Method for calculating total tidal energy based on multi-source fusion data

Abstract

This invention relates to the field of electrical digital data processing technology and discloses a method for calculating the total tidal energy based on multi-source fusion data. The method includes: collecting multi-source observation data and geographic topology data; mapping the data to a three-dimensional voxel grid to construct a spatiotemporal velocity tensor; calculating the rate of change of the inner product of the grid velocity vectors; adaptively refining the grid resolution using an octree structure; determining the occlusion state of the grid addressing path based on the geographic topology data; establishing anisotropic completion constraints by setting the weights to zero; removing outlier sampling points by combining the tidal harmonic constant; and calculating the total tidal energy using the filled tensor. This invention couples geographic constraints to the tensor completion logic, eliminating the passivation defects in the velocity field simulation caused by abrupt topographic changes, and ensuring the physical fidelity of tidal energy calculation in large-scale ocean areas.

Description

Technical Field

[0001] This invention relates to a method for calculating the total tidal energy based on multi-source fusion data, belonging to the field of electrical digital data processing technology. Background Technology

[0002] The current calculation of total tidal energy involves data processing steps such as spatial interpolation and temporal alignment, which are used to convert discrete sampling points into dense flow field models, serving as the basis for grid dispatch and marine energy operation, and supporting grid-connected power allocation.

[0003] Large-scale marine observations are constrained by environmental and cost factors, resulting in sparse sensor deployment, discontinuous sampling data, hardware limitations, and deficiencies in software control methods. For example, Chinese invention patent application CN120182489A discloses a method for reconstructing three-dimensional ocean vortex flow fields based on physical constraints. This method constructs vertical mode functions of the vortex and superimposes sea surface and deep-water velocity constraints to recover the three-dimensional flow field. However, this scheme relies on ideal dynamics assumptions in open ocean areas, presupposes zero deep-water velocity, and extrapolates vertical modes based on climatological averages. In tidal energy calculation scenarios, the kinetic energy of the flow field exhibits anisotropy and nonlinear abrupt changes due to the influence of seabed topography and nearshore topology. Existing low-temperature tidal energy... The simplified first-order modal scheme cannot perceive the local jet and turbulence characteristics caused by abrupt topographic changes; the isotropic numerical diffusion logic cannot achieve momentum cutoff when facing physical obstacles such as seabed ridges and islands, resulting in feature passivation in numerical smoothing at topographic abrupt changes. The existing data processing architecture follows the isotropic principle, and the numerical recovery of missing nodes depends on the effective nodes in the neighborhood. The seabed topography constitutes the anisotropic boundary of fluid momentum. Numerical smoothing that ignores physical topological interference produces feature passivation at topographic abrupt changes. Since the total tidal energy is proportional to the cubic velocity modulus, the numerical dissipation in the data reconstruction stage produces error amplification in the spatial integration, which restricts the reliability of grid-connected power prediction.

[0004] Therefore, how to construct a data addressing logic with physical boundary constraints to achieve high-fidelity calculation of total power flow energy is the technical problem to be solved by this invention. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: A method for calculating the total tidal power based on multi-source fusion data, comprising the following steps: Step S1: Obtain multi-source observation data and geographic topology data of the sea area to be measured. The multi-source observation data includes fixed-point profile velocity data collected by a fixed acoustic Doppler current profiler, surface flow field height data collected by a satellite altimeter, and background field data generated by numerical simulation. Step S2: Map the multi-source observation data to a three-dimensional voxel grid, establish a discrete multidimensional data matrix characterizing the spatiotemporal features of the current field in the sea area to be measured, and construct the spatiotemporal velocity tensor. Step S3: Calculate the rate of change of the velocity vector inner product between adjacent grids in the three-dimensional voxel grid, and trigger adaptive refinement of the three-dimensional voxel grid based on the rate of change of the velocity vector inner product. The resolution of the local velocity gradient abrupt change region is refined by using an octree structure. Step S4: Determine whether there is geographic topological occlusion on the addressing path of adjacent 3D voxel grids based on geographic topological data. When geographic topological occlusion is determined to exist, reset the transmission weight between the addresses of adjacent 3D voxel grids to zero in order to construct anisotropic completion constraints for the spatiotemporal velocity tensor. Step S5: Extract the seabed topographic height using geographic topological data, and construct the background field variance envelope by combining it with the tidal harmonic constant, and remove outlier sampling points in the spatiotemporal velocity tensor that exceed the background field variance envelope. Step S6: Fill missing values ​​in the spatiotemporal velocity tensor after removing outlier sampling points based on anisotropic completion constraints, and calculate the total global tidal energy based on the filled spatiotemporal velocity tensor.

