应的层数不同，每个海底模型的每一层中的地声参数为待反演参数，所述地声参数包括：密度、横波声速、纵波声速、横波衰减、纵波衰减和海底厚度；
针对每个海底模型，分别获取每个地声参数对应的预设变化范围，基于所述每 个地声参数对应的预设变化范围随机生成各个所述地声参数的值，基于所述各个所述地声参数的值计算得到理论声压值；
获取实际测量得到的实际声压值；
将所述理论声压值和实际声压值进行比较，根据比较结果调整更新所述各个地声参数的值，返回执行所述基于所述各个所述地声参数的值计算得到理论声压值的步骤，直到得到的理论声压值和实际声压值相匹配，将相匹配的理论声压值对应的各个所述地声参数的值作为所述待反演参数对应的目标地声参数值；
根据所述每个海底模型对应的目标地声参数值采用贝叶斯理论计算得到每个海底模型对应的BIC值；
将BIC值最小的海底模型作为目标海底模型，将所述目标海底模型对应的目标地声参数值作为目标反演参数值。
一种浅海多层海底地声参数反演装置，包括：
建立模块，用于建立多个海底模型，不同海底模型对应的层数不同，每个海底模型的每一层中的地声参数为待反演参数，所述地声参数包括：密度、横波声速、纵波声速、横波衰减、纵波衰减和海底厚度；
生成模块，用于针对每个海底模型，分别获取每个地声参数对应的预设变化范围，基于所述每个地声参数对应的预设变化范围随机生成各个所述地声参数的值，基于所述各个所述地声参数的值计算得到理论声压值；
获取模块，用于获取实际测量得到的实际声压值；
更新模块，用于将所述理论声压值和实际声压值进行比较，根据比较结果调整更新所述各个地声参数的值，返回执行所述基于所述各个所述地声参数的值计算得到理论声压值的步骤，直到得到的理论声压值和实际声压值相匹配，将相匹配的理论声压值对应的各个所述地声参数的值作为所述待反演参数对应的目标地声参数值；
计算模块，用于根据所述每个海底模型对应的目标地声参数值采用贝叶斯理论计算得到每个海底模型对应的BIC值；
确定模块，将BIC值最小的海底模型作为目标海底模型，将所述目标海底模型对应的目标地声参数值作为目标反演参数值。
一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述计算机程序被所述处理器执行时，使得所述处理器执行以下步骤：
建立多个海底模型，不同海底模型对应的层数不同，每个海底模型的每一层中的地声参数为待反演参数，所述地声参数包括：密度、横波声速、纵波声 速、横波衰减、纵波衰减和海底厚度；
针对每个海底模型，分别获取每个地声参数对应的预设变化范围，基于所述每个地声参数对应的预设变化范围随机生成各个所述地声参数的值，基于所述各个所述地声参数的值计算得到理论声压值；
获取实际测量得到的实际声压值；
将所述理论声压值和实际声压值进行比较，根据比较结果调整更新所述各个地声参数的值，返回执行所述基于所述各个所述地声参数的值计算得到理论声压值的步骤，直到得到的理论声压值和实际声压值相匹配，将相匹配的理论声压值对应的各个所述地声参数的值作为所述待反演参数对应的目标地声参数值；
根据所述每个海底模型对应的目标地声参数值采用贝叶斯理论计算得到每个海底模型对应的BIC值；
将BIC值最小的海底模型作为目标海底模型，将所述目标海底模型对应的目标地声参数值作为目标反演参数值。
一种计算机可读存储介质，存储有计算机程序，所述计算机程序被处理器执行时，使得所述处理器执行以下步骤：
建立多个海底模型，不同海底模型对应的层数不同，每个海底模型的每一层中的地声参数为待反演参数，所述地声参数包括：密度、横波声速、纵波声速、横波衰减、纵波衰减和海底厚度；
针对每个海底模型，分别获取每个地声参数对应的预设变化范围，基于所述每个地声参数对应的预设变化范围随机生成各个所述地声参数的值，基于所述各个所述地声参数的值计算得到理论声压值；
获取实际测量得到的实际声压值；
将所述理论声压值和实际声压值进行比较，根据比较结果调整更新所述各个地声参数的值，返回执行所述基于所述各个所述地声参数的值计算得到理论声压值的步骤，直到得到的理论声压值和实际声压值相匹配，将相匹配的理论声压值对应的各个所述地声参数的值作为所述待反演参数对应的目标地声参数值；
根据所述每个海底模型对应的目标地声参数值采用贝叶斯理论计算得到每个海底模型对应的BIC值；
将BIC值最小的海底模型作为目标海底模型，将所述目标海底模型对应的目标地声参数值作为目标反演参数值。
上述浅海多层海底地声参数反演方法、装置、计算机设备及存储介质，首 先，建立多个海底模型，不同海底模型对应不同的层数，然后再针对每个海底模型来随机生成各个地声参数的值，并基于各个地声参数的值计算得到理论声压值，通过将理论声压值与实际声压值进行比较，来确定与实际声压值匹配的理论声压值，并进而确定每个海底模型对应的目标地声参数值，最后采用贝叶斯理论计算得到每个海底模型的BIC值，将BIC值最小的海底模型作为目标海底模型。上述过程中，通过将计算得到的理论声压值和实际声压值进行比较来反演得到目标地声参数值，并且针对每个海底模型采用贝叶斯理论计算得到BIC值，根据BIC值确定了最优的海底模型结构，该方法不仅实现了高效、准确地通过反演得到了目标海底模型中的地声参数值，而且确定了最优的海底模型层数。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
其中：
图1是一个实施例中浅海多层海底地声参数反演方法的流程图；
图2是一个实施例中多层海底参数化模型图；
图3是一个实施例中浅海多层海底地声参数反演装置的结构框图；
图4是一个实施例中计算机设备的内部结构图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
