Underwater topographic survey system and method based on multi-sensor data fusion

Abstract

The invention belongs to the technical field of underwater topographic survey, and particularly relates to an underwater topographic survey system based on multi-sensor data fusion, which comprises a multi-beam sounding sonar, a laser radar sensor, an ultrasonic bathymeter, an optical fiber hydrophone, a side scan sonar and a data processing and fusion module. The invention further relates to an underwater topographic survey method, and the method adopts a Kalman filtering data fusion algorithm. According to the invention, multiple types of sensors such as a multi-beam sounding sonar, a laser radar, an ultrasonic bathymeter, an optical fiber hydrophone, a side-scan sonar and the like are integrated into the same system for the first time, and an acoustics, optics and mechanics multi-dimensional data acquisition system is formed.

Description

Technical Field

[0001] The present invention belongs to the technical field of underwater topography survey, and in particular relates to an underwater topography survey system based on multi-sensor data fusion, and an underwater topography survey method based on multi-sensor data fusion. Background Art

[0002] Underwater topographic surveys are crucial for ocean mapping, seabed resource development, and marine engineering construction. Traditional underwater topographic surveying methods primarily rely on single sensors, such as single-beam or multi-beam sonar. While capable of acquiring depth data with a certain degree of accuracy, these methods are limited in terms of measurement efficiency and data integrity in complex environments. For example, single-beam sonar only provides point-like depth data, resulting in low measurement efficiency. While multi-beam sonar improves measurement efficiency, it can still result in missing data or insufficient accuracy in shallow waters or areas with complex terrain.

[0003] Furthermore, with the deepening of ocean development, higher requirements are being placed on the accuracy and efficiency of underwater topographic surveys. This is especially true in the early planning and construction monitoring phases of submarine engineering construction and marine resource development, which require rapid and accurate acquisition of comprehensive underwater topographic information. Therefore, developing an underwater topographic survey system and method based on multi-sensor data fusion to improve the accuracy and efficiency of underwater topographic surveys is of great practical significance. Summary of the Invention

[0004] In response to the problems in the prior art, the present invention proposes an underwater topographic survey system and method based on multi-sensor data fusion, which are used to obtain underwater topographic data with high precision and high efficiency. The technical solution of the present invention is as follows:

[0005] The underwater terrain survey system based on multi-sensor data fusion includes the following components:

[0006] Multi-beam bathymetric sonar: Used to transmit ultra-wide sound beams to the seabed and receive reflected signals to obtain high-precision water depth data; Multi-beam bathymetric sonar transmits ultra-wide sound beams to the seabed, receives reflected signals from the seabed, and obtains high-density water depth data points based on time difference and angle calculation;

[0007] LiDAR sensor: Scanning obliquely or vertically toward the water surface, the LiDAR sensor emits high-frequency laser pulses, utilizing the laser's reflection off the water surface and shallow penetration characteristics to generate high-precision 3D point cloud data.

[0008] Ultrasonic depth meter: By emitting single-beam ultrasonic waves and measuring the echo time, it calculates the vertical distance from the hull to the bottom of the water in real time;

[0009] Fiber optic hydrophone: Used to collect underwater acoustic signals. By analyzing the spectral characteristics of the acoustic signals, it helps determine the seabed type and the acoustic reflection pattern caused by the terrain. This provides acoustic feature constraints for terrain modeling and helps determine underwater terrain characteristics.

[0010] Side-scan sonar: It emits fan-shaped acoustic beams to scan along both sides of the route, receives backscattered signals from the seabed, and generates high-contrast two-dimensional grayscale images of the seabed. It clearly shows the roughness of the seabed, the distribution of obstacles, and small landform units. It is used to generate two-dimensional images of the seabed topography and detect underwater objects.

[0011] In addition, the data processing and fusion module analyzes and processes the signals collected by all the above components, fuses the data collected by each sensor, and constructs an underwater terrain model.

[0012] As an improvement to the above technical solution, the data processing and fusion module includes the following modules:

[0013] Data preprocessing module, preprocesses multi-source data;

[0014] Data fusion module, which uses spatiotemporal registration and fusion algorithms;

[0015] and, 3D modeling and visualization modules.