[0006] Preferably, the specific process of step S3 is as follows: obtain the gradient value of the inner product of the velocity vectors of adjacent voxel points in the three-dimensional voxel grid in the spatial grid dimension; determine whether the gradient value exceeds the preset mutation threshold; if the gradient value exceeds the mutation threshold, perform octree recursive partitioning on the three-dimensional voxel grid to divide it into 8 sub-voxel grids until the gradient value in the sub-voxel grid is not greater than the mutation threshold or reaches the preset maximum partitioning depth.

[0007] Preferably, the specific process of step S4 is as follows: extract the seabed topographic elevation coordinates from the geographic topology data and map them to blocking nodes in a three-dimensional voxel grid; when iteratively completing the spatiotemporal velocity tensor, determine the transmission weight W based on the topological connectivity state between adjacent three-dimensional voxel grids, and the determination logic of the transmission weight W follows the following formula: ,in, The propagation weight between the i-th voxel and the j-th voxel in the 3D voxel mesh; Represents the i-th voxel. With the j-th individual point The straight path between them passes through the blocking node. This indicates that the straight path does not pass through any blocking nodes; and These are the 3D mesh coordinates of the corresponding voxel points; σ is the preset kernel function width parameter.

[0008] Preferably, step S2 further includes: using an asynchronous spatial interpolation algorithm to align the fixed-point profile velocity data and the surface flow field height data to the same sampling frequency, and using multidimensional tensor algebra operations to map the unaligned spatiotemporal discrete data set to a discrete multidimensional data matrix.

[0009] Preferably, in step S4, the anisotropic completion constraint also includes a mass conservation constraint based on the fluid continuity operator. By transforming the mass conservation constraint into an algebraic addressing condition between three-dimensional voxel meshes, the iterative completion path of the spatiotemporal velocity tensor is restricted.

[0010] Preferably, the specific process of step S5 is as follows: retrieve the historical tidal harmonic constant of the sea area to be measured, and calculate the mean value of the background field of the flow field accordingly; calculate the real-time deviation between the real-time sampled value in the spatiotemporal velocity tensor and the mean value of the background field of the flow field; determine whether the real-time deviation falls within the range of the background field variance envelope obtained based on the statistical data of the same historical period; and remove outlier sampling points whose real-time deviation is outside the background field variance envelope.

[0011] Preferably, after calculating the total global tidal energy, the land area voxel nodes are addressed and masked using the filled spatiotemporal velocity tensor to construct a data quality control closed loop.

[0012] Preferably, the process for calculating the total global tidal energy is as follows: extracting the velocity component from the filled spatiotemporal velocity tensor; calculating the energy density within each three-dimensional voxel grid based on the seawater density parameter of the sea area to be measured; and performing numerical integration on the global three-dimensional voxel grid in both the time and spatial dimensions; wherein, the seawater density parameter is set to 1025 kg / m³. 3 .

[0013] Preferably, the geographic topological data includes seabed topographic elevation data and coastline vector data. By mapping the coastline vector data to the zero potential energy boundary of a three-dimensional voxel grid, numerical diffusion in the land area is blocked.

[0014] Preferably, the total global tidal energy is used for the site selection decision of the tidal energy generator. By comparing the energy distribution of different candidate grid areas within a set time period, the layout coordinates of the tidal energy generator in the three-dimensional voxel grid are determined.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In the calculation of total tidal energy of multi-source fusion data, a dynamic mask truncation mechanism based on elevation gradient is introduced during the tensor iteration completion process to solve the common isotropic diffusion error in electrical digital data processing. Spatial gradient vectors are extracted from static seabed topographic point cloud data and compared with the velocity feature vectors by performing a dot product comparison, thereby generating the corresponding zero-value mask factor at the memory addressing level. This mechanism physically truncates invalid data flow paths that cross seabed topographic obstacles, prevents low velocity region data from causing numerical pollution to the features of high kinetic energy jet regions, and ensures that the reconstructed global dense velocity tensor can accurately retain the local flow field distortion features excited by topographic abrupt changes, thereby avoiding distortion in spatial integration calculation. 2. By combining the iterative calculation method of fluid topology transmission weights, the physical constraint of mass conservation is directly transformed into algebraic addressing conditions between discrete voxel meshes, realizing deep coupling between physical priors and pure digital matrix completion. This processing method changes the logic of traditional interpolation algorithms that rely solely on geometric distance for numerical filling, making the tensor completion process controlled by the anisotropic boundary of fluid motion. By reducing the dimensionality of the complex four-dimensional nonlinear coupling optimization problem into a reduced-order algebraic iteration guided by asymmetric transmission weights, the existing processor resources in the system are used to realize the rapid reconstruction of the global flow field. While maintaining low computational overhead, the fidelity of the three-dimensional data model in energy accounting is improved. 3. The mesh adaptive splitting mechanism triggered by the rate of change of the inner product of the velocity vectors of adjacent voxels enables the system to have the ability to dynamically respond to local velocity gradients. When a sudden change in the velocity vector is detected in the spatial topology, the resolution of local voxels is improved by performing an octree splitting operation, which enables the system to capture subtle velocity gradient changes in strong turbulent regions. This synergistic effect of local refinement and global low-dimensional tensor completion avoids the smoothing distortion problem caused by fixed resolution meshes when dealing with complex fields such as island and reef wakes, and enhances the system's environmental perception sensitivity and modeling robustness under different marine topography. Attached Figure Description