如图1所示，提出了一种浅海多层海底地声参数反演方法，该浅海多层海底地声参数反演方法可以应用于终端，本实施例以应用于终端举例说明。该浅海多层海底地声参数反演方法具体包括以下步骤：
步骤102，建立多个海底模型，不同海底模型对应的层数不同，每个海底模型的每一层中的地声参数为待反演参数，地声参数包括：密度、横波声速、纵波声速、横波衰减、纵波衰减和海底厚度。
其中，建立多个不同的海底模型，是为了后续找到与实际情况相符的最佳的海底模型。不同海底模型的层数不同，即不同的海底模型具有不同的海底结构。海底模型是基于波动理论建立的，建立的海底模型是采用方程来表示的， 建立的海底模型的方程中涉及到了地声参数和声压，即地声参数和声压是海底模型中的参数。为了得到更加准确的海底模型，本申请中的地声参数比较全面，每一层中同时考虑了密度、横波声速、纵波声速、横波衰减、纵波衰减和海底厚度等多个因素。
不同层数的海底模型是不同的，其计算过程也是不同的，海底层数越多需要反演的参数越多，模型是水平分成的，每增加一层海底，在计算过程中将会增加4个方程。如图2所示，为多层海底参数化模型图，每一层中都包含有相应的地声参数，图中，c p、c s、ρ b、α p和α s分别代表纵波声速、横波声速、海底密度、纵波声速衰减和横波声速衰减，z为水深，z s为声源水深，r为传播距离，f 0为声源频率，下标分别代表所在层数。
步骤104，针对每个海底模型，分别获取每个地声参数对应的预设变化范围，基于每个地声参数对应的预设变化范围随机生成各个地声参数的值，基于各个地声参数的值计算得到理论声压值。
其中，每个地声参数的预设变化范围是指预先设置的该地声参数的值的变化范围。由于地声参数和声压都是海底模型方程中的未知参数，且地声参数的个数很多，所以该方程是无法直接求解的。这里为了进行反演出地声参数的值，是通过给各个地声参数进行赋值，不断地进行寻优的过程。赋值的方式是每个地声参数在预设变化范围内随机生成相应的地声参数的值，然后基于各个地声参数的值就可以计算得到声压值，即理论声压值。各个地声参数的变化范围可以根据实际情况自定义设定，在一个实施例中，预设变化范围设定如下：密度g·cm -3(1-2)，纵波声速m/s(1800-2000)，横波声速(900-1100),纵波衰减dB/λ(0.09-0.11),横波衰减dB/λ(0.09/0.11)，厚度几十米不等(15-25)。然后基于预设范围生成地声参数的值，比如，密度：1.5、纵波声速1950、横波声速900、纵波衰减0.095、横波衰减0.096、海底厚度18。
步骤106，获取实际测量得到的实际声压值。
其中，实际声压值是可以测量得到的，测量的方式可以采用水听器来测量，通过水听器监听声源发出的声波，然后对监听的声波进行处理得到实际声压值，在一个实施例中，水听器测量到的结果为wav格式的音频，导入到matlab中转换为数值的形式，然后经过傅里叶变换得到这组数据的频谱，去幅值即为声压值。一般得到的实际声压值是一组声压值组成的，比如，一组声压值中包含有1000个数值。
步骤108，将理论声压值和实际声压值进行比较，根据比较结果调整更新各个地声参数的值，返回执行基于各个地声参数的值计算得到理论声压值的步骤，直到得到的理论声压值和实际声压值相匹配，将相匹配的理论声压值对应 的各个地声参数的值作为待反演参数对应的目标地声参数值。
其中，通过采用误差函数计算理论声压值和实际声压值之间的误差值，当理论声压值和实际声压值不匹配时，需要在各个地声参数的预设范围内更新各个地声参数的值，再次进行理论声压值的计算，然后再进行比较，就这样通过多次迭代计算，直到计算得到的理论声压值和实际声压值相匹配。理论声压值和实际声压值相匹配的判断条件可以有多种形式，一种是预先设置最小误差值，当两者的误差中小于最小误差值时判断两者匹配。一种是经过多次迭代后，误差值达到了收敛，然后停止迭代，将最后得到的理论声压值作为和实际声压值相匹配的声压值，然后将相匹配的理论声压值对应的各个地声参数的值作为待反演参数对应的目标地声参数值。目标地声参数值即为经过反演得到的地声参数的值。
上述针对多层海底模型，采用给各个地声参数赋值，计算得到理论声压值，将理论声压值和实际声压值进行比较来实现对各个地声参数的反演。该多层海底模型中采用比较理论声压值和实际声压值的方法来进行各个地声参数的反演，从而实现了高效、准确地确定各个地声参数的值。
步骤110，根据每个海底模型对应的目标地声参数值采用贝叶斯理论计算得到每个海底模型对应的BIC值。
其中，BIC(Bayesian Information Criterion，贝叶斯信息准则)的引入是为了判别实际声压所处的海洋环境是如何分层的，因为我们测到的实际声压数据就是一组数，针对不同的海底模型都是可以计算得到对应的目标地声参数的，但是到底哪种海底模型才是最佳的呢，这里创新性地基于BIC准则计算得到每个海底模型的BIC值，通过比较BIC值确定哪个海底模型反演得到的目标地声参数是最佳的。
步骤112，将BIC值最小的海底模型作为目标海底模型，将目标海底模型对应的目标地声参数值作为目标反演参数值。
其中，BIC值越小，说明该海底模型越接近真实海底环境。将得到的目标海底模型对应的目标地声参数值作为地声参数的目标反演参数值(即目标反演结果)。
上述浅海多层海底地声参数反演方法，首先，建立多个海底模型，不同海底模型对应不同的层数，然后再针对每个海底模型来随机生成各个地声参数的值，并基于各个地声参数的值计算得到理论声压值，通过将理论声压值与实际声压值进行比较，来确定与实际声压值匹配的理论声压值，并进而确定每个海底模型对应的目标地声参数值，最后采用贝叶斯理论计算得到每个海底模型的BIC值，将BIC值最小的海底模型作为目标海底模型。上述过程中，通过将计算得到的理论声压值和实际声压值进行比较来反演得到目标地声参数值，并且 针对每个海底模型采用贝叶斯理论计算得到BIC值，根据BIC值确定了最优的海底模型结构，该方法不仅实现了高效、准确地通过反演得到了目标海底模型中的地声参数值，而且确定了最优的海底模型层数。