[0016] As an improvement to the above technical solution, the data preprocessing module and preprocessing method include:

[0017] Step 1: Correct the sound velocity profile and compensate for the ship's attitude on the sonar data;

[0018] Step 2: Correct the water refraction error and denoise the laser point cloud;

[0019] Step 3: Perform time-frequency analysis on the fiber optic hydrophone signal to extract terrain-related acoustic features.

[0020] As an improvement to the above technical solution, the data fusion module adopts a spatiotemporal registration and fusion algorithm, and the steps are as follows:

[0021] Step 1: Based on a unified geographic coordinate system, project the sensor data into the same grid coordinate system through timestamp synchronization and spatial transformation matrix.

[0022] Step 2: Adopt a multi-level fusion strategy, including:

[0023] a. Data layer fusion: Based on Bayesian networks or evidence theory, similar data are integrated;

[0024] b. Feature layer fusion: Combines the geometric features of the laser point cloud with the texture features of the side-scan sonar to construct a terrain feature vector;

[0025] c. Decision-making layer fusion: Through DS evidence theory or neural networks, multi-source features are integrated to achieve terrain classification and target recognition.

[0026] As an improvement to the above technical solution, the three-dimensional modeling and visualization module has the following working steps:

[0027] Step 1: Use triangulated mesh or grid model to generate seamless underwater terrain surface;

[0028] Step 2: Superimpose optical textures and acoustic properties to construct an immersive three-dimensional terrain scene, supporting real-time interaction and measurement analysis.

[0029] As an improvement to the above technical solution, the data fusion module adopts a multi-level fusion strategy, including a three-level fusion architecture of data layer, feature layer, and decision layer:

[0030] a. Data layer fusion: Directly process raw data, including multi-beam bathymetric points and lidar point clouds, retaining the lowest level of information to avoid feature extraction loss;

[0031] b. Feature layer fusion: Extracts features from each sensor, including terrain slope and object outlines, eliminating data redundancy and improving processing efficiency;

[0032] c. Decision-making layer fusion: Each sensor makes an independent decision and then integrates it, including side-scan sonar to identify object type and multi-beam to confirm location, reducing the risk of misjudgment by a single sensor.

[0033] An underwater terrain survey method based on multi-sensor data fusion, wherein the underwater terrain survey method adopts a Kalman filter data fusion algorithm, includes:

[0034] Step 1: First, define the state of the underwater terrain survey system. Assume that the state vector X k represents the underwater terrain related state at time k, where we assume that X k Contains only water depth information, i.e. X k =[h k ] T , where h k is the depth of the water at time k;

[0035] Step 2: Secondly, the prediction step is performed, including state prediction and error covariance prediction:

[0036] Step 2-1, state prediction

[0037] State prediction is based on the state estimate at the previous moment To predict the current state The calculation formula is:

[0038]

[0039] Among them, A k It is the state transfer matrix, which represents the relationship between the system state and time. In the water depth prediction, if it is assumed that the water depth changes smoothly in a short time, A can be set to k =[1];

[0040] Step 2-2, Error Covariance Prediction

[0041] The error covariance prediction is based on the error covariance matrix P at the previous moment k-1 To predict the error covariance matrix at the current moment The calculation formula is:

[0042]

[0043] Among them, Q k is the process noise covariance matrix, which represents the uncertainty in the process of system state change and reflects uncertain factors such as water flow and geological changes. Assume that Q k =[q], where q is a positive number representing the variance of the process noise;

[0044] Step 3, update step

[0045] Step 3-1, measurement prediction

[0046] The measurement prediction is based on the current state prediction value To predict the sensor's measurement value The calculation formula is:

[0047]

[0048] Among them, H k is the observation matrix, which represents the relationship between the system state and the sensor measurement value. For sensors that directly measure water depth, such as multi-beam depth sonar and ultrasonic depth meter, H can be k =[1];

[0049] Step 3-2, Kalman gain calculation

[0050] Kalman gain K k The weight used to weigh the state prediction value and the sensor measurement value is calculated as follows:

[0051]

[0052] Among them, R k is the measurement noise covariance matrix, which represents the uncertainty in the sensor measurement process;

[0053] Step 3-3, Status Update

[0054] The state update is based on the Kalman gain K k , the actual measured value of the sensor z k and measured predicted values To update the current state estimate The calculation formula is:

[0055]

[0056] Step 3-4, error covariance update

[0057] The error covariance update is based on the Kalman gain K k and the forecast error covariance matrix To update the error covariance matrix P at the current moment k , the calculation formula is:

[0058]

[0059] Where I is the identity matrix.