[0016] Figure 1 This is the core flowchart for calculating the total tidal energy of multi-source fusion data in this invention; Figure 2 This is a block diagram of the tensor anisotropic completion logic of the present invention, which integrates the seabed geographic topology.

[0017] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0019] A method for calculating total tidal current energy based on multi-source fusion data includes the following steps: Step S1: Obtain multi-source observation data and geographic topology data of the sea area to be measured. The multi-source observation data includes fixed-point profile velocity data collected by a fixed acoustic Doppler current profiler, surface flow field height data collected by a satellite altimeter, and background field data generated by numerical simulation. Step S2: Map the multi-source observation data to a three-dimensional voxel grid, establish a discrete multidimensional data matrix characterizing the spatiotemporal features of the current field in the sea area to be measured, and construct the spatiotemporal velocity tensor. Step S3: Calculate the rate of change of the velocity vector inner product between adjacent grids in the three-dimensional voxel grid, and trigger adaptive refinement of the three-dimensional voxel grid based on the rate of change of the velocity vector inner product. The resolution of the local velocity gradient abrupt change region is refined by using an octree structure. Step S4: Determine whether there is geographic topological occlusion on the addressing path of adjacent 3D voxel grids based on geographic topological data. When geographic topological occlusion is determined to exist, reset the transmission weight between the addresses of adjacent 3D voxel grids to zero in order to construct anisotropic completion constraints for the spatiotemporal velocity tensor. Step S5: Extract the seabed topographic height using geographic topological data, and construct the background field variance envelope by combining it with the tidal harmonic constant, and remove outlier sampling points in the spatiotemporal velocity tensor that exceed the background field variance envelope. Step S6: Fill missing values ​​in the spatiotemporal velocity tensor after removing outlier sampling points based on anisotropic completion constraints, and calculate the total global tidal energy based on the filled spatiotemporal velocity tensor.

[0020] Preferably, the specific process of step S3 is as follows: obtain the gradient value of the inner product of the velocity vectors of adjacent voxel points in the three-dimensional voxel grid in the spatial grid dimension; determine whether the gradient value exceeds the preset mutation threshold; if the gradient value exceeds the mutation threshold, perform octree recursive partitioning on the three-dimensional voxel grid to divide it into 8 sub-voxel grids until the gradient value in the sub-voxel grid is not greater than the mutation threshold or reaches the preset maximum partitioning depth.

[0021] Preferably, the specific process of step S4 is as follows: extract the seabed topographic elevation coordinates from the geographic topology data and map them to blocking nodes in a three-dimensional voxel grid; when iteratively completing the spatiotemporal velocity tensor, determine the transmission weight W based on the topological connectivity state between adjacent three-dimensional voxel grids, and the determination logic of the transmission weight W follows the following formula: ,in, The propagation weight between the i-th voxel and the j-th voxel in the 3D voxel mesh; Represents the i-th voxel. With the j-th individual point The straight path between them passes through the blocking node. This indicates that the straight path does not pass through any blocking nodes; and These are the 3D mesh coordinates of the corresponding voxel points; σ is the preset kernel function width parameter.

[0022] Preferably, step S2 further includes: using an asynchronous spatial interpolation algorithm to align the fixed-point profile velocity data and the surface flow field height data to the same sampling frequency, and using multidimensional tensor algebra operations to map the unaligned spatiotemporal discrete data set to a discrete multidimensional data matrix.