在一个实施例中，所述建立多个海底模型，不同海底模型对应的层数不同，包括：根据波动理论，构建每个海底模型中每层对应的位移势函数方程；根据所述位移势函数方程计算得到每个位移势函数的通解，所述每个位移势函数的通解中包含有多个不确定系数，所述多个不确定系数与所述地声参数相关，所述理论声压值是根据位移势函数计算得到的。
其中，在柱坐标下各层海底物理量通过位移势函数表示，各层位移势函数建立满足波动方程方程组，在声场条件下结合点源条件和流体/弹性体分界面处的边界条件，可具体表示出各层位移势函数。然后依据快速场方法(Fast Field Method，FFM)，实际上是求解出方程组的各个系数，从而得到个各层位移势函数。再利用流体层中声压p与势函数φ 1之间的关系：p＝ρ 1ω 2φ 1，即可获得流体层中各点声压值。
海底模型建立以及声压计算公式推导如下：根据波动理论，在频域中，多层海底模型中各层位移势函数满足如下方程：
Figure PCTCN2021090789-appb-000001
Figure PCTCN2021090789-appb-000002
Figure PCTCN2021090789-appb-000003
其中，δ(r,z)表示声源方程，k mn＝ω/c m(m＝1,p,sn＝1…N)是每层的波数，ω＝2πf 0是点源在f 0处的角频率，r表示信号传播距离；z表示垂直深度；▽表示拉普拉斯算子；φ 1表示流体层中的位移势函数；φ p为弹性海底中位标量移势函数，ψ s表示矢量位移势函数；k 1＝ω/c 1表示流体层中的波数，k p＝ω/c p'表示弹性海底中的纵波波数(c p'为加入声速衰减后声速值，其中：
Figure PCTCN2021090789-appb-000004
)；k s＝ω/c s'表示弹性海底中的横波波数；ω＝2πf 0表示声源频率f 0所对应的角频率。海底模型的建立是基于波动理论建立的，具有可靠性。
在一个实施例中，所述获取实际测量得到的实际声压值，包括：采用水听器监听声源发出的声波，所述声波是通过发射换能器在水中发射产生的，所述水听器和所述发射换能器是通过相对移动来对完成测量的；将水听器检测到的 wav格式的音频导入到matlab中转换为一组数值；采用傅里叶变换对所述一组数值进行处理，得到所述一组数值对应的频谱；计算所述频谱的幅值得到所述实际声压值，所述实际声压值包括多个位置的声压值。
其中，wav是一种声音文件格式。matlab是一种数学软件，用于数据分析等。
实际海洋环境，一般是低频信号，这样传播距离会更远一些，达到携带更多海底信息的目的，一般在100HZ左右，发射位置水下几米或者几十米都可。这个方法在海上实际使用的时候，一般是用船搭载水听器或者声源进行移动。既声源位置固定，实验船携带水听器移动完成测量；或者水听器固定，实验船携带声源移动完成测量；发射换能器就是实验中的声源设备，通过发射换能器才能将声波向水中发射，在理论写作、分析的时候会用声源来描述，在实验中的发声设备就是发射换能器。实际实验的时候，发射换能器和水听器一般都是电源端固定在船上，发射和接收端通过缆绳下放到水里，深度根据实验设计决定。将测到的wav格式的音频导入到matlab中转换为数值的形式，经过傅里叶变换得到这组数据的频谱，取幅值即为声压值。在测量的时候不会只测一个位置的声压，而是通过船搭载水听器或者声源进行移动而测量不同位置的声压，从而得到的是一组声压值。上述实际声压值的测量是基于实际海洋环境进行测量的，且测量得到的是一组声压值，可靠准确。
以实验室消声水池实验为例说明。采用聚氯乙烯材料的板子模拟海底，高频水声由固定位置的声源发射，接收水听器每隔固定距离测量一次，发射换能器固定在一端水中，接收水听器固定在移动微型工作台，工作台每次移动2mm，测量误差小于20um。利用计算机对移动工作台进行控制，测量并获取数据。当完成在一个位置测量后，工作台自动移动到下一个位置，共测量1000个点。在一个实施例中，所述将所述理论声压值和实际声压值进行比较，根据比较结果调整更新所述各个地声参数的值，包括：采用误差函数计算理论声压值和实际声压值之间的误差值，所述误差函数的公式如下：
Figure PCTCN2021090789-appb-000005
其中，
Figure PCTCN2021090789-appb-000006
表示理论声压值，
Figure PCTCN2021090789-appb-000007
表示实际声压值，m表示海底模型参数；*表示共轭转置，f表示频点数，F为采用的频点总数，K表示实验中水听器的个数。当所述误差值大于预设误差值时，更新调整所述各个地声参数的值。
其中，理论声压值和实际声压值之间的误差值采用误差函数计算得到，误差函数是贝叶斯理论设计的，在贝叶斯理论下结合似然函数建立理论声压和实际声压的关系的误差函数，在该理论下误差函数达到最小值时表明理论声压和实际声压相似度达到最大，即此时的理论声压等于实际声压。该误差函数可以准确地反映出理论声压值和实际声压值的差异，从而有利于更好地匹配得到与实际声压值匹配的理论声压值。
在一个实施例中，所述针对每个海底模型，分别获取每个地声参数对应的预设变化范围，基于所述每个地声参数对应的预设变化范围随机生成各个所述地声参数的值，基于所述各个所述地声参数的值计算得到理论声压值，包括：获取各个地声参数的初始值，所述初始值是基于所述预设变化范围随机生成的；基于所述各个地声参数的初始值和所述预设变化范围采用改进的模拟退火法进行扰动生成新的各个所述地声参数的值；根据各个新的所述地声参数的值计算得到相应的新的理论声压值；