[0060] As an improvement to the above technical solution, the data fusion algorithm adopts sequential fusion of multi-sensor data and performs Kalman filtering updates on the data of each sensor in turn, including:

[0061] Assume that at time k there are n sensor measurements z k,1 ,z k,2, …,z k,n , you can perform fusion by following the steps below:

[0062] Step 1: Initialize the state estimate and the error covariance matrix

[0063] Step 2: For each sensor i=1,2,…,n:

[0064] a. Calculate the measured predicted value

[0065] b. Calculate Kalman gain

[0066] c. Update state estimate

[0067] d. Update the error covariance matrix

[0068] e. Final state estimate This is the estimated water depth after fusion.

[0069] As an improvement to the above technical solution, the data fusion algorithm adopts parallel fusion of multi-sensor data and performs Kalman filtering update on the data of each sensor in turn, including:

[0070] All sensor data are integrated simultaneously. Assume that at time k there are n sensor measurements z k,1 ,z k,2 ,…,z k,n , these measurements can be combined into a vector And the observation matrix H k Merge into one matrix Measurement noise covariance matrix R k Merge into a matrix R k =diag(R k，1 , R k，2 ,…,R k，n ), directly on the quantized Z k 、H k and R k Perform prediction and update steps.

[0071] The beneficial effects of the present invention are:

[0072] 1. The underwater terrain survey system based on multi-sensor data fusion of the present invention is aimed at existing methods that rely on a single sonar technology, such as single-beam or multi-beam. The present invention is the first to integrate multiple types of sensors such as multi-beam bathymetric sonar, lidar, ultrasonic depth meter, fiber optic hydrophone, side-scan sonar, etc. into the same system, forming an "acoustic + optical + mechanical" multi-dimensional data acquisition system.

[0073] 2. The underwater terrain survey system based on multi-sensor data fusion of the present invention adopts a three-level data fusion system and proposes a three-level fusion architecture of data layer, feature layer and decision layer. Through layered processing, it balances data accuracy and processing efficiency. In the feature layer fusion, invalid data can be quickly filtered out, shortening the complex terrain modeling time by more than 40%.

[0074] 3. The underwater terrain survey system based on multi-sensor data fusion of the present invention can achieve multi-modal data complementarity: deep water areas rely on full coverage depth measurement of multi-beam sonar, shallow water areas use lidar to supplement fine structures, and fiber optic hydrophones and side-scan sonars provide acoustic and texture constraints, eliminating the detection blind spots of a single sensor.

[0075] 4. The underwater topographic survey system based on multi-sensor data fusion of the present invention can realize a dynamic calibration mechanism: the ultrasonic depth meter tracks the vertical movement of the hull in real time, and combines the inertial navigation system (INS) data to realize dynamic tidal compensation and attitude correction, thereby improving the temporal and spatial consistency of the data.

[0076] 5. The underwater terrain survey system based on multi-sensor data fusion of the present invention can implement an intelligent fusion algorithm: it automatically mines the implicit associations of multi-source data (such as the correlation between terrain elevation and sonar backscatter intensity) through machine learning, reduces manual intervention, and improves modeling efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 Shown is a schematic diagram of the structure of an underwater terrain survey system;

[0078] Figure 2 It shows a schematic diagram of the projection distance measurement of the multi-beam bathymetric plane calculation model in the underwater terrain survey method;

[0079] Figure 3 Shown is Figure 2 A simplified diagram of the positions related to the projected distance measurement. DETAILED DESCRIPTION

[0080] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments.

[0081] like Figure 1 The underwater terrain survey system based on multi-sensor data fusion of the present invention includes the following components:

[0082] Multi-beam bathymetric sonar: used to transmit ultra-wide sound beams to the seabed and receive reflected signals to obtain high-precision water depth data; multi-beam bathymetric sonar is installed on the bottom of the survey vessel or towed platform, emitting ultra-wide sound beams to the seabed with a coverage angle of more than 150°. By receiving the reflected signals from the seabed and solving the time difference and angle, high-density water depth data points (with a resolution of up to centimeters) are obtained to form a "full coverage" bathymetric strip of the seabed topography. Multi-beam bathymetric sonar provides basic elevation data of the seabed topography and is suitable for large-scale topographic mapping in deep water areas.