[0023] Preferably, in step S4, the anisotropic completion constraint also includes a mass conservation constraint based on the fluid continuity operator. By transforming the mass conservation constraint into an algebraic addressing condition between three-dimensional voxel meshes, the iterative completion path of the spatiotemporal velocity tensor is restricted.

[0024] Preferably, the specific process of step S5 is as follows: retrieve the historical tidal harmonic constant of the sea area to be measured, and calculate the mean value of the background field of the flow field accordingly; calculate the real-time deviation between the real-time sampled value in the spatiotemporal velocity tensor and the mean value of the background field of the flow field; determine whether the real-time deviation falls within the range of the background field variance envelope obtained based on the statistical data of the same historical period; and remove outlier sampling points whose real-time deviation is outside the background field variance envelope.

[0025] Preferably, after calculating the total global tidal energy, the land area voxel nodes are addressed and masked using the filled spatiotemporal velocity tensor to construct a data quality control closed loop.

[0026] Preferably, the process for calculating the total global tidal energy is as follows: extracting the velocity component from the filled spatiotemporal velocity tensor; calculating the energy density within each three-dimensional voxel grid based on the seawater density parameter of the sea area to be measured; and performing numerical integration on the global three-dimensional voxel grid in both the time and spatial dimensions; wherein, the seawater density parameter is set to 1025 kg / m³. 3 .

[0027] Preferably, the geographic topological data includes seabed topographic elevation data and coastline vector data. By mapping the coastline vector data to the zero potential energy boundary of a three-dimensional voxel grid, numerical diffusion in the land area is blocked.

[0028] Preferably, the total global tidal energy is used for the site selection decision of the tidal energy generator. By comparing the energy distribution of different candidate grid areas within a set time period, the layout coordinates of the tidal energy generator in the three-dimensional voxel grid are determined.

[0029] Example 1: In a real-time dispatch scenario for grid-connected power generation of tidal energy in a marine area including island reefs and submarine canyons, the coastal power grid dispatch center faces dual data processing limitations: a mismatch in spatiotemporal resolution of multi-source observation data and nonlinear flow field distortion caused by terrain. Fixed acoustic Doppler current profilers are sparsely deployed and their sampling frequencies are asynchronously misaligned with those of satellite altimeters. Conventional data processing systems employ isotropic smooth diffusion algorithms based on Euclidean geometric distance, which diffuse low-velocity data from one side of the terrain obstruction to the high-velocity jet region on the other side. This results in feature passivation in the generated global three-dimensional flow field model at abrupt terrain changes, and further issues in subsequent cubic voxel space integration calculations. The system's processor acquires multi-source observation data and geographic topology data of the sea area under test. Using an asynchronous spatial interpolation algorithm, it aligns the fixed-point profile velocity data and surface flow field height data to the same sampling frequency. Through multidimensional tensor algebra operations, it maps these data to a three-dimensional voxel grid constructed based on the spatial coordinates of the seabed topographic point cloud data, generating an initial state spatiotemporal velocity tensor containing missing nodes. The system's multi-source observation data alignment process relies on historical tidal phase scales to perform temporal assimilation operations, extracting the historical tidal level variation curves of the sea area under test over a continuous year to obtain the high and low tide times of the main and secondary tides, constructing a reference phase time axis. The processor retrieves the absolute sampling timestamps of fixed-point profile velocity data and surface flow field height data, respectively. Based on the nearest neighbor matching rule, it maps and projects these two sets of heterogeneous timestamps onto a unified phase coordinate interval of the reference phase time axis. Using a cubic spline interpolation function, it resamples and calculates the discrete velocity and height sampling points projected onto the same phase interval, generating a three-dimensional voxel mesh initialization node sequence with a unified step size and aligned timestamps. It handles data sampling misalignment caused by differences in sensor physical revisit periods. The system retrieves the historical tidal harmonic constant of the sea area under test to calculate the mean value of the flow field background. It calculates the real-time deviation between the real-time sampled value in the spatiotemporal velocity tensor and the mean value of the flow field background, and discards the deviations. Outlier sampling points outside the background field variance envelope obtained from historical data are marked as missing. Before filling missing values, the seabed elevation coordinates in the geographic topology data are extracted and mapped to blocking nodes in the three-dimensional voxel grid. When geographic topological occlusion exists on the addressing path of adjacent three-dimensional voxel grids based on the geographic topology data, i.e., when the straight path between adjacent voxel nodes passes through the blocking node, the system forces the transmission weight between the addresses of adjacent three-dimensional voxel grids to be reset to zero in the underlying memory data structure, generates an asymmetric topological transmission weight tensor containing a zero-value mask factor, and constructs the anisotropic completion constraint of the spatiotemporal velocity tensor.

