An ADCP-based multi-beam coordinated current profile dynamic correction system and method

Abstract

The present application relates to the technical field of water flow profile correction, in particular to a multi-beam cooperative water flow profile dynamic correction system based on ADCP, which comprises a multi-beam ADCP cooperative sensing array, an environmental parameter sensing module, a dynamic correction and fusion module, a data feedback and communication network and a visualization terminal. The multi-beam cooperative water flow profile dynamic correction system and method based on ADCP breaks through the single-beam measurement blind area by constructing a multi-beam ADCP cooperative measurement network and synchronously collecting water flow velocity data at different azimuths and depths. At the same time, a dynamic correction algorithm is designed to combine real-time water flow turbulence intensity, water sound velocity profile and other environmental parameters to perform real-time fusion and deviation correction on multi-beam data. A data feedback closed loop is built to real-time return the corrected water flow profile data to the measurement terminal, realize dynamic iteration of measurement-correction-output, and improve the accuracy and timeliness of water flow profile measurement in complex water areas (such as estuary and reservoir area).

Description

Technical Field

[0001] This invention relates to the field of flow profile correction technology, specifically to a multi-beam collaborative flow profile dynamic correction system and method based on ADCP. Background Technology

[0002] Multibeam collaborative dynamic flow profile correction is a technology based on acoustic Doppler current profiler (ADCP) for precise measurement and dynamic optimization of flow profile parameters. Its core lies in the three-dimensional deployment and spatiotemporal synchronization of multibeam ADCP nodes, combined with real-time feedback of aquatic environmental parameters, to correct systematic errors in the measurement process, ultimately outputting high-confidence, three-dimensional flow field data with no blind spots across the entire domain.

[0003] Because traditional ADCP (Acoustic Doppler Current Profiler) is easily affected by water turbulence and water impurities when performing single-beam measurement of water flow profiles, and multi-beam collaborative measurement has problems such as data deviation and correction lag, a multi-beam collaborative dynamic correction system and method for water flow profile based on ADCP is proposed. Summary of the Invention

[0004] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a multi-beam collaborative dynamic flow profile correction system and method based on ADCP. It has the advantage of constructing multi-beam ADCP collaborative measurement and dynamic correction to achieve dynamic iteration of measurement-correction-output. It solves the problems of existing acoustic Doppler current profilers being susceptible to interference in single-beam measurement and prone to data deviation and correction lag in multi-beam collaborative measurement.

[0005] (II) Technical Solution To achieve the aforementioned goal of constructing a multi-beam ADCP collaborative measurement and dynamic correction system to realize the dynamic iteration of measurement-correction-output, this invention provides the following technical solution: a multi-beam collaborative flow profile dynamic correction system based on ADCP, comprising a multi-beam ADCP collaborative sensing array, employing a heterogeneous node network with a master-slave architecture, and based on three-dimensional deployment and high-precision spatiotemporal synchronization, realizing redundant measurement and spatial complementarity of the flow profile; The environmental parameter sensing module integrates a synchronous sensing unit and acquires the acoustic characteristics and physical state of the water body in real time, providing physical driving parameters for dynamic correction. The dynamic correction and fusion module performs ray tracking correction, turbulence adaptive filtering, and mass-weighted vector fusion to achieve real-time conversion of multi-source heterogeneous data into a three-dimensional flow field. The data feedback and communication network, based on a closed-loop communication architecture with bidirectional transmission and priority scheduling, ensures online self-calibration and adaptive optimization of the system. The visualization terminal is a decision support system that integrates data management, 3D rendering, and interactive control, enabling immersive visualization and real-time interactive analysis of calibration results.

[0006] Preferably, the multi-beam ADCP collaborative sensing array includes an array communication control unit, a cluster of sensor nodes, and an array power unit. The array communication control unit consists of a main control computer that runs the array control, a clock synchronizer that deploys the IEEE 1588 precision time protocol, and a central communication hub with integrated communication interfaces. The central communication hub is responsible for establishing bidirectional data links with each node. The sensor node cluster is divided into a master ADCP node deployed in the cross-section or the core area of ​​the water flow, and several slave ADCP nodes deployed in an asymmetric three-dimensional manner. The asymmetric three-dimensional deployment technology includes: 1) Near-shore nodes are deployed near both banks, and the beam can be tilted towards the center of the river to measure the bank flow velocity and complex backflow areas that are difficult to detect with traditional vertical beams; 2) Set up layered nodes anchored at different water depths, and the beam can be emitted horizontally or at an angle to measure vertical flow velocity shear and bottom flow velocity profile; 3) The bottom node is deployed on the bottom with the beam emitting upwards to compensate for the near-bottom blind zone measurement of the surface ADCP; The master ADCP node and the slave ADCP node adopt multi-beam ADCP and are configured with attitude and position reference subsystem, including: 1) using an inertial measurement unit (IMU) to measure the roll, pitch and heading angle data of each sensor node in real time, and converting the beam coordinate system velocity into the absolute geodetic coordinate system. 2) Use GNSS receivers for surface nodes to provide the absolute latitude and longitude coordinates of the nodes; 3) Use an underwater acoustic locator for underwater and bottom-sitting nodes, and determine the underwater coordinates by acoustic ranging with a reference source at a known position on the water surface; The array energy unit uses underwater buoys, buoys, or ship-mounted towed bodies as deployment platforms to build a power supply system with a hybrid architecture of shore power grid and solar power generation.