所述将所述理论声压值和实际声压值进行比较，根据比较结果调整更新所述各个地声参数的值，返回执行所述基于所述各个所述地声参数的值计算得到理论声压值的步骤，直到得到的理论声压值和实际声压值相匹配，包括：根据新的所述理论声压值和所述实际声压值计算得到新的误差值，将所述新的误差值与之前的误差值进行比较，保留较小的误差值以及对应的地声参数，返回执行所述基于所述各个地声参数的初始值和所述预设变化范围进行扰动生成新的各个所述地声参数值的步骤，直到达到收敛条件，将最后保留的各个地声参数值对应的理论声压值作为与实际声压值匹配的声压值。
其中，确定与实际声压值匹配的理论声压值的过程就是地声参数的反演过程。首先，设置各个地声参数的预设变化范围，然后初始化地声参数，初始化地声参数的过程是在预设变化范围内随机生成各个地声参数的初始值。然后将各个初始值带入到海底模型计算得到理论声压值，将理论声压值和实际声压值带入误差函数得到误差值。误差值用来衡量理论声压值和实际声压值之间的差异，误差值越小，表明理论声压值和实际声压值越接近。然后采用扰动算法以初始值为中心通过扰动在预设范围内生成新值，得到新的各个所述地声参数的值，然后计算得到新的理论声压值，将新的理论声压值与实际声压值通过误差函数计算得到新的误差中，将新的误差值和初始误差值进行比较，保留较小的误差值和相应的地声参数值。之后循环执行上述采用扰动算法以初始值为中心通过扰动在预设范围内生成新值，得到新的各个所述地声参数的值，并计算得到新的误差值，将新的误差值和保留的误差值进行比较，然后保留误差值较小的理论声压值以及对应的地声参数，直到达到收敛，将最后保留的理论声压值 对应的地声参数的值作为反演得到的值。上述给各个地声参数赋值的过程中，为每个地声参数都设置有预设变化范围，从而保证了随机生成的地声参数不会偏离实际，且又可以保证随机性，从而实现了准确地确定目标地声参数的值。
在一个实施例中，所述基于所述各个地声参数的初始值和所述预设变化范围采用改进的模拟退火法进行扰动生成新的各个所述地声参数的值，包括：获取当当前迭代次数，根据当前迭代次数确定扰动系数；获取扰动条件，所述扰动条件为多层海底模型中下层海底参数大于上层海底参数；根据所述预设变化范围、所述扰动系数和所述扰动条件随机生成新的各个所述地声参数的值。
其中，迭代次数决定了随机扰动的幅度，迭代次数和扰度幅度成反相关，迭代次数和模拟退火温度成反相关，模拟退火温度越低，迭代次数越高，相应的扰动幅度也越小。扰动条件为多层海底模型中下层海底参数大于上层海底参数。通过设置扰动条件有效地遵循了一般情况下多沉积层海底声阻抗随深度增加而增大的客观规律。通过设置扰动条件，保证了多层海底模型中各个地声参数能够遵循客观规律，从而有利于生成准确的地声参数的值。
在一个实施例中，扰动过程的计算如下：
第一步：为待反演的参数设置预设变化范围(即上下边界)，在算法执行扰动后的结果均保持在该范围中，超出该范围的参数值将通过越界函数剔除掉，设置初始温度Tmax，终止温度Tmin(即设置外循环终止条件)和马尔科夫链(Markov)的长度L，用于表示初始设置的种群数，就是研究多少组数，例如，纵波声速设置种群数为1000，即每次扰动寻优都是1000个纵波声速。
第二步：为每个参数随机产生初始值，其中m 0表示待反演参数的初始值，S min表示每个参数区间的下边界；S L表示参数区间宽度，即上边界减去下边界；rand(0,1)为matlab函数，能够产生0-1之间的随机数。
m 0＝S min+S L·rand(0,1)
第三步：将生成的初始值代入海底模型计算出该组参数对应的误差值并保留E(m i)。
第四步：在初始值的基础上通过扰动生成新解，引入函数randi()，令R＝randi([0,1])，使得R值非0即1。
当R＝0时，在初始值的基础上左移，即
m new＝m now+(S max-m now)·a
其中m new表示扰动后的新解，m now表示当前解(第一次循环时为初始解)，S max表示参数区间上边界，a表示扰动系数。其中，a＝(1-rand(0,1)^(1-(t/T)^b)),其中t表示当前迭代次数，T表示预设总迭代次数，b控制搜寻步长，b的经验取值一般为2。随着温度的逐渐降低，t值在不断增大，即a值在温度较高时 保持较大的值，温度较小时保持较小值，能够使得在初始搜索时保证较大的扰动量，随着温度的降低搜索区间逐渐减小直至最后算法收敛。
反之右移，即
m new＝m now-(m now–S min)·a
第五步：将扰动后的新解代入海底模型计算出新的误差值E(m i)，并与上一个误差值作差得到△E＝E(m i+1)-E(m i)。判断△E值，若△E<0，则接受新解，若△E>0，则根据Metropolis准则(以概率接受新状态)接受新解，若都不满足则不接受新解，保留原始参数解，用于下次进行误差值的比较。
第六步：判断是否满足内循环终止条件(误差值是否达到收敛)，若不满足则返回步骤四，若满足则判断是否满足外循环终止条件(温度是否小于Tmin)，若不满足执行降温，若满足则终止计算，输出结果。
举个例子：将这些参数带入正演海底模型(即计算理论声压)得到一组声压，这组声压是和实际声压维度相同的一组数，比如实测得到的是一组1000个点的声压值，那理论计算出的声压值也是1000个点的声压值，将理论声压和实际声压带入到误差函数中得到误差值，比如是-5，然后经过扰动之后再给出一组值，1.5、2000、1000、0.01、0.01、20.再次将这组参数带入正演海底模型计算理论声压，将得到的理论声压和不变的实际声压带入误差函数中计算出误差值比如-6，因为-6更小，所以1.5、2000、1000、0.01、0.01、20。这组参数更优，然后保存这组参数以及误差值，下边的计算重复上述步骤，而下一个误差值和保存的这个误差值-6进行比较，如果-7那么保存新的一组参数和误差值，如果-3了，那么还是保留刚才-6那组参数，就这样一直须有直到误差值不变，收敛停止计算，保留最后一代参数作为反演的解。