[0083] LiDAR sensor: Installed on a vessel above the water, LiDAR provides extremely high spatial resolution, generating high-precision three-dimensional point cloud data that clearly displays the details of underwater terrain and objects. Mounted on the mast or elevated platform of a surface vessel, LiDAR sensors scan at an angle or vertically toward the water surface. They emit high-frequency laser pulses (typically at a wavelength of 532nm or 1064nm) and utilize the laser's reflection off the water surface and shallow penetration characteristics (suitable for clear waters, with penetration depths of up to several meters) to generate high-precision three-dimensional point cloud data (with spatial resolution down to millimeters). This clearly depicts near-surface terrain (such as shoals, coral reefs, and underwater vegetation) and the outlines of objects (such as shipwrecks and reefs) beneath the surface. LiDAR sensors compensate for the insufficient resolution of sonar in shallow waters or areas with complex structures, providing refined terrain details.

[0084] Ultrasonic Depth Gauge: Installed on the bottom of a ship, directly in contact with the water, it measures water depth in real time. By emitting a single-beam ultrasonic wave and measuring the echo time, the ultrasonic depth gauge calculates the vertical distance from the ship to the bottom in real time (with a response speed of milliseconds). The ultrasonic depth gauge provides real-time depth reference data for dynamic calibration of the multibeam sonar depth reference, eliminating errors caused by changes in the ship's draft or wave interference.

[0085] Fiber-optic hydrophones are deployed via cables or buoy systems at different depths underwater (e.g., near the seabed, mid-water, and below the surface). Based on the principle of fiber-optic interferometry, they sense changes in underwater sound pressure fields and collect acoustic signals (frequency range: 10 Hz to 100 kHz) such as fish activity, water disturbances, and reflections from the seabed. Fiber-optic hydrophones collect underwater acoustic signals and analyze their spectral characteristics to help determine the seabed type (e.g., sand, rock, silt) and the reflection patterns caused by terrain undulations. This provides acoustic feature constraints for terrain modeling and aids in determining underwater terrain characteristics.

[0086] Side-scan sonar: Using a towed probe or fixed on both sides of the ship's bottom, the side-scan sonar emits a fan-shaped acoustic beam that scans along both sides of the route, receives the seabed backscatter signal, and generates a high-contrast two-dimensional grayscale image of the seabed (with a lateral resolution of up to decimeters). It clearly displays the roughness of the seabed, the distribution of obstacles (such as pipelines and reefs), and small landform units (such as gullies and sand waves). It is used to generate two-dimensional images of the seabed topography and detect underwater objects. The side-scan sonar provides texture characteristics of the seabed topography and target detection capabilities, and assists multi-beam data in terrain classification and object recognition.

[0087] In addition, the data processing and fusion module analyzes and processes the signals collected by all the above components, fuses the data collected by each sensor, and constructs a comprehensive and accurate underwater terrain model.

[0088] The hardware architecture of the data processing and fusion module is integrated into the shipboard industrial computer or edge computing unit, and has high-speed data interfaces (such as PCIe, Gigabit Ethernet) and parallel computing capabilities (GPU / TPU acceleration).

[0089] Data processing and fusion module, including the following modules:

[0090] 1. Data preprocessing module, which preprocesses multi-source data. The preprocessing methods include:

[0091] Step 1: Perform sound velocity profile correction and hull attitude compensation (roll, pitch, and bow) on the sonar data;

[0092] Step 2: Correct the water refraction error and denoise the laser point cloud (such as RANSAC filtering);

[0093] Step 3: Perform time-frequency analysis on the fiber optic hydrophone signal to extract terrain-related acoustic features (such as bottom reflection coefficient).

[0094] 2. Data fusion module uses spatiotemporal registration and fusion algorithms. The main steps are as follows:

[0095] Step 1: Based on a unified geographic coordinate system (such as WGS84), project the sensor data into the same grid coordinate system through timestamp synchronization (nanosecond accuracy) and spatial transformation matrices (rotation, translation, and scaling).