[0030] For connected voxel grids that do not pass through blocking nodes, the system quantifies the conduction relationship of fluid parameters based on the spatial distance attenuation mechanism, and the processor of the scheduling system extracts the spatial coordinates of the i-th voxel point in the three-dimensional voxel grid. and the spatial coordinates of the j-th voxel. The square of the Euclidean distance between these two spatial coordinates is calculated, divided by the square of twice the smoothing scale parameter σ, and then the negative exponent of the natural logarithm is taken to generate a spatial attenuation factor. The transmission weight is determined by the overall topological connectivity of the system. Transmission weight The calculation follows the formula ,in, The propagation weight between the i-th voxel and the j-th voxel in the 3D voxel mesh. This indicates that the straight-line path between the i-th voxel and the j-th voxel passes through the blocking node. This indicates that the straight path between the i-th and j-th voxels does not pass through any blocking nodes. The smoothing scale parameter σ is equal to the geometric mean of the deployment spacing of multiple fixed acoustic Doppler current profilers within the tested sea area. The system integrates local spatial attenuation values ​​and discrete Boolean topological masks to physically truncate data smoothing links crossing terrain obstacles while maintaining the continuity of fluid kinetic energy transmission. It outputs an asymmetric topological transmission weight tensor containing seabed geometric boundary constraints. Combining this with the mass conservation constraint transformed into an algebraic addressing condition, the system uses this asymmetric topological transmission weight tensor to limit the geometric boundary of numerical diffusion. It performs a reduced-order algebraic completion iteration on the spatiotemporal velocity tensor after removing outlier sampling points, and performs physical... To address invalid data smoothing links crossing seabed topographic obstacles, the system invokes the alternating least squares algorithm framework during the reduced-order algebraic completion iteration. The asymmetric topological propagation weight tensor is synchronously injected into the gradient update process of the alternating least squares algorithm as a mask matrix. The system establishes a segmented weight mapping rule in the underlying addressing logic. When a straight path between voxels i and j crosses an obstruction node, the weight is set to zero; otherwise, the squared Euclidean distance between the two points is extracted, divided by twice the squared smoothing scale parameter, and the negative exponent of the natural logarithm is used as the weight value. During each iteration of the alternating least squares algorithm, the processor updates the spatiotemporal velocity tensor eigenvector update matrix and mask matrix element by element. The dot product operation forces the addressing gradient step size to zero in the direction of crossing the seabed topographic obstacle, truncates the numerical diffusion path of heterogeneous velocity characteristics on both sides of the topography, and transforms the mass conservation constraint of the fluid continuity operator into a local flux algebraic sum inspection procedure in the discrete grid domain. The processor extracts the three-dimensional velocity components of the central three-dimensional voxel grid and the six orthogonally adjacent sub-voxel grids in the iterative completion state, and calculates the algebraic sum of the velocity differences in the three spatial dimensions using the central difference scheme. The system extracts a preset mass conservation tolerance threshold set to 5% of the mean value of the flow field background field. When the algebraic sum of the velocity differences exceeds the mass conservation tolerance threshold, the processor determines that the current iteration path generates numerical source and sink anomalies and dynamically increases the objective function of the alternating least squares algorithm. The proportional penalty term has the following control logic: a mass conservation penalty weight with an initial value of 0.1 is preset in the processor memory. Whenever the algebraic sum of velocity differences calculated by the central difference scheme is detected to exceed the mass conservation tolerance threshold of 1.0%, the penalty weight is automatically incremented by 0.05. The maximum limit for this accumulation process is 2.0. The processor uses the penalty weight accumulated in real time as a coefficient to directly multiply into the step size factor of the gradient update matrix, forcibly suppressing numerical fluctuations that do not meet physical continuity. The transmission weight between adjacent voxels is reduced step by step according to the iteration rounds until the algebraic sum of velocity differences converges to the global velocity field spatial distribution after constraint completion within the mass conservation tolerance threshold range.