在一个实施例中，所述根据所述每个海底模型对应的目标地声参数值采用贝叶斯理论计算得到每个海底模型对应的BIC值，包括：根据所述每个海底模型对应的目标地声参数值和误差值采用改进的贝叶斯理论计算得到每个海底模型对应的BIC值，所述BIC值的计算采用如下公式实现：
Figure PCTCN2021090789-appb-000008
其中，M是模型中参数的个数，N为数据个数，
Figure PCTCN2021090789-appb-000009
表示根据误差函数计算得到的误差值。
其中，BIC值的计算公式是经过推导得到的，该BIC值的大小由误差函数、模型参数个数和数据个数共同决定，从而避免了欠参数化和过参数化模型，更有效的选择出最优海底模型。
在一个实施例中，贝叶斯理论、误差函数以及BIC公式推导如下：
随机变量d和m分别表示缩比实验中提取的实验数据和海底模型参数， N和M分别表示向量d和向量m的个数。向量d和m满足贝叶斯定理：
P(m|d)＝P(d|m)P(m)/P(d)   (10)
其中，P(m|d)为后验概率密度(PPD)，d的条件概率P(d|m)通常用似然函数L(m)来表示，P(m)是m的先验概率密度函数，表示独立于数据的可用模型参数先验信息，P(d)是参数d的概率密度函数。由于P(d)与参数m无关，可以看作一个常数，上式可改为:
P(m|d)∝L(m)P(m)   (11)
似然函数由数据形式和数据误差的统计分布决定。考虑到在实际应用过程中，误差的统计特征很难独立获得，在处理过程中采用无偏高斯误差的假设，似然函数的形式为:
L(m)＝P(d|m)∝exp[-E(m)]   (12)
其中E(m)为误差函数，归一化后可得
Figure PCTCN2021090789-appb-000010
其中，积分域跨越M维参数空间，M为待反演参数的个数。在贝叶斯理论中，后验概率密度(PPD)可作为反演问题的解。由于反演中存在对维参数问题，为更合理的解释参数反演结果还需对模型参数间的相关特性进行研究，例如：参数的MAP值、均值、和一维概率密度分布，分别定义为:
Figure PCTCN2021090789-appb-000011
Figure PCTCN2021090789-appb-000012
P(m i|d)＝∫δ(m i-m i′)P(m′|d)dm′   (16)
在贝叶斯反演理论中，求解参数PPD需要获得似然函数L(m)，似然函数与数据误差(包括测量误差和理论误差)的统计分布有关，是定量描述参数不确定性的重要指标。本文假设数据误差是独立同分布的随机变量，则似然函数可以表示为：
Figure PCTCN2021090789-appb-000013
其中，p f mea表示在频率为f下单个传感器在位置k接收到的测量声压，在相同情况下，p f pre和C f m分别表示模型预测声压和协方差矩阵。
预测声压p f pre可通过下式表达
Figure PCTCN2021090789-appb-000014
其中，p f FFM表示通过快速场方法(FFM)计算的声压,A f和θ f为未知声源在每一个频率上的幅度和相位。令
Figure PCTCN2021090789-appb-000015
可得到声源的
最大似然估计值为：
Figure PCTCN2021090789-appb-000016
其中*表示共轭转置。忽略数据的空间相关性，对角线协方差近似处理为C f m＝v fI，其中方差v f只与频率有关，I为单位阵，此时似然函数可以简化为:
Figure PCTCN2021090789-appb-000017
其中B f(m)表示归一化Bartlett失配器。
Figure PCTCN2021090789-appb-000018
令
Figure PCTCN2021090789-appb-000019
得到方差v f的最大似然估计值，即：
Figure PCTCN2021090789-appb-000020
将公式(22)带入公式(12)和公式(20)得到满足最大似然函数估计值时对应的误差函数E(m)
Figure PCTCN2021090789-appb-000021
合理参数化模型是贝叶斯反演的关键，欠参数化模型使得结构无法完整解析，导致模型的不确定性偏低；过参数化模型对参数的约束不够，导致模型的不确定性增加，欠参数化和过参数化模型都会对反演结果造成一定的影响。本文应用贝叶斯信息准则(BIC)选择与实测数据最符合的参数化模型。BIC值是从多维变量的正态分布中得到的，并不是一个精确值，是模型I的贝叶斯定理P(d|I)的渐进近似，也就是假定测量数据d，模型I的似然函数，其表达式为：
Figure PCTCN2021090789-appb-000022
其中，M是模型I中参数的个数，N为数据参数个数，用误差函数代替似然函数可以得到：
Figure PCTCN2021090789-appb-000023
BIC值最小的模型即为最优模型。从公式(25)可以看出，BIC数值的大小由误差函数、模型参数个数和数据个数共同决定，从而避免了欠参数化和过参数化模型，更有效的选择出最优海底模型。
如图3所示，一种浅海多层海底地声参数反演装置，包括：
建立模块302，用于建立多个海底模型，不同海底模型对应的层数不同， 每个海底模型的每一层中的地声参数为待反演参数，所述地声参数包括：密度、横波声速、纵波声速、横波衰减、纵波衰减和海底厚度；