[0007] Preferably, the environmental parameter sensing module includes a sound velocity profile measurement unit, an optical scattering sensing unit, a turbulence intensity quantification unit, and a data integration interface unit. The sound velocity profile measurement unit consists of a vertical chain sensor array composed of multiple sound velocity probes and a temperature-depth-salinity profiler. The optical scattering sensing unit includes an optical turbidity sensor composed of a turbidimeter and a laser particulate matter analyzer, which measures the light intensity scattered by suspended particles in the water body based on the emitted beam and evaluates the turbidity. It also includes a backscattering intensity monitoring module built into the ADCP for recording the echo intensity of each depth unit. The turbulence intensity quantization unit includes an acoustic Doppler velocimeter for acquiring instantaneous flow velocity fluctuations and a turbulence inversion model for quantifying turbulence. The ADCP echo signal-to-noise ratio and velocity signal-to-noise ratio are input to the turbulence inversion model, and the spatial distribution of turbulence intensity is inverted based on the original ADCP data.

[0008] Preferably, the dynamic correction and fusion module includes a data cache pool for the access data integration interface unit, and the data cache pool stores the following data based on a first-in-first-out data queue: 1) Raw velocity data packets from each ADCP node; 2) Parameter matrices such as sound velocity profile, turbidity, and turbulence intensity index from the environmental perception module; 3) IMU attitude and positioning data from each node; The dynamic correction and fusion module also includes a correction module and a fusion module. The correction module is divided into a sound velocity path corrector, a turbulence adaptive filter and a mass weight dynamic calculator. The sound velocity path corrector takes the real-time sound velocity profile as input and uses the ray tracing method to simulate the bending propagation path of each sound beam in the real sound velocity profile. It outputs the real geometric position and beam incident angle of each depth unit to correct the depth deviation and flow velocity calculation error. The turbulence adaptive filter takes the real-time turbulence intensity index and the original flow velocity data as input, and outputs a flow velocity estimate optimized by turbulence error based on the adaptive low-pass filter. The quality weight dynamic calculator takes signal-to-noise ratio, echo intensity, turbulence index, turbidity and beam incidence angle as input, runs a weighting function to calculate the comprehensive quality weight value for each data point, and outputs weighting factors for subsequent data fusion. In the fusion module, for each grid node, all corrected velocity data points within its surrounding range are searched, and a weighted vector average fusion is performed using the optimal estimation fusion algorithm.

[0009] Preferably, the real-time data feedback and communication network includes a sensing layer network connecting underwater and surface sensing nodes, and the sensing layer network is divided into an underwater acoustic communication network, a short-range wireless mesh network, and a surface communication gateway. The underwater acoustic communication network integrates an underwater acoustic communication modem on each underwater ADCP node and environmental sensor to enable data transmission between underwater nodes and between underwater nodes and the surface gateway; the short-range wireless mesh network includes Wi-Fi and LoRa modules for surface nodes, which self-organize the nodes on the surface platform into a network and hop data to the gateway; the surface communication gateway is a shipborne computer integrating multiple network interfaces to collect data from all nodes. The real-time data feedback and communication network includes a decision-making layer network connecting node gateways and shore-based cloud. The decision-making layer network is divided into broadband wireless links and wired links. The broadband wireless links include 5G communication modules and wireless private networks, and the wired links are Ethernet. The real-time data feedback and communication network also includes a control layer network connecting the central processing unit and the sensor nodes. The control layer network includes an encoding and compression module and a communication channel for issuing correction instructions and control commands from the central processing unit.

[0010] Preferably, the visualization terminal includes a functional engine layer and a human-computer interaction layer. The functional engine layer includes a scene management and rendering engine, which constructs a virtual underwater environment based on a 3D graphics engine and receives flow velocity data to generate a 3D water flow field visualization model. The functional engine layer also includes a data analysis and alarm module, which has a built-in algorithm library for providing streamline calculation, vorticity identification, cross-sectional flow integral and vertical average flow velocity calculation and analysis, and system alarms when real-time data triggers rules according to set thresholds. The human-computer interaction layer includes a multi-view collaborative interface and an interactive control panel. The multi-view collaborative interface is equipped with a three-dimensional dynamic flow field diagram that dynamically displays the direction, velocity and three-dimensional structure of the water flow, a two-dimensional cross-sectional diagram for quantitative analysis of flow velocity, a data quality heat map that displays the confidence level of cross-sectional area data, and an instrument status panel that displays the data of each ADCP node in real time. The interactive control panel is equipped with a time axis controller, a view controller and a layer controller.

[0011] A multi-beam cooperative dynamic flow profile correction method based on ADCP includes the following specific steps: S1. Based on the precision clock synchronization protocol, the distributed ADCP node array is triggered to perform synchronous beam transmission and data acquisition, and a unified spatiotemporal reference is established; S2. Perform coordinate transformation on the original flow velocity data of each node, and implement outlier removal and initial quality code assignment based on signal coherence and echo intensity; S3. Construct a weighting function to dynamically calculate the weights of multi-source data and interpolate and match them to a unified spatial grid; S4. Use real-time sound velocity profiles for ray tracking, correct beam geometry and depth deviations, and combine turbulence intensity index to drive an adaptive filter to suppress flow velocity pulsation noise. S5. Perform vector weighted fusion based on mass weights on a standardized spatial grid to generate a high-confidence velocity field with no blind spots across the entire domain; S6. Compare the fusion results with the original node data to identify systematic deviations, generate correction parameters and inject them down to the corresponding nodes to achieve online self-calibration of measurement parameters; S7. Output the final corrected flow profile and quality assessment report, and realize closed-loop dynamic optimization of measurement-correction-output.