[0096] Step 2: Adopt a multi-level fusion strategy:

[0097] a. Data layer fusion: Based on Bayesian networks or evidence theory, similar data (such as multi-beam and ultrasonic water depth) are integrated;

[0098] b. Feature layer fusion: Combines the geometric features (curvature, height) of the laser point cloud with the texture features (grayscale entropy, edge intensity) of the side-scan sonar to construct a terrain feature vector;

[0099] c. Decision-making layer fusion: Through DS evidence theory or neural networks, multi-source features are integrated to achieve terrain classification (such as plains, hills, and steep slopes) and target recognition.

[0100] 3. The main steps of the 3D modeling and visualization module are as follows:

[0101] Step 1: Use triangulated network (TIN) or grid (DEM) model to generate seamless underwater terrain surface;

[0102] Step 2: Superimpose optical textures (such as side-scan sonar images) and acoustic properties (such as bottom classification results) to construct an immersive three-dimensional terrain scene that supports real-time interaction and measurement analysis.

[0103] The underwater terrain survey system based on multi-sensor data fusion of the present invention adopts a three-level data fusion system and proposes a three-level fusion architecture of data layer, feature layer and decision layer:

[0104] a. Data layer fusion: Directly process raw data, such as multi-beam bathymetric points and lidar point clouds, retaining the lowest level of information to avoid feature extraction loss;

[0105] b. Feature layer fusion: Extracts sensor features, such as terrain slope and object outline, to eliminate data redundancy and improve processing efficiency;

[0106] c. Decision-making layer fusion: Each sensor makes an independent decision and then integrates it. For example, side-scan sonar identifies object type and multi-beam confirms location, reducing the risk of misjudgment by a single sensor.

[0107] The underwater terrain survey system based on multi-sensor data fusion of the present invention balances data accuracy and processing efficiency through layered processing. In feature layer fusion, invalid data can be quickly filtered out, shortening the complex terrain modeling time by more than 40%.

[0108] The underwater terrain survey system based on multi-sensor data fusion of the present invention can achieve multi-modal data complementarity: deep water areas rely on full coverage depth measurement of multi-beam sonar, shallow water areas use lidar to supplement fine structures, and fiber optic hydrophones and side-scan sonars provide acoustic and texture constraints, eliminating the detection blind spots of a single sensor.

[0109] The underwater topographic survey system based on multi-sensor data fusion of the present invention can realize a dynamic calibration mechanism: the ultrasonic depth meter tracks the vertical movement of the hull in real time, and combines the inertial navigation system (INS) data to realize dynamic tidal compensation and attitude correction, thereby improving the temporal and spatial consistency of the data.

[0110] The underwater terrain survey system based on multi-sensor data fusion of the present invention can implement an intelligent fusion algorithm: it automatically mines the implicit associations of multi-source data (such as the correlation between terrain elevation and sonar backscatter intensity) through machine learning, reduces manual intervention, and improves modeling efficiency and accuracy.

[0111] The following uses the Kalman filter data fusion algorithm as an example to explain how an underwater terrain survey system uses Kalman filtering to fuse multi-sensor data. Kalman filtering is a recursive optimal linear estimation method that consists of two main steps: a prediction step and an update step.

[0112] Step 1: First, define the state of the underwater terrain survey system. Assume that the state vector X k represents the underwater terrain related state at time k, such as water depth, terrain slope, etc. For simplicity, it is assumed here that X k Contains only water depth information, i.e. X k =[h k ] T , where h k is the water depth at time k.

[0113] Step 2, secondly, the prediction step is performed, including state prediction and error covariance prediction.

[0114] Step 2-1, state prediction

[0115] State prediction is based on the state estimate at the previous moment To predict the current state The calculation formula is:

[0116]

[0117] Among them, A k Is the state transfer matrix, which represents the relationship between the system state and time. In water depth prediction, if it is assumed that the water depth changes slowly in a short time, A k =[1].

[0118] Step 2-2, Error Covariance Prediction

[0119] The error covariance prediction is based on the error covariance matrix P at the previous moment k-1 To predict the error covariance matrix at the current moment The calculation formula is:

[0120]

[0121] Among them, Q k is the process noise covariance matrix, which represents the uncertainty in the process of system state change. It reflects the uncertainty factors such as water flow and geological changes. Assume that Q k =[q], where q is a positive number representing the variance of the process noise.

[0122] Step 3, update step

[0123] Step 3-1, measurement prediction

[0124] The measurement prediction is based on the current state prediction value To predict the sensor's measurement value The calculation formula is:

[0125]

[0126] Among them, H k is the observation matrix, which represents the relationship between the system state and the sensor measurement value. For sensors that directly measure water depth, such as multi-beam depth sonar and ultrasonic depth meter, H k =[1].