[0031] The system acquires the gradient value of the velocity vector dot product of adjacent voxels within a 3D voxel grid along the spatial grid dimension. When this gradient value exceeds a preset abrupt change threshold, the system performs octree recursive partitioning of the 3D voxel grid until the gradient value within the sub-voxel grid is no greater than the abrupt change threshold or reaches a preset maximum partitioning depth. This is based on the velocity vector dot product change rate triggering adaptive mesh refinement and anisotropic completion constraints to form a positive feedback loop in data flow. Strong velocity gradients captured by local high-density grids are preserved by asymmetric masks and guide global tensor low-rank addressing. The system utilizes the processor to perform dot product and local spatial partitioning based on conditional masks, solving the problems of high-fidelity 3D flow field reconstruction accuracy and electronic / digital integration. To address the engineering conflict caused by limited system computing resources, the velocity component in the filled spatiotemporal velocity tensor was extracted. The energy density within each three-dimensional voxel grid was calculated using a seawater density parameter of 1025 kg / m³. By mapping the coastline vector data to the zero potential energy boundary of the three-dimensional voxel grid, numerical diffusion in the land area was blocked. Numerical integration was performed on the global three-dimensional voxel grid in both time and space dimensions, outputting large-scale calculation data of the total tidal energy of the ocean area, which includes local distortion characteristics. Based on this calculation data, the dispatch center compared the energy distribution of different candidate grid areas within a set time period, output the deployment coordinates of the tidal power generator units in the three-dimensional voxel grid, and issued grid connection and peak shaving commands.

[0032] Example 2: Addressing the engineering problem of numerical dissipation caused by abrupt topographic changes in the reconstruction of a three-dimensional tidal energy flow field in large-scale ocean areas, an experimental environment was constructed to verify the three-dimensional flow field reconstruction logic. A dataset of measured Doppler velocity profiles from the South China Sea's exclusive island and reef areas was extracted from the National Marine Science Data Center. Fluid dynamic simulation flow field data obtained by solving the incompressible three-dimensional Navier-Stokes equations was used as the baseline. The simulation spatial resolution was set to 5m, and the time step to 0.1s. Gaussian white noise with a signal-to-noise ratio of 15.2dB was superimposed on the fixed-point profile velocity data to simulate acoustic scattering interference from suspended marine particles. A mutation threshold for octree recursive partitioning was set, which balances the accuracy of flow field spatial feature restoration with the memory consumption of three-dimensional voxel mesh addressing. The discrete Laplace operator of the velocity vector inner product is used as the criterion. When the spatial characteristic wavelength corresponding to the local kinetic energy change rate is less than the Nyquist sampling interval of the current three-dimensional voxel mesh, the mutation threshold is reduced and the subdivision depth is increased. Based on this constraint model, for the velocity range distribution from 0.5 m / s to 2.5 m / s, the preferred value of the mutation threshold is determined to be 0.45. The experiment sets up the sample group of this invention and four control groups. Control group one adopts the isotropic spatial inverse distance weighted interpolation algorithm; control group two introduces the asymmetric topological transmission weight tensor and turns off the octree adaptive refinement module; control group three sets the mutation threshold to 0.10 to define the lower limit limit condition; control group four sets the mutation threshold to 0.85 to define the upper limit limit condition.

[0033] Multi-source observation data with superimposed noise were input, the mean value of the background flow field was calculated, and the variance envelope of the background field was constructed. The processor identified discrete noise pulses with an amplitude of up to 3.85 m / s, and the memory addresses of the corresponding outlier sampling points were cleared. The data flow status of the intermediate layer on both sides of the seabed ridge obstacle model was monitored. In the baseline ground truth, the measured velocity on the water-facing side was 2.46 m / s, and the measured velocity on the obstructed back side was 0.34 m / s. The interpolation results output from the control group showed the velocity components on the water-facing and back sides. The average velocity reached 1.41 m / s, exhibiting geometric diffusion dissipation. In control group 2, the addressing path of adjacent three-dimensional voxel grids was determined to pass through the ridge blocking node. The transmission weight W was set to zero, and the output flow velocity on the upstream side was 2.23 m / s, and the flow velocity on the downstream side was 0.42 m / s. The calculated kinetic energy integral error measurement value was 8.6%. Under the constraint of setting the weight to zero, the sample group of this invention identified that the gradient value of the inner product of the velocity vector of adjacent voxel points at the top of the ridge reached 0.63, which is greater than the mutation threshold of 0.45, triggering the octree recursive partitioning.