[0012] (III) Beneficial Effects Compared with the prior art, the present invention provides a multi-beam cooperative dynamic flow profile correction system and method based on ADCP, which has the following beneficial effects: This multi-beam collaborative flow profile dynamic correction system and method based on ADCP constructs a multi-beam ADCP collaborative measurement network to simultaneously collect flow velocity data at different azimuths and depths, overcoming the blind spots of single-beam measurement. Simultaneously, a dynamic correction algorithm is designed, combining real-time flow turbulence intensity, water body sound velocity profile, and other environmental parameters to perform real-time fusion and deviation correction of the multi-beam data. A data feedback closed loop is established to transmit the corrected flow profile data back to the measurement terminal in real time, realizing dynamic iteration of measurement-correction-output. This improves the accuracy and timeliness of flow profile measurement in complex water areas, thus providing more reliable data support for water conservancy project scheduling and water environment monitoring. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of the multi-beam cooperative dynamic flow profile correction system of the present invention; Figure 2 This is a flowchart of the multi-beam coordinated dynamic correction method for water flow profiles according to the present invention. Detailed Implementation

[0014] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] Example 1 In this embodiment, the specific working process of the multi-beam ADCP cooperative sensing array is as follows: 1) Initialization and task synchronization: The main control computer sends the measurement task to all sensor nodes through the communication network. The task includes the measurement start time, pulse interval, sampling frequency, beam mode and working duration. The high-precision clock synchronization module is activated to provide a unified timestamp for synchronization of all nodes.

[0016] 2) Synchronous acquisition and data appending: At the same preset time, all ADCP nodes simultaneously emit acoustic pulses into the water. Each sensing node receives backscattered signals from water bodies at different depths and independently calculates the three-dimensional flow velocity vector (east, north, and sky directions) of each depth unit. While generating the original flow velocity data, the IMU and positioning module of each node work synchronously to append the current attitude data and position data to each frame of flow velocity data.

[0017] 3) Data uplink transmission: All slave nodes transmit raw data packets with spatiotemporal references to the master computer in real time through the communication network.

[0018] The specific working process of the environmental parameter sensing module is as follows: 1) Synchronous acquisition: When the ADCP array starts to emit acoustic pulses, a synchronous trigger signal is also sent to the environmental parameter sensing module. All sensors synchronously acquire the environmental data at the current moment. The high-frequency turbulence sensor continuously performs high-speed sampling and aligns its data with the ADCP measurement cycle.

[0019] 2) Data processing and parameterization: Sound velocity profile generation: The temperature, depth and salt profiler will collect the raw data of T, C and P and calculate a complete sound velocity-depth curve in real time through the sound velocity formula. The vertical chain sensor array will directly output the curve, the turbidity meter will output the turbidity value, and the internal circuit of ADCP will output the echo intensity value of each beam and each depth unit. The standard deviation or turbulent kinetic energy of the velocity data is directly calculated. Based on the inversion model, the signal-to-noise ratio and correlation of the ADCP data are read, and a turbulence intensity index profile is output in real time.

[0020] 3) Generate an environmental parameter matrix: All processed environmental parameters are integrated into an environmental parameter matrix according to their corresponding depths. The parameters of this environmental parameter matrix include depth, sound velocity, turbidity, echo intensity, and turbulence intensity.

[0021] The specific working process of the dynamic correction and fusion module is as follows: 1) Spatiotemporal synchronization and standardization: Raw data enters the data cache pool, and all data is time-aligned with the system clock as the reference. The coordinate system is unified using IMU and positioning data.

[0022] 2) Parallel dynamic correction: The sound velocity path corrector reads the sound velocity profile, performs ray tracing on each depth unit of each beam, calculates its true spatial coordinates and beam geometry, and corrects the flow velocity value; the turbulence adaptive filter reads the turbulence intensity index, dynamically adjusts the filter parameters, smooths the flow velocity data, and filters out unreasonable turbulence pulsation noise; the quality weight dynamic calculator integrates various quality indicators and assigns a dynamic weight Q to each data point.

[0023] 3) Quality weighting and mesh fusion: The entire measurement area is divided into a regular 3D mesh. For each mesh, all neighboring data points are searched, and the quality weights are calculated based on their weights Q. i By performing a weighted vector average, the optimal velocity estimate V for that grid point is obtained. fused .

[0024] The specific working process of data feedback and communication networks is as follows: 1) Synchronous data uplink: At the beginning of a measurement cycle, all sensing nodes synchronously complete data acquisition, and each node sends data packets with a unified timestamp to the gateway through the sensing layer network; The master node allocates communication time slots for each underwater node using the Time Division Multiple Access (TDMA) protocol to achieve underwater acoustic communication; the mesh network routing protocol selects the best path to the gateway to achieve surface wireless communication, and the data is packaged and uploaded to the central processing unit through the backhaul network.

[0025] 2) Central Processing and Instruction Generation: The central processing unit completes the dynamic correction and fusion of data and triggers feedback control logic. It compares and analyzes the fused true value with the original data of each node. If a systematic deviation is found in a certain node, a digital correction value is generated. If it is determined that the turbulence in a certain area is enhanced, an instruction is generated.

[0026] 3) Low-latency instruction downlink: The compressed instruction is sent by the central processing unit to the surface gateway through the backhaul network, and the gateway searches for the target node through the perception layer network.

[0027] 4) Node Adaptation and Loop Formation: After receiving the instruction, the target node's internal microprocessor will immediately parse and apply the instruction. In the next measurement cycle, the node will work with the new parameters, thereby outputting optimized new data.

[0028] The specific working process of the visualization terminal is as follows: 1) Scene Update and Rendering: The engine reads the latest flow field and node location data from the database. Based on the user's view settings, the engine performs scientific computing visualization, including: Vector Arrow: Generates an arrow at each grid point pointing in the direction of water flow, with a length proportional to the speed, and performs color mapping.