[0127] Step 3-2, Kalman gain calculation

[0128] Kalman gain K k The weight used to weigh the state prediction value and the sensor measurement value is calculated as follows:

[0129]

[0130] Among them, R k is the measurement noise covariance matrix, which represents the uncertainty in the sensor measurement process. Different sensors have different measurement accuracy, so R kIt needs to be set according to the characteristics of the sensor. For example, the measurement accuracy of multi-beam depth sounding sonar is high, and its measurement noise variance r1 is small; the measurement accuracy of ultrasonic depth meter is relatively low, and its measurement noise variance r2 is large. Assume R k = [r], where r is the measurement noise variance of the sensor.

[0131] Step 3-3, Status Update

[0132] The state update is based on the Kalman gain K k , the actual measured value of the sensor z k and measured predicted values To update the current state estimate The calculation formula is:

[0133]

[0134] Step 3-4, error covariance update

[0135] The error covariance update is based on the Kalman gain K k and the forecast error covariance matrix To update the error covariance matrix P at the current moment k , the calculation formula is:

[0136]

[0137] Where I is the identity matrix.

[0138] In underwater topographic survey systems, multiple sensors can measure water depth information, such as multi-beam bathymetric sonars and ultrasonic depth meters. To fuse the data from these sensors, either sequential or parallel multi-sensor data fusion can be used.

[0139] Multi-sensor data is sequentially fused, and the Kalman filter is updated for each sensor’s data in turn. Assume that at time k there are n sensor measurements z k,1 ,z k,2, …,z k,n , you can perform fusion by following the steps below:

[0140] Step 1: Initialize the state estimate and the error covariance matrix

[0141] Step 2: For each sensor i=1,2,…,n:

[0142] a. Calculate the measured predicted value

[0143] b. Calculate Kalman gain

[0144] c. Update state estimate

[0145] d. Update the error covariance matrix

[0146] e. Final state estimate This is the estimated water depth after fusion.

[0147] Parallel fusion of multi-sensor data: fusion of all sensor data at the same time. Assume that at time k there are n sensor measurements z k,1 ,z k,2 ,…,z k,n , these measurements can be combined into a vector And the observation matrix H k Merge into one matrix Measurement noise covariance matrix R k Merge into a matrix R k =diag(R k，1 , R k，2 ,…,R k，n ), directly on the quantized Z k 、H k and R k Perform prediction and update steps.

[0148] By sequential fusion of multi-sensor data or parallel fusion of multi-sensor data, the Kalman filter algorithm can be used to fuse the data of multiple sensors in the underwater terrain survey system to obtain more accurate underwater terrain information. In practical applications, it is necessary to reasonably set the state transfer matrix A according to the characteristics of the sensor and the requirements of the system. k , observation matrix H k , process noise covariance matrix Q k and the measurement noise covariance matrix R k .

[0149] The above embodiment only takes the water depth as an example, that is, assuming that X k It only contains water depth information. In fact, the data fusion algorithm of the present invention can also be extended to the data fusion of other information collected in other underwater terrain survey methods.

[0150] like Figure 2 and Figure 3 As shown in FIG, it is the multi-beam bathymetric plane calculation model in the underwater terrain survey method.

[0151] In order to adapt to different scenarios, the calculation of the overlap rate needs to be expanded, that is, the overlap rate is equal to the ratio of the length of the overlapping part to the total coverage length.

[0152] η=1-(WL 重 ) / W Formula (1)

[0153] Where η is the overlap rate between adjacent strips, W is the coverage width of the survey line, and L 重 is the length of the overlapping parts of adjacent strips.

[0154] Draw a straight line parallel to the seabed slope, project the line spacing d onto the straight line with the two rightmost lines as parallel lines, and translate it to the seabed slope, as shown in the following example: Figure 2 As shown, at this time d'=WL 重 , which can eliminate the influence of slope on overlap rate calculation.

[0155] The overlap ratio calculation method shown in formula (1) can effectively avoid overlap ratio failure caused by different scenarios, that is, the overlap ratio is equal to the ratio of the overlap length to the total coverage length.