生成模块304，用于针对每个海底模型，分别获取每个地声参数对应的预设变化范围，基于所述每个地声参数对应的预设变化范围随机生成各个所述地声参数的值，基于所述各个所述地声参数的值计算得到理论声压值；
获取模块306，用于获取实际测量得到的实际声压值；
更新模块308，用于将所述理论声压值和实际声压值进行比较，根据比较结果调整更新所述各个地声参数的值，返回执行所述基于所述各个所述地声参数的值计算得到理论声压值的步骤，直到得到的理论声压值和实际声压值相匹配，将相匹配的理论声压值对应的各个所述地声参数的值作为所述待反演参数对应的目标地声参数值；
计算模块310，用于根据所述每个海底模型对应的目标地声参数值采用贝叶斯理论计算得到每个海底模型对应的BIC值；
确定模块312，将BIC值最小的海底模型作为目标海底模型，将所述目标海底模型对应的目标地声参数值作为目标反演参数值。
在一个实施例中，建立模块302还用于根据波动理论，构建每个海底模型中每层对应的位移势函数方程；根据所述位移势函数方程计算得到每个位移势函数的通解，所述每个位移势函数的通解中包含有多个不确定系数，所述多个不确定系数与所述地声参数相关，所述理论声压值是根据位移势函数计算得到的。
在一个实施例中，获取模块306还用于采用水听器监听声源发出的声波，所述声波是通过发射换能器在水中发射产生的，所述水听器和所述发射换能器是通过相对移动来对完成测量的；将水听器检测到的wav格式的音频导入到matlab中转换为一组数值；采用傅里叶变换对所述一组数值进行处理，得到所述一组数值对应的频谱；计算所述频谱的幅值得到所述实际声压值，所述实际声压值包括多个位置的声压值。
在一个实施例中，更新模块308还用于采用误差函数计算理论声压值和实际声压值之间的误差值，所述误差函数的公式如下：
Figure PCTCN2021090789-appb-000024
其中，
Figure PCTCN2021090789-appb-000025
表示理论声压值，
Figure PCTCN2021090789-appb-000026
表示实际声压值，m表示海底模型参数；*表示共轭转置，F表示包含的声压值的个数。当所述误差值大于预设误差值时，更新调整所述各个地声参数的值。
在一个实施例中，生成模块304还用于获取各个地声参数的初始值，所述初始值是基于所述预设变化范围随机生成的；基于所述各个地声参数的初始值和所述预设变化范围采用改进的模拟退火法进行扰动生成新的各个所述地声参数的值；根据各个新的所述地声参数的值计算得到相应的新的理论声压值；更新模块308还用于根据新的所述理论声压值和所述实际声压值计算得到新的误差值，将所述新的误差值与之前的误差值进行比较，保留较小的误差值以及对应的地声参数，返回执行所述基于所述各个地声参数的初始值和所述预设变化范围进行扰动生成新的各个所述地声参数值的步骤，直到达到收敛条件，将最后保留的各个地声参数值对应的理论声压值作为与实际声压值匹配的声压值。
在一个实施例中，生成模块304还用于获取当前退火温度，根据所述当前退火温度确定扰动系数；根据所述扰动系数确定扰动幅度；获取扰动条件，所述扰动条件为多层海底模型中下层海底参数大于上层海底参数；根据所述预设变化范围、所述扰动幅度和所述扰动条件随机生成新的各个所述地声参数的值。
在一个实施例中，计算模块310还用于根据所述每个海底模型对应的目标地声参数值和误差值采用改进的贝叶斯理论计算得到每个海底模型对应的BIC值，所述BIC值的计算采用如下公式实现：
Figure PCTCN2021090789-appb-000027
其中，M是模型中参数的个数，N为数据个数，
Figure PCTCN2021090789-appb-000028
表示根据误差函数计算得到的误差值。
图4示出了一个实施例中计算机设备的内部结构图。该计算机设备具体可以是终端，也可以是服务器。如图4所示，该计算机设备包括通过系统总线连接的处理器、存储器和网络接口。其中，存储器包括非易失性存储介质和内存储器。该计算机设备的非易失性存储介质存储有操作系统，还可存储有计算机程序，该计算机程序被处理器执行时，可使得处理器实现上述的浅海多层海底地声参数反演方法。该内存储器中也可储存有计算机程序，该计算机程序被处理器执行时，可使得处理器执行上述的浅海多层海底地声参数反演方法。本领域技术人员可以理解，图4中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的计算机设备的限定，具体的计算机设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
在一个实施例中，提出了一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述计算机程序被所述处理器执行时，使得所述处理器执行上述浅海多层海底地声参数反演方法的步骤。
在一个实施例中，提出了一种计算机可读存储介质，存储有计算机程序，所述计算机程序被处理器执行时，使得所述处理器执行上述浅海多层海底地声参数反演方法的步骤。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的程序可存储于一非易失性计算机可读取存储介质中，该程序在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限，RAM以多种形式可得，诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。
以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对本申请专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。