[0029] Streamlines and Particles: Virtual particles are created in the flow field and their trajectories over time are calculated to form dynamic streamlines, which visually display the water flow path and vortices.

[0030] Isosurface: Generates a three-dimensional isosurface for a specific value.

[0031] The engine merges the calculated graphic elements with the 3D terrain model to generate a complete image and output it to the display device.

[0032] 2) Analysis and Interactive Response: Users interact with the terminal through the interface, query the data and display the precise velocity value, depth, time and other information of the point, draw lines in the 3D scene to define a new measurement cross section, and then immediately calculate and pass the instantaneous flow rate of the cross section. By dragging the time axis, the terminal quickly loads the corresponding time data from the historical database and re-renders the scene. Then, by adjusting the system parameters, the terminal encapsulates this instruction into a control package, which is sent to the central processing unit through the communication network to control the sensing array.

[0033] 3) Alarm and report generation: The background continuously compares real-time data with preset rules. When an anomaly is detected (such as the flow rate in a certain area exceeding the safe value or a node losing connection), an alarm is immediately triggered. Daily and weekly reports can be automatically generated on the command terminal, or data for a specific period can be exported.

[0034] Example 2 In this embodiment, step S1 ensures that all distributed sensing units start up and operate under a unified spatiotemporal reference, providing a data source for data fusion. The specific steps include: 1. After the system powers on, the master ADCP node broadcasts a "network discovery" command to the network. Upon receiving the command, all online slave ADCP nodes and environment awareness modules register their IDs, types, and basic capabilities with the master node. The master node then constructs and maintains a real-time system topology map, clarifying the hierarchical relationships and network status of each node.

[0035] 2. The master control node encapsulates the specific task parameters into an instruction package according to the preset measurement scheme (such as cross-sectional measurement and fixed-point profile measurement), and sends it to all nodes through the communication network. The parameter content includes the transmission parameters: pulse length, transmission frequency, and blind zone setting. Beam parameters: beamforming mode, beam tilt angle; Sampling parameters: measurement start time, pulse repetition interval, number of beam cycles for each transmission sequence; 3. Achieving collaborative measurement includes: Reference synchronization: The master node acquires and maintains high-precision UTC time by using a built-in GPS disciplined clock or by receiving a master clock signal from the IEEE 1588 (PTP) precision time protocol.

[0036] Synchronization signal distribution: The master node distributes this clock signal to all slave nodes via wired (e.g., IEEE 1588) or wireless (e.g., periodic broadcast synchronization frames). For underwater nodes, synchronization sound signals can be sent via an underwater acoustic modem.

[0037] Slave clock correction: After receiving the synchronization signal, each slave node corrects its local clock based on the signal propagation delay (which can be calculated through pre-calibration or bidirectional timestamp exchange).

[0038] 4. At the preset measurement start time, all ADCP nodes simultaneously transmit acoustic pulses based on the synchronized clock and begin receiving and recording echo data, while the environmental parameter sensing module synchronously starts sampling.

[0039] 5. When each node generates a frame of raw data (including flow rate, echo intensity, correlation, etc.), it uses its calibrated local clock to stamp the frame of data with a high-precision timestamp. At the same time, the node's IMU and positioning data are also recorded and appended with the same time base.

[0040] Compensating for clock skew and drift between slave and master nodes in high-precision clock synchronization requires the use of a linear clock model and its correction algorithm, specifically including: 1. Clock skew model Assume the master node clock time is The node clock time is The relationship between them is modeled as follows:

[0041] in Use the master node clock time as a reference value; The local clock time of the slave node needs to be corrected; This is the clock drift factor, ideally... =1, if >1 indicates that the slave node's clock is faster than the master node's clock; if If the value is less than 1, it means that the slave node's clock is faster than the master node's clock. Clock offset is the fixed offset of the slave node clock relative to the master node clock.

[0042] 2. Bias Estimation and Correction Algorithm By exchanging timestamps between master and slave nodes, parameters a and b can be estimated using linear regression with the least squares method, specifically including: Data Acquisition: The master and slave nodes perform N (N≥2) timestamp exchanges. For the i-th exchange, the slave node's local time... When a synchronization request packet is sent, the master node records its own time upon receiving the packet. and immediately reply with a message containing and master node sending time The response packet, from the node at local time Upon receiving the response packet, for the i-th exchange, we obtain a data point: the node time. and the corresponding master node time or However, processing delays need to be considered.

[0043] The set of data points collected is defined as follows: Using the least squares method, the best straight line is fitted as follows: The slope and intercept The calculation formula is:

[0044]

[0045] Real-time correction: Once a and b are calculated, the node can perform the correction at any subsequent local time. All of them can be converted to the standard time synchronized with the master node using the following formula. , is represented as:

[0046] All collected data will be used As its timestamp; in The number of timestamp data pairs used for clock synchronization calculations. For the i-th timestamp exchange, the local time recorded by the node; This represents the standard time recorded by the master node during the i-th timestamp exchange.

[0047] Example 3 In this embodiment, step S2 involves preprocessing the multi-source data acquired from the cooperative sensing array and assigning an initial quality assessment label to each data unit. The specific steps include: 1. Receive raw data packets from different models of ADCP nodes and environmental sensors with different protocol formats, and extract core data fields according to predefined parsing rules, including: ADCP data: Radial velocity, echo intensity, signal correlation, beam pointing angle, etc. for each beam; IMU data: roll, pitch, and yaw angles; Location data: latitude and longitude coordinates; Environmental data: sound velocity, turbidity, etc.; Then output to a standardized data table with a unified format.