[0156] The left part is used as the deeper area to draw the relevant position. Figure 3 As shown, use the geometric relationship in the figure to write the calculation method of each parameter, where AB is recorded as W i左 , BC is recorded as W i右 .

[0157] Seawater depth: if x Dj It is known that referring to the trigonometric formula, we have:

[0158] x Di =x Dj +Δd·tanα Formula (2)

[0159] Among them, x Di and x Dj Represent the water depth on the left and right sides of the figure respectively.

[0160] Coverage width: calculated using the sine theorem

[0161]

[0162] Among them, W i is the coverage width of the left survey line, and other angles are indicated in the figure.

[0163] Overlap rate:

[0164] η=1-d' / W Formula (4)

[0165]

[0166] Multibeam bathymetry 3D computational model:

[0167] Assume the heading normal is n 航=(a,b,c), consider this vector as representing a 航 The plane with normal direction is n. Since the sea level is horizontal, the plane is perpendicular to the sea level. 海 =(a',b',c'), then the slope line is the intersection line of two planes, and the direction of the slope line can be calculated by Cramer's rule as (bc'-b'c,ca'-ac',ab'-ba'). Therefore, when the heading of the survey ship is determined (the direction of the survey line), the slope is uniquely determined and is a constant value.

[0168] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. The underwater terrain survey system based on multi-sensor data fusion is characterized by: Includes the following components: Multi-beam bathymetric sonar: used to transmit ultra-wide sound beams to the seabed and receive reflected signals to obtain high-precision water depth data; Multi-beam bathymetric sonar transmits an ultra-wide sound beam to the seabed, receives the reflected signal from the seabed, and obtains high-density water depth data points based on time difference and angle calculation; LiDAR sensor: Scanning obliquely or vertically toward the water surface, the LiDAR sensor emits high-frequency laser pulses, utilizing the laser's reflection off the water surface and shallow penetration characteristics to generate high-precision 3D point cloud data. Ultrasonic depth meter: By emitting single-beam ultrasonic waves and measuring the echo time, it calculates the vertical distance from the hull to the bottom of the water in real time; Fiber optic hydrophone: Used to collect underwater acoustic signals. By analyzing the spectral characteristics of the acoustic signals, it helps determine the seabed type and the acoustic reflection pattern caused by the terrain. This provides acoustic feature constraints for terrain modeling and helps determine underwater terrain characteristics. Side-scan sonar: It emits fan-shaped acoustic beams to scan along both sides of the route, receives backscattered signals from the seabed, and generates high-contrast two-dimensional grayscale images of the seabed. It clearly shows the roughness of the seabed, the distribution of obstacles, and small landform units. It is used to generate two-dimensional images of the seabed topography and detect underwater objects. In addition, the data processing and fusion module analyzes and processes the signals collected by all the above components, fuses the data collected by each sensor, and constructs an underwater terrain model.

2. The underwater topographic survey system based on multi-sensor data fusion according to claim 1, characterized in that: The data processing and fusion module includes the following modules: Data preprocessing module, preprocesses multi-source data; Data fusion module, which uses spatiotemporal registration and fusion algorithms; and, 3D modeling and visualization modules.

3. The underwater topographic survey system based on multi-sensor data fusion according to claim 2, characterized in that: The data preprocessing module and the preprocessing method include: Step 1: Correct the sound velocity profile and compensate for the ship's attitude on the sonar data; Step 2: Correct the water refraction error and denoise the laser point cloud; Step 3: Perform time-frequency analysis on the fiber optic hydrophone signal to extract terrain-related acoustic features.

4. The underwater topographic survey system based on multi-sensor data fusion according to claim 2, characterized in that: The data fusion module adopts the spatiotemporal registration and fusion algorithm, and the steps are as follows: Step 1: Based on a unified geographic coordinate system, project the sensor data into the same grid coordinate system through timestamp synchronization and spatial transformation matrix. Step 2: Adopt a multi-level fusion strategy, including: a. Data layer fusion: Based on Bayesian networks or evidence theory, similar data are integrated; b. Feature layer fusion: Combines the geometric features of the laser point cloud with the texture features of the side-scan sonar to construct a terrain feature vector; c. Decision-making layer fusion: Through DS evidence theory or neural networks, multi-source features are integrated to achieve terrain classification and target recognition.