[0034] The present invention generates a high-density sub-voxel mesh around the ridge, and the asymmetric mask preserves the strong velocity gradient characteristics. The output velocity component on the upstream side is 2.41 m / s, and the velocity component on the downstream side is 0.37 m / s. The absolute error converges to 2.2%. In contrast, the control group's three-mutation threshold of 0.10 leads to over-subdivision of the velocity-smooth region, and the velocity reconstruction error converges to 1.8%. The measured time for matrix iteration solution increases nonlinearly from 1.4s in the present invention's sample to 18.6s, triggering the memory overflow protection threshold. The control group's four-mutation threshold of 0.85 does not respond to the initial kinetic energy change at the terrain edge, resulting in the loss of local jet characteristics and a jump in kinetic energy integration error to 14.3%. Experimental data confirm that the asymmetric topological transmission weight tensor containing the zero-value mask factor truncates invalid data smoothing links across obstacles, the collaborative mesh adaptive refinement module forms a positive feedback for data flow, and the selected mutation threshold range constitutes an engineering window to maintain high-fidelity three-dimensional flow field reconstruction accuracy and computing resource stability. The total tidal energy calculation data output by global integration eliminates the power prediction deviation caused by nonlinear flow field distortion.

[0035] Example 3: In the real-time dispatching of tidal energy grid-connected power generation covering the complete cycle of spring tides and neap tides, the coastal power grid dispatching center faces technical limitations in processing long-term multi-source observation data, namely, grid refinement scale control and heterogeneous grid integration overlap. The fixed threshold for the rate of change of the velocity vector inner product leads to excessive subdivision of the three-dimensional voxel grid under the high kinetic energy state during spring tides, increasing the memory address addressing load. Under the low kinetic energy state during neap tides, the velocity shear layer at the topographic edge is missed. When the processor of the dispatching system aggregates and sums the heterogeneous sub-voxel grids generated by the octree recursive partitioning, it adopts the geometric volume weight mapping rule across the level boundary to prevent the repeated calculation of energy in the spatial numerical integration stage.

[0036] The processor acquires multi-source observation data from a fixed acoustic Doppler current profiler and a satellite altimeter, extracts the real-time astronomical tide amplitude and periodic average amplitude from the historical tidal harmonic constant, calculates the quotient of the real-time astronomical tide amplitude and the periodic average amplitude, and generates a dynamic tidal adjustment factor. The system acquires a preset benchmark mutation threshold, multiplies the benchmark mutation threshold by the dynamic tidal adjustment factor, and generates a mutation threshold applicable to the current sampling time window. The processor calculates the rate of change of the velocity vector inner product between adjacent grids in the three-dimensional voxel grid and compares it with the mutation threshold. When the rate of change of the velocity vector inner product is greater than the mutation threshold, the system triggers adaptive refinement of the three-dimensional voxel grid. The processor adjusts the mutation threshold according to the tidal amplitude, increases the grid refinement trigger benchmark during the stage of increased flow field kinetic energy, and controls the computational load of the algorithm addressing.

[0037] After establishing anisotropic completion constraints and completing adaptive refinement, the system generates a spatiotemporal velocity tensor containing multi-level sub-voxel grids. To calculate the total global tidal energy, the processor extracts the geometric side length of each sub-voxel grid, calculates the corresponding three-dimensional spatial volume scalar, extracts the velocity vector at the geometric center of each sub-voxel grid, calculates the kinetic energy density at the geometric center based on the seawater density of 1025 kg / m³, multiplies the kinetic energy density by the corresponding three-dimensional spatial volume scalar, and generates the local total energy value of the sub-voxel grid. The processor traverses the effective voxel addresses in the spatiotemporal velocity tensor in ascending order according to the octree hierarchy depth, sequentially accumulates the local total energy value of each sub-voxel grid, and outputs the large-scale ocean tidal energy total calculation data. This voxel-level scalar mapping and ascending accumulation operation defines the mass conservation integral boundary of the cross-level heterogeneous grid space. The scheduling center determines the deployment coordinates of the tidal power generator units in the three-dimensional voxel grid and issues grid connection and peak shaving commands based on the global tidal energy total calculation data.

[0038] Example 4: In the initial calibration of deploying the total tidal energy calculation system to a specific sea area, to address the hardware limitation of the mismatch between the computational cost of mesh refinement and the physical memory capacity of the server, the processor of the scheduling system extracts the preset flow field step test vector and injects it into the initialized three-dimensional voxel mesh. The system performs octree recursive partitioning for the boundary of the abrupt change in velocity gradient caused by the flow field step test vector, reads the memory addressing increment scalar generated by mesh refinement in the underlying register, and compares the accumulated memory addressing increment scalar with the preset system safe memory threshold. When the memory addressing increment scalar reaches the system safe memory threshold, the processor extracts the current octree partitioning level index value, writes the octree partitioning level index value into the underlying configuration file, and sets it as the maximum partitioning depth for controlling the adaptive refinement of the spatiotemporal velocity tensor.