[0048] 2. Transform the velocity vector measured by each ADCP node, based on its own carrier coordinate system and beam coordinate system, to a unified geodetic coordinate system (East-North-Sky, ENU). Use a three-dimensional rotation matrix for coordinate transformation, as shown below:

[0049] in These represent the velocity components in the east, north, and sky directions, respectively, in the geodetic coordinate system. The radial velocity vector in the beam coordinate system. This is the rotation matrix from the carrier coordinate system to the geodetic coordinate system; this matrix consists of the attitude angles of the nodes (including roll). , swaying ,course The calculation yields the result, which is expressed as follows: ,in These are the basic rotation matrices around the x, y, and z axes, respectively.

[0050] 3. Based on signal quality filtering and the signal correlation provided by ADCP, a correlation threshold is set. The formula for determining outliers and wild values ​​is as follows:

[0051] in Let i be the signal correlation value of the i-th depth unit. The correlation threshold is set based on experience and the aquatic environment.

[0052] 4. Construct a multi-factor quality scoring model to calculate a comprehensive quality score for each valid data unit that passes the initial screening. This score quantifies the reliability of the data. The model is expressed as:

[0053] in This is a comprehensive quality score, ranging from 0 to 1; Signal-to-noise ratio (SNR) is a key feature; a higher SNR indicates better data quality. This means normalizing it to the interval [0,1]. This represents signal correlation; the higher the value, the better the data quality. This means normalizing it to the interval [0,1]. The turbulence intensity index is used; the higher the value, the greater the data is affected by turbulence and the worse the quality. Turbidity, if its value is too high, may mean that the sound waves are being over-scattered or absorbed, resulting in a decrease in data quality; , , , These are the weighting coefficients for each quality factor, and their sum is 1. The weights can be adjusted according to the characteristics of different aquatic environments.

[0054] 5. Based on the overall quality score Each data unit is labeled with a different quality level, represented as: High quality: ≥0.8, marked as Flag=1; Medium quality: 0.5≤ <0.8, marked as Flag=2; Low quality: <0.5, marked as Flag=3; Invalid data: Removed and marked with Flag=9.

[0055] After this step, we can obtain: Standardized data: All data are in a unified geodetic coordinate system and have a consistent format.

[0056] Clean dataset: Obvious outliers and low-reliability data are identified and labeled.

[0057] Data streams with quality labels: Each data unit is accompanied by a quantified quality score and quality level code.

[0058] Example 4 In this embodiment, in step S3, different flow velocity data are quantized and weighted to achieve optimal data fusion. The specific steps include: 1. Based on the geometric extent of the measurement area (e.g., river cross-section, reservoir water area), define a regular three-dimensional spatial grid, setting a fixed resolution for the grid in the horizontal direction (east-north) and the vertical direction (sky), and then output a data structure containing (…). , , A digital spatial framework of grid cells, ensuring that each cell has a unique index and center point coordinates.

[0059] 2. A multi-factor coupled weighting function model is used to calculate a weight value for fusion for each valid original data point. This weight comprehensively reflects the reliability of the data, geometric relationships, and environmental interference. Its model is represented as follows:

[0060] in The fusion weight of the i-th data point is such that the larger the value, the greater the contribution of that point in the fusion. The overall quality score; The exponential adjustment factor for the quality fraction (usually) ≥1), used to control the degree of advantage of high-quality data; The distance from the i-th data point to its target grid center is used to implement spatial constraints; the greater the distance, the lower the weight. The spatially relevant scale parameter determines how quickly the weight decays with distance, and is set according to the physical spatial correlation of the flow field; The incident angle is the angle between the beam direction and the zenith direction. The larger the incident angle, the greater the geometric error in the velocity measurement. The exponential adjustment factor for the incident angle (usually) ≥1), used to control the degree of penalty for large incident angle data; This is the turbulence intensity index. A higher value indicates that the data at that point is more severely contaminated by turbulent fluctuations, and therefore less reliable. This is a linear adjustment factor for turbulence intensity, used to control the degree to which turbulence suppresses the weights.

[0061] 3. Inverse distance weighted interpolation is used to convert each data point (with spatial coordinates and weights) into its corresponding data points. The data points are assigned to one or more spatial grid cells. For a given grid cell, its initial velocity estimate is determined by all data points within a certain search radius R around it. The contribution of each point is determined by its inverse distance to the grid center and its weight. The decision is made jointly, and its formula is expressed as:

[0062] in For the data points (velocity vectors) assigned to grid cell k and weight The set of ) Let i be the spatial coordinates of the i-th data point. Let the center coordinates of the k-th grid cell be... The search radius defines the spatial extent of the data points participating in the fusion of each grid cell. 4. Traverse all grid cells and generate the corresponding data set for each cell k. Then, a structured data volume is output, in which each grid cell is associated with a list containing all data points that match this cell and their corresponding calculated dynamic weights. If a grid cell contains no data points and is therefore an empty set, it is marked as a blind zone and will be filled by interpolation using data from the surrounding grid cells.

[0063] After this step, we can obtain: Structured spatial grid framework: The entire measurement area is discretized into a regular three-dimensional grid.

[0064] Weighted grid datasets: Each grid cell contains a set of data points from different sources, each with a dynamically fused weight.

[0065] Spatial matching relationship: Clarify the mapping relationship between the original data and the standardized output grid.