5. The underwater topographic survey system based on multi-sensor data fusion according to claim 2, characterized in that: The three-dimensional modeling and visualization module operates as follows: Step 1: Use triangulated mesh or grid model to generate seamless underwater terrain surface; Step 2: Superimpose optical textures and acoustic properties to construct an immersive three-dimensional terrain scene, supporting real-time interaction and measurement analysis.

6. The underwater topographic survey system based on multi-sensor data fusion according to claim 4, characterized in that: The data fusion module adopts a multi-level fusion strategy, including a three-level fusion architecture of data layer, feature layer, and decision layer: a. Data layer fusion: Directly process raw data, including multi-beam bathymetric points and lidar point clouds, retaining the lowest level of information to avoid feature extraction loss; b. Feature layer fusion: Extracts features from each sensor, including terrain slope and object outlines, eliminating data redundancy and improving processing efficiency; c. Decision-making layer fusion: Each sensor makes an independent decision and then integrates it, including side-scan sonar to identify object type and multi-beam to confirm location, reducing the risk of misjudgment by a single sensor.

7. The underwater terrain survey method based on multi-sensor data fusion according to any one of claims 1 to 6, characterized in that: The underwater terrain survey method adopts a Kalman filter data fusion algorithm, including: Step 1: First, define the state of the underwater terrain survey system. Assume that the state vector X k represents the underwater terrain related state at time k, where it is assumed that X k Contains only water depth information, i.e. X k =[h k ] T , where h k is the depth of the water at time k; Step 2: Secondly, the prediction step is performed, including state prediction and error covariance prediction: Step 2-1, state prediction State prediction is based on the state estimate at the previous moment To predict the current state The calculation formula is: Among them, A k It is the state transfer matrix, which represents the relationship between the system state and time. In the water depth prediction, if it is assumed that the water depth changes smoothly in a short time, A can be set to k =[1]; Step 2-2, Error Covariance Prediction The error covariance prediction is based on the error covariance matrix P at the previous moment k-1 To predict the error covariance matrix at the current moment The calculation formula is: Among them, Q k is the process noise covariance matrix, which represents the uncertainty in the process of system state change and reflects uncertain factors such as water flow and geological changes. Assume that Q k =[q], where q is a positive number representing the variance of the process noise; Step 3, update step Step 3-1, measurement prediction The measurement prediction is based on the current state prediction value To predict the sensor's measurement value The calculation formula is: Among them, H k is the observation matrix, which represents the relationship between the system state and the sensor measurement value. For sensors that directly measure water depth, such as multi-beam depth sonar and ultrasonic depth meter, H can be k =[1]; Step 3-2, Kalman gain calculation Kalman gain K k The weight used to weigh the state prediction value and the sensor measurement value is calculated as follows: Among them, R k is the measurement noise covariance matrix, which represents the uncertainty in the sensor measurement process; Step 3-3, Status Update The state update is based on the Kalman gain K k , the actual measured value of the sensor z k and measured predicted values To update the current state estimate The calculation formula is: Step 3-4, error covariance update The error covariance update is based on the Kalman gain K k and the forecast error covariance matrix To update the error covariance matrix P at the current moment k , the calculation formula is: Where I is the identity matrix.

8. The underwater terrain survey method based on multi-sensor data fusion according to claim 7, characterized in that: The data fusion algorithm adopts the sequential fusion of multi-sensor data and performs Kalman filter update on the data of each sensor in turn, including: Assume that at time k there are n sensor measurements z k,1 ,z k,2, …,z k,n , you can perform fusion by following the steps below: Step 1: Initialize the state estimate and the error covariance matrix Step 2: For each sensor i=1,2,…,n: a. Calculate the measured predicted value b. Calculate Kalman gain c. Update state estimate d. Update the error covariance matrix e. Final state estimate This is the estimated water depth after fusion.

9. The underwater terrain survey method based on multi-sensor data fusion according to claim 7, characterized in that: The data fusion algorithm adopts parallel fusion of multi-sensor data and performs Kalman filter update on the data of each sensor in turn, including: All sensor data are integrated simultaneously. Assume that at time k there are n sensor measurements z k,1 ,z k,2 ,…,z k,n , these measurements can be combined into a vector And the observation matrix H k Merge into one matrix Measurement noise covariance matrix R k Merge into a matrix R k =diag(R k，1 ,R k，2 ,…,R k，n ), directly on the quantized Z k 、H k and R k Perform prediction and update steps.

Images