Claims (10)

  1. 一种浅海多层海底地声参数反演方法，其特征在于，所述方法包括：建立多个海底模型，不同海底模型对应的层数不同，每个海底模型中的每一层中的地声参数为待反演参数，所述每一层地声参数包括：密度、横波声速、纵波声速、横波衰减、纵波衰减和海底厚度；
    针对每个海底模型，分别获取每个地声参数对应的预设变化范围，基于所述每个地声参数对应的预设变化范围随机生成各个所述地声参数的值，基于所述各个所述地声参数的值计算得到理论声压值；
    获取实际测量得到的实际声压值；
    将所述理论声压值和实际声压值进行比较，根据比较结果调整更新所述各个地声参数的值，返回执行所述基于所述各个所述地声参数的值计算得到理论声压值的步骤，直到得到的理论声压值和实际声压值相匹配，将相匹配的理论声压值对应的各个所述地声参数的值作为所述待反演参数对应的目标地声参数值；
    根据所述每个海底模型对应的目标地声参数值采用贝叶斯理论计算得到每个海底模型对应的BIC值；
    将BIC值最小的海底模型作为目标海底模型，将所述目标海底模型对应的目标地声参数值作为目标反演参数值。
  2. 根据权利要求1所述的方法，其特征在于，所述建立多个海底模型，不同海底模型对应的层数不同，包括：
    根据波动理论，构建每个海底模型中每层对应的位移势函数方程；
    根据所述位移势函数方程计算得到每个位移势函数的通解，所述每个位移势函数的通解中包含有多个不确定系数，所述多个不确定系数与所述地声参数相关，所述理论声压值是根据位移势函数计算得到的。
  3. 根据权利要求1所述的方法，其特征在于，所述获取实际测量得到的实际声压值，包括：
    采用水听器监听声源发出的声波，所述声波是通过发射换能器在水中发射产生的，所述水听器和所述发射换能器是通过相对移动来对完成测量的；
    将水听器检测到的wav格式的音频导入到matlab中转换为一组数值；
    采用傅里叶变换对所述一组数值进行处理，得到所述一组数值对应的频谱；
    计算所述频谱的幅值得到所述实际声压值，所述实际声压值包括多个位置的声压值。
  4. 根据权利要求1所述的方法，其特征在于，所述将所述理论声压值和实际声压值进行比较，根据比较结果调整更新所述各个地声参数的值，包括：
    采用误差函数计算理论声压值和实际声压值之间的误差值，所述误差函数的公式如下：
    Figure PCTCN2021090789-appb-100001
    其中，
    Figure PCTCN2021090789-appb-100002
    Figure PCTCN2021090789-appb-100003
    表示理论声压值，
    Figure PCTCN2021090789-appb-100004
    表示实际声压值，m表示海底模型参数；*表示共轭转置，f表示频点数，F为采用的频点总数，K表示水听器的个数；
    当所述误差值大于预设误差值时，更新调整所述各个地声参数的值。
  5. 根据权利要求1所述的方法，其特征在于，所述针对每个海底模型，分别获取每个地声参数对应的预设变化范围，基于所述每个地声参数对应的预设变化范围随机生成各个所述地声参数的值，基于所述各个所述地声参数的值计算得到理论声压值，包括：
    获取各个地声参数的初始值，所述初始值是基于所述预设变化范围随机生成的；
    基于所述各个地声参数的初始值和所述预设变化范围采用改进的模拟退火法进行扰动生成新的各个所述地声参数的值；
    根据各个新的所述地声参数的值计算得到相应的新的理论声压值；
    所述将所述理论声压值和实际声压值进行比较，根据比较结果调整更新所述各 个地声参数的值，返回执行所述基于所述各个所述地声参数的值计算得到理论声压值的步骤，直到得到的理论声压值和实际声压值相匹配，包括：
    根据新的所述理论声压值和所述实际声压值计算得到新的误差值，将所述新的误差值与之前的误差值进行比较，保留较小的误差值以及对应的地声参数，返回执行所述基于所述各个地声参数的初始值和所述预设变化范围进行扰动生成新的各个所述地声参数值的步骤，直到达到收敛条件，将最后保留的各个地声参数值对应的理论声压值作为与实际声压值匹配的声压值。
  6. 根据权利要求1所述的方法，其特征在于，所述基于所述各个地声参数的初始值和所述预设变化范围采用改进的模拟退火法进行扰动生成新的各个所述地声参数的值，包括：
    获取当前迭代次数，根据所述当前迭代次数确定扰动系数；
    获取扰动条件，所述扰动条件为多层海底模型中下层海底参数大于上层海底参数；
    根据所述预设变化范围、所述扰动系数和所述扰动条件随机生成新的各个所述地声参数的值。
  7. 根据权利要求1所述的方法，其特征在于，所述根据所述每个海底模型对应的目标地声参数值采用贝叶斯理论计算得到每个海底模型对应的BIC值，包括：
    根据所述每个海底模型对应的目标地声参数值和误差值采用贝叶斯理论计算得到每个海底模型对应的BIC值，所述BIC值的计算采用如下公式实现：
    Figure PCTCN2021090789-appb-100005
    其中，M是模型中参数的个数，N为数据个数，
    Figure PCTCN2021090789-appb-100006
    表示根据误差函数计算得到的误差值。
  8. 一种浅海多层海底地声参数反演装置，其特征在于，包括：
    建立模块，用于建立多个海底模型，不同海底模型对应的层数不同，每个 海底模型的每一层中的地声参数为待反演参数，所述地声参数包括：密度、横波声速、纵波声速、横波衰减、纵波衰减和海底厚度；
    生成模块，用于针对每个海底模型，分别获取每个地声参数对应的预设变化范围，基于所述每个地声参数对应的预设变化范围随机生成各个所述地声参数的值，基于所述各个所述地声参数的值计算得到理论声压值；
    获取模块，用于获取实际测量得到的实际声压值；
    更新模块，用于将所述理论声压值和实际声压值进行比较，根据比较结果调整更新所述各个地声参数的值，返回执行所述基于所述各个所述地声参数的值计算得到理论声压值的步骤，直到得到的理论声压值和实际声压值相匹配，将相匹配的理论声压值对应的各个所述地声参数的值作为所述待反演参数对应的目标地声参数值；
    计算模块，用于根据所述每个海底模型对应的目标地声参数值采用贝叶斯理论计算得到每个海底模型对应的BIC值；
    确定模块，将BIC值最小的海底模型作为目标海底模型，将所述目标海底模型对应的目标地声参数值作为目标反演参数值。
  9. 一种计算机可读存储介质，存储有计算机程序，所述计算机程序被处理器执行时，使得所述处理器执行如权利要求1至7中任一项所述的浅海多层海底地声参数反演方法的步骤。
  10. 一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述计算机程序被所述处理器执行时，使得所述处理器执行如权利要求1至7中任一项所述的浅海多层海底地声参数反演的步骤。
PCT/CN2021/090789 2021-04-29 2021-04-29 浅海多层海底地声参数反演方法、装置、计算机设备及存储介质 WO2022226856A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
PCT/CN2021/090789 WO2022226856A1 (zh) 2021-04-29 2021-04-29 浅海多层海底地声参数反演方法、装置、计算机设备及存储介质
CN202180000994.5A CN113330439A (zh) 2021-04-29 2021-04-29 浅海多层海底地声参数反演方法、装置、计算机设备及存储介质
US17/588,224 US20220357476A1 (en) 2021-04-29 2022-01-29 Inversion method and apparatus for multilayer seabed geoacoustic parameter in shallow sea, computer device and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/090789 WO2022226856A1 (zh) 2021-04-29 2021-04-29 浅海多层海底地声参数反演方法、装置、计算机设备及存储介质