[0066] Example 5 In this embodiment, in step S4, a physical model is applied to systematically correct the data for deviations, including geometric positioning errors caused by changes in the sound velocity profile and velocity measurement errors caused by turbulence. The specific steps include: 1. Using real-time sound velocity profile data, establish a functional model of sound velocity variation with depth. Then, precise ray tracing is performed on the propagation path of each sound beam to correct for beam footprint positioning errors and depth calculation deviations caused by the non-uniform distribution of sound speed. The formula for calculating the sound ray trajectory based on Snell's law is as follows:

[0067] The numerical solution to the differential equations of the sound ray trajectory is as follows:

[0068]

[0069] in The velocity of sound at depth z is measured directly by a sound velocity profiler or calculated by a temperature-salinity-depth profiler. Let be the velocity of sound at the transducer surface (z=0). Let be the angle between the sound ray at depth z and the horizontal plane. The initial emission angle (beam tilt angle) when the acoustic beam leaves the transducer. This refers to the horizontal displacement of the vocal tract. This is the propagation time of the sound wave.

[0070] The numerical solution process is as follows: The water body from the transducer to the target depth is discretized into multiple thin layers, and the sound velocity is assumed to change linearly within each layer. Starting from z=0, it is known that and For each layer k, the angle at which the ray enters the next layer is calculated using Snell's law. Then we have:

[0071] Calculate the horizontal displacement of the sound ray in this layer and transmission time ; The calculation is iterated until the target depth is reached, and the total horizontal displacement x and total propagation time t of the sound ray are obtained by summing them up.

[0072] The true depth z and horizontal position x relative to ADCP of each depth cell are accurately calculated to correct for spatial matching. The original depth calculation based on the propagation time t and average sound speed assumptions is replaced by the results of this physical model.

[0073] 2. Using the turbulence intensity index, a simplified turbulence error model is established to drive an adaptive filter, suppressing turbulence fluctuations in the velocity data and extracting a more stable time-averaged velocity. The adaptive Kalman filter based on turbulence intensity is expressed as: State-space model: State equations (system model):

[0074] Observation equations (measurement model):

[0075] in Let k be the state vector of the system at time k. Here is the state transition matrix. The observation vector is the original velocity vector measured by ADCP. This is the observation matrix, which maps the state vector to the observation space. It is usually also the identity matrix I. The process noise represents the uncertainty of state evolution, and its covariance matrix is... ; Let be the observation noise, representing the measurement error, and let its covariance matrix be R.

[0076] Introducing an adaptive mechanism to address process noise covariance Make dynamic adjustments

[0077] in The basic process noise covariance represents the model uncertainty under calm water flow conditions; Let k be the turbulence intensity index at time k; The adjustment coefficient controls the degree of influence of turbulence intensity on process noise; Its adaptive correction mechanism is as follows , The basic observation noise covariance represents the measurement accuracy of ADCP under ideal conditions; Adaptive filtering is achieved based on the Kalman filter recursive formula, expressed as: predict: ,

[0078] renew:

[0079] ,

[0080] in This is for estimating the prior state at time k based on the state at time k-1; The posterior state estimate (i.e., the corrected flow velocity) is obtained by combining the observations at time k; P is the covariance matrix of the state estimate error. The Kalman gain determines the weight of the observation in the state update.

[0081] After this step, we can obtain: Geometrically corrected spatial coordinates of the data: The true three-dimensional position of each data point is corrected by ray tracing to eliminate systematic positioning errors caused by inaccurate sound speed assumptions.

[0082] Physically corrected velocity estimates: The velocity value of each data point is subjected to adaptive filtering based on turbulence intensity, which effectively suppresses the influence of turbulence fluctuations and provides a velocity estimate that is closer to the real situation.

[0083] Example 6 In this embodiment, in step S5, the corrected multi-source data with dynamic weights are optimally synthesized within each spatial grid cell to generate a three-dimensional full-domain water flow velocity field. The specific steps include: 1. Traverse each spatial grid cell k and examine its dataset. The effectiveness; 2. For each valid grid cell k, set its data set. All velocity vectors in the vectors are assigned their corresponding dynamic weights. The weighted average is calculated using the following formula:

[0084] in is the fused velocity vector of grid cell k, which is the final and optimal velocity estimate of that cell; For data sets The velocity vector of the i-th data point; is the dynamic fusion weight for the i-th data point.

[0085] 3. Fusion results for each grid cell To calculate an uncertainty measure to assess the reliability of the result, the formula for calculating the weighted standard deviation is:

[0086] 4. For grid cells marked as blind zones, spatial interpolation is performed using the fusion results of the surrounding effective grids to generate a physically reasonable continuous flow field.

[0087] After this step, we can obtain: High-precision three-dimensional velocity field: A three-dimensional velocity field that covers the entire measurement area and is regularly gridded, with each grid cell containing an optimally estimated velocity vector.

[0088] Data quality field: An uncertainty metric field corresponding to the velocity field, which intuitively reflects the credibility of the fusion results at different spatial locations.

[0089] Continuous flow profile: By interpolating and filling in the gaps, a flow profile map of the entire watershed is finally output.

[0090] Example 7 In this embodiment, in step S6, by analyzing the differences between the fusion result and the original data, the system diagnoses the performance status of each node and generates correction instructions to optimize the measurement process in real time. The specific steps include: 1. The obtained high-confidence fused flow field is used as the reference true value and compared with the original measurement value of each ADCP node to identify whether there are systematic, modelable biases; 2. Determine whether the calculated deviation is statistically significant, and thus decide whether to generate a feedback instruction; 3. Encapsulate the correction parameters to be sent into a compact, machine-readable instruction package to adapt to low-bandwidth, high-latency communication links, and send the generated feedback instructions to the target node with the highest priority through the communication network; 4. After receiving the instruction, the target node's embedded system parses and applies the new parameters.

[0091] After this step, the system achieves the following: Online self-calibration: The system can automatically identify and compensate for systematic deviations in nodes (such as installation errors or long-term drift) without manual intervention.

[0092] The measurement process is adaptive: the system can dynamically adjust the signal processing strategy of the nodes according to real-time hydrological conditions (such as enhanced turbulence).

[0093] Continuous performance optimization: Through a closed-loop optimization process of design awareness-analysis-adjustment, the entire system is continuously optimized and improved as runtime increases.

[0094] Example 8 In this embodiment, step S7 outputs the final correction data product to the user and manages the continuous operation cycle of the entire system, realizing dynamic tracking and adaptive optimization. The specific steps include: 1. Standardize and encapsulate the generated 3D fused velocity field and uncertainty field, as well as the related metadata; 2. Distribute the formatted data products to various users and systems in parallel through different channels; 3. The system automatically records key performance indicators within this cycle for system health monitoring and long-term performance analysis. The recorded content includes node data packet loss rate, signal-to-noise ratio statistics; differences in data before and after fusion; issuance and execution status of feedback commands; and processor computing load and communication latency. 4. The system can intelligently adjust the correction cycle according to the degree of change in water flow.

[0095] After this step, the system achieves the following: Standardized data product output: Output standardized three-dimensional water flow field data containing complete quality information in a time sequence; Multi-user real-time service: Provides real-time data services to users with different needs through multiple channels; System performance is traceable: complete operation logs provide data support for system maintenance and optimization; Intelligent dynamic iteration: The system runs continuously at the optimal sampling frequency.

[0096] In summary, this ADCP-based multi-beam collaborative flow profile dynamic correction system and method overcomes the blind spots of single-beam measurement by constructing a multi-beam ADCP collaborative measurement network to simultaneously collect flow velocity data at different azimuths and depths. Simultaneously, a dynamic correction algorithm is designed to combine real-time flow turbulence intensity and water body sound velocity profiles to perform real-time fusion and deviation correction of multi-beam data. Furthermore, a data feedback closed loop is established to transmit the corrected flow profile data back to the measurement terminal in real time, realizing dynamic iteration of measurement-correction-output. This improves the accuracy and timeliness of flow profile measurements in complex water areas (such as estuaries and reservoirs), thereby providing more reliable data support for water conservancy project scheduling and water environment monitoring.

[0097] The relevant modules involved in this system are all hardware system modules or functional modules that combine computer software programs or protocols with hardware in the prior art. The computer software programs or protocols involved in these functional modules are technologies known to those skilled in the art and are not improvements to this system. The improvement of this system lies in the interaction or connection between the modules, that is, in improving the overall structure of the system to solve the corresponding technical problems that this system aims to address.

[0098] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-beam cooperative dynamic flow profile correction system based on ADCP, characterized in that, This includes a multi-beam ADCP collaborative sensing array, a heterogeneous node network with a master-slave architecture, and redundant measurement and spatial complementarity of water flow profiles based on three-dimensional deployment and high-precision spatiotemporal synchronization. The environmental parameter sensing module integrates a synchronous sensing unit and acquires the acoustic characteristics and physical state of the water body in real time, providing physical driving parameters for dynamic correction. The dynamic correction and fusion module performs ray tracking correction, turbulence adaptive filtering, and mass-weighted vector fusion to achieve real-time conversion of multi-source heterogeneous data into a three-dimensional flow field. The data feedback and communication network, based on a closed-loop communication architecture with bidirectional transmission and priority scheduling, ensures online self-calibration and adaptive optimization of the system. The visualization terminal is a decision support system that integrates data management, 3D rendering, and interactive control, enabling immersive visualization and real-time interactive analysis of calibration results.

2. The multi-beam cooperative flow profile dynamic correction system based on ADCP according to claim 1, characterized in that, The multi-beam ADCP collaborative sensing array includes an array communication control unit, a cluster of sensor nodes, and an array power unit. The array communication control unit consists of a main control computer that runs the array control, a clock synchronizer that deploys the IEEE 1588 precision time protocol, and a central communication hub with integrated communication interfaces. The central communication hub is responsible for establishing bidirectional data links with each node. The sensor node cluster is divided into a master ADCP node deployed in the cross-section or the core area of ​​the water flow, and several slave ADCP nodes deployed in an asymmetric three-dimensional manner. The asymmetric three-dimensional deployment technology includes: 1) Near-shore nodes are deployed near both banks, and the beam can be tilted towards the center of the river to measure the bank flow velocity and complex backflow areas that are difficult to detect with traditional vertical beams; 2) Set up layered nodes anchored at different water depths, and the beam can be emitted horizontally or at an angle to measure vertical flow velocity shear and bottom flow velocity profile; 3) The bottom node is deployed on the bottom with the beam emitting upwards to compensate for the near-bottom blind zone measurement of the surface ADCP; The master ADCP node and the slave ADCP node adopt multi-beam ADCP and are configured with attitude and position reference subsystem, including: 1) using an inertial measurement unit (IMU) to measure the roll, pitch and heading angle data of each sensor node in real time, and converting the beam coordinate system velocity into the absolute geodetic coordinate system. 2) Use GNSS receivers for surface nodes to provide the absolute latitude and longitude coordinates of the nodes; 3) Use an underwater acoustic locator for underwater and bottom-sitting nodes, and determine the underwater coordinates by acoustic ranging with a reference source at a known position on the water surface; The array energy unit uses underwater buoys, buoys, or ship-mounted towed bodies as deployment platforms to build a power supply system with a hybrid architecture of shore power grid and solar power generation.