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US17/588,224 Continuation US20220357476A1 (en) 2021-04-29 2022-01-29 Inversion method and apparatus for multilayer seabed geoacoustic parameter in shallow sea, computer device and storage medium

Publications (1)

Publication Number Publication Date
WO2022226856A1 true WO2022226856A1 (zh) 2022-11-03

Family

ID=77427043

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/090789 WO2022226856A1 (zh) 2021-04-29 2021-04-29 浅海多层海底地声参数反演方法、装置、计算机设备及存储介质

Country Status (3)

Country Link
US (1) US20220357476A1 (zh)
CN (1) CN113330439A (zh)
WO (1) WO2022226856A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115824458A (zh) * 2023-02-22 2023-03-21 自然资源部第一海洋研究所 一种海底底层水温反演的方法和装置
CN116910473A (zh) * 2023-07-03 2023-10-20 中山大学 海底沉积物宽频声速预测方法、计算机装置和存储介质

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116644295B (zh) * 2023-07-03 2023-12-12 中山大学 海底沉积物声衰减宽频预测方法、计算机装置和存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107765310A (zh) * 2017-09-26 2018-03-06 浙江海洋大学 一种针对浅海分层弹性海底的浅海声场计算方法
CN109214025A (zh) * 2017-07-06 2019-01-15 中国石油化工股份有限公司 基于贝叶斯分类的储层参数预测方法及系统
CN109472069A (zh) * 2018-10-26 2019-03-15 浙江海洋大学 一种由贝叶斯理论和声压场反演海底参数的方法
CN111563233A (zh) * 2020-04-12 2020-08-21 西北工业大学 基于非线性贝叶斯理论的水声信号频散特征被动反演方法
CN112163378A (zh) * 2020-10-28 2021-01-01 浙江海洋大学 根据非线性贝叶斯理论的浅海多层海底地声参数反演方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2479172B (en) * 2010-03-31 2016-02-10 Statoil Petroleum As Estimating interval velocities
EP3359982B1 (en) * 2015-10-07 2022-12-14 Services Pétroliers Schlumberger Seismic sensor orientation

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109214025A (zh) * 2017-07-06 2019-01-15 中国石油化工股份有限公司 基于贝叶斯分类的储层参数预测方法及系统
CN107765310A (zh) * 2017-09-26 2018-03-06 浙江海洋大学 一种针对浅海分层弹性海底的浅海声场计算方法
CN109472069A (zh) * 2018-10-26 2019-03-15 浙江海洋大学 一种由贝叶斯理论和声压场反演海底参数的方法
CN111563233A (zh) * 2020-04-12 2020-08-21 西北工业大学 基于非线性贝叶斯理论的水声信号频散特征被动反演方法
CN112163378A (zh) * 2020-10-28 2021-01-01 浙江海洋大学 根据非线性贝叶斯理论的浅海多层海底地声参数反演方法

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115824458A (zh) * 2023-02-22 2023-03-21 自然资源部第一海洋研究所 一种海底底层水温反演的方法和装置
CN116910473A (zh) * 2023-07-03 2023-10-20 中山大学 海底沉积物宽频声速预测方法、计算机装置和存储介质
CN116910473B (zh) * 2023-07-03 2024-01-19 中山大学 海底沉积物宽频声速预测方法、计算机装置和存储介质

Also Published As

Publication number Publication date
US20220357476A1 (en) 2022-11-10
CN113330439A (zh) 2021-08-31

Similar Documents

Publication Publication Date Title
WO2022226856A1 (zh) 浅海多层海底地声参数反演方法、装置、计算机设备及存储介质
CN107885934B (zh) 基于耦合fem-pe的海洋信道下弹性结构声辐射预报方法
CN111123192A (zh) 一种基于圆形阵列和虚拟扩展的二维doa定位方法
CN110399680B (zh) 一种浅海弹性结构辐射声场计算方法
CN112163378A (zh) 根据非线性贝叶斯理论的浅海多层海底地声参数反演方法
CN113064147A (zh) 一种低信噪比下新型匹配场被动定位方法
WO2023221655A1 (zh) 一种水下无线传感网目标位置和环境传播参数的联合估计方法
CN113253248B (zh) 一种基于迁移学习的小样本垂直阵目标距离估计方法
CN110389335B (zh) 一种基于海洋背景场的水声传播损失可靠计算方法
Jin et al. Semi-supervised underwater acoustic source localization based on residual convolutional autoencoder
CN112014791A (zh) 存在阵列误差的阵列pca-bp算法的近场源定位方法
CN105138743B (zh) 一种基于单矢量水听器的相干信号源方位估计方法
Jenkins et al. Bayesian optimization with Gaussian process surrogate model for source localization
Prasad et al. Deep learning techniques for detection of underwater acoustic sources
Li et al. Deep learning-based DOA estimation using CRNN for underwater acoustic arrays
Wang et al. A comparative study for shallow water match-field inversion using surrogate models
CN113688571B (zh) 一种基于卷积相似度的水声环境参数分类方法
Margolina et al. BRS Sound Exposure Modeling Tool: a system for planning, visualization and analysis
CN116879952B (zh) 点源弹性波海底反射系数的计算方法、存储介质和设备
Gerstoft et al. Parametric bootstrapping of array data with a generative adversarial network
Yayu et al. Research on Location of Underwater Sound Source Target in Deep Sea Sound Field Based on Bellhop Model
Patra et al. AI & ML based Implementation of the Underwater Channel Model in the Tropical Littoral Waters of the Indian Ocean Region (IOR)
CN116338574B (zh) 一种基于匹配波束的稀疏贝叶斯学习水下声源定位方法
Qu et al. 3D Deep Residual Convolutional Neural Network for Underwater Acoustic Source Localization Using Local Acoustic Intensity Field
CN111505582B (zh) 基于差分进化算法的水听器阵列定位方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21938341

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21938341

Country of ref document: EP

Kind code of ref document: A1