3. The multi-beam cooperative dynamic flow profile correction system and method based on ADCP according to claim 1, characterized in that, The environmental parameter sensing module includes a sound velocity profile measurement unit, an optical scattering sensing unit, a turbulence intensity quantification unit, and a data integration interface unit. The sound velocity profile measurement unit consists of a vertical chain sensor array composed of multiple sound velocity probes and a temperature-depth-salinity profiler. The optical scattering sensing unit includes an optical turbidity sensor composed of a turbidimeter and a laser particulate matter analyzer, which measures the light intensity scattered by suspended particles in the water body based on the emitted beam and evaluates the turbidity. It also includes a backscattering intensity monitoring module built into the ADCP, which is used to record the echo intensity of each depth unit. The turbulence intensity quantization unit includes an acoustic Doppler velocimeter for acquiring instantaneous flow velocity fluctuations and a turbulence inversion model for quantifying turbulence. The ADCP echo signal-to-noise ratio and velocity signal-to-noise ratio are input to the turbulence inversion model, and the spatial distribution of turbulence intensity is inverted based on the original ADCP data.

4. The multi-beam cooperative flow profile dynamic correction system based on ADCP according to claim 3, characterized in that, The dynamic correction and fusion module includes a data cache pool for the access data integration interface unit. The data cache pool stores the following data based on a first-in-first-out data queue: 1) Raw velocity data packets from each ADCP node; 2) Parameter matrices such as sound velocity profile, turbidity, and turbulence intensity index from the environmental perception module; 3) IMU attitude and positioning data from each node; The dynamic correction and fusion module also includes a correction module and a fusion module. The correction module is divided into a sound velocity path corrector, a turbulence adaptive filter and a mass weight dynamic calculator. The sound velocity path corrector takes the real-time sound velocity profile as input and uses the ray tracing method to simulate the bending propagation path of each sound beam in the real sound velocity profile. It outputs the real geometric position and beam incident angle of each depth unit to correct the depth deviation and flow velocity calculation error. The turbulence adaptive filter takes the real-time turbulence intensity index and the original flow velocity data as input, and outputs a flow velocity estimate optimized by turbulence error based on the adaptive low-pass filter. The quality weight dynamic calculator takes signal-to-noise ratio, echo intensity, turbulence index, turbidity and beam incidence angle as input, runs a weighting function to calculate the comprehensive quality weight value for each data point, and outputs weighting factors for subsequent data fusion. In the fusion module, for each grid node, all corrected velocity data points within its surrounding range are searched, and a weighted vector average fusion is performed using the optimal estimation fusion algorithm.

5. The multi-beam cooperative flow profile dynamic correction system based on ADCP according to claim 1, characterized in that, The real-time data feedback and communication network includes a sensing layer network connecting underwater and surface sensing nodes. The sensing layer network is divided into an underwater acoustic communication network, a short-range wireless mesh network, and a surface communication gateway. The underwater acoustic communication network integrates an underwater acoustic communication modem on each underwater ADCP node and environmental sensor to enable data transmission between underwater nodes and between underwater nodes and the surface gateway; the short-range wireless mesh network includes Wi-Fi and LoRa modules for surface nodes, which self-organize the nodes on the surface platform into a network and hop data to the gateway; the surface communication gateway is a shipborne computer integrating multiple network interfaces to collect data from all nodes. The real-time data feedback and communication network includes a decision-making layer network connecting node gateways and shore-based cloud. The decision-making layer network is divided into broadband wireless links and wired links. The broadband wireless links include 5G communication modules and wireless private networks, and the wired links are Ethernet. The real-time data feedback and communication network also includes a control layer network connecting the central processing unit and the sensor nodes. The control layer network includes an encoding and compression module and a communication channel for issuing correction instructions and control commands from the central processing unit.

6. The multi-beam cooperative flow profile dynamic correction system based on ADCP according to claim 1, characterized in that, The visualization terminal includes a functional engine layer and a human-computer interaction layer. The functional engine layer includes a scene management and rendering engine, which constructs a virtual underwater environment based on a 3D graphics engine and receives flow velocity data to generate a 3D water flow field visualization model. The functional engine layer also includes a data analysis and alarm module, which has a built-in algorithm library for providing streamline calculation, vorticity identification, cross-sectional flow integral and vertical average flow velocity calculation and analysis, and system alarms when real-time data triggers rules according to set thresholds. The human-computer interaction layer includes a multi-view collaborative interface and an interactive control panel. The multi-view collaborative interface is equipped with a three-dimensional dynamic flow field diagram that dynamically displays the direction, velocity and three-dimensional structure of the water flow, a two-dimensional cross-sectional diagram for quantitative analysis of flow velocity, a data quality heat map that displays the confidence level of cross-sectional area data, and an instrument status panel that displays the data of each ADCP node in real time. The interactive control panel is equipped with a time axis controller, a view controller and a layer controller.

7. A method for dynamic correction of multi-beam cooperative flow profiles based on ADCP, characterized in that, The specific steps include the following: S1. Based on the precision clock synchronization protocol, the distributed ADCP node array is triggered to perform synchronous beam transmission and data acquisition, and a unified spatiotemporal reference is established; S2. Perform coordinate transformation on the original flow velocity data of each node, and implement outlier removal and initial quality code assignment based on signal coherence and echo intensity; S3. Construct a weighting function to dynamically calculate the weights of multi-source data and interpolate and match them to a unified spatial grid; S4. Use real-time sound velocity profiles for ray tracking, correct beam geometry and depth deviations, and combine turbulence intensity index to drive an adaptive filter to suppress flow velocity pulsation noise. S5. Perform vector weighted fusion based on mass weights on a standardized spatial grid to generate a high-confidence velocity field with no blind spots across the entire domain; S6. Compare the fusion results with the original node data to identify systematic deviations, generate correction parameters and inject them down to the corresponding nodes to achieve online self-calibration of measurement parameters; S7. Output the final corrected flow profile and quality assessment report, and realize closed-loop dynamic optimization of measurement-correction-output.

Images