[0039] The system imports a historical continuous multi-source acoustic detection baseline data stream with a preset period into a three-dimensional voxel mesh with a set maximum subdivision depth. The processor calculates the moving average and discrete standard deviation of the temporal velocity vector for each independent sub-voxel mesh node. The system constructs a spatially heterogeneous static reference variance matrix based on the discrete standard deviation of each node, extracts the extreme boundary of the static reference variance matrix at the spatial coordinates of each node, maps and stores it as the background field variance envelope. The background field variance envelope output by this initial calibration procedure, together with the maximum subdivision depth, constitutes the technical criteria used to remove outlier sampling points and constrain the mesh refinement boundary during the real-time data acquisition and calculation stage.

[0040] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0041] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for calculating total tidal current energy based on multi-source fusion data, characterized in that, Includes the following steps: Step S1: Obtain multi-source observation data and geographic topology data of the sea area to be measured. The multi-source observation data includes fixed-point profile velocity data collected by a fixed acoustic Doppler current profiler, surface flow field height data collected by a satellite altimeter, and background field data generated by numerical simulation. Step S2: Map the multi-source observation data to a three-dimensional voxel grid, establish a discrete multidimensional data matrix characterizing the spatiotemporal features of the current field in the sea area to be measured, and construct the spatiotemporal velocity tensor. Step S3: Calculate the rate of change of the velocity vector inner product between adjacent grids in the three-dimensional voxel grid, and trigger adaptive refinement of the three-dimensional voxel grid based on the rate of change of the velocity vector inner product. The resolution of the local velocity gradient abrupt change region is refined by using an octree structure. Step S4: Determine whether there is geographic topological occlusion on the addressing path of adjacent 3D voxel grids based on geographic topological data. When geographic topological occlusion is determined to exist, reset the transmission weight between the addresses of adjacent 3D voxel grids to zero in order to construct anisotropic completion constraints for the spatiotemporal velocity tensor. Step S5: Extract the seabed topographic height using geographic topological data, and construct the background field variance envelope by combining it with the tidal harmonic constant, and remove outlier sampling points in the spatiotemporal velocity tensor that exceed the background field variance envelope. Step S6: Fill missing values ​​in the spatiotemporal velocity tensor after removing outlier sampling points based on anisotropic completion constraints, and calculate the total global tidal energy based on the filled spatiotemporal velocity tensor.

2. The method for calculating total tidal power based on multi-source fusion data according to claim 1, characterized in that, The specific process of step S3 is as follows: obtain the gradient value of the inner product of the velocity vectors of adjacent voxel points in the three-dimensional voxel grid in the spatial grid dimension; determine whether the gradient value exceeds the preset mutation threshold; if the gradient value exceeds the mutation threshold, perform octree recursive partitioning on the three-dimensional voxel grid to divide it into 8 sub-voxel grids until the gradient value in the sub-voxel grid is not greater than the mutation threshold or reaches the preset maximum partitioning depth.

3. The method for calculating total tidal power based on multi-source fusion data according to claim 1, characterized in that, The specific process of step S4 is as follows: extract the seabed topographic elevation coordinates from the geographic topology data and map them to blocking nodes in a three-dimensional voxel grid; when iteratively completing the spatiotemporal velocity tensor, determine the transmission weight W based on the topological connectivity state between adjacent three-dimensional voxel grids. The determination logic of the transmission weight W follows the following formula: ,in, The propagation weight between the i-th voxel and the j-th voxel in the 3D voxel mesh; Represents the i-th voxel. With the j-th individual point The straight path between them passes through the blocking node. This indicates that the straight path does not pass through any blocking nodes; and These are the 3D mesh coordinates of the corresponding voxel points; σ is the preset kernel function width parameter.

4. The method for calculating total tidal power based on multi-source fusion data according to claim 1, characterized in that, Step S2 further includes: using an asynchronous spatial interpolation algorithm to align the fixed-point profile velocity data and the surface flow field height data to the same sampling frequency, and using multidimensional tensor algebra operations to map the unaligned spatiotemporal discrete data set to a discrete multidimensional data matrix.

5. The method for calculating total tidal power based on multi-source fusion data according to claim 1, characterized in that, In step S4, the anisotropic completion constraint also includes the mass conservation constraint based on the fluid continuity operator. By transforming the mass conservation constraint into an algebraic addressing condition between three-dimensional voxel meshes, the iterative completion path of the spatiotemporal velocity tensor is restricted.

6. The method for calculating total tidal power based on multi-source fusion data according to claim 1, characterized in that, The specific process of step S5 is as follows: retrieve the historical tidal harmonic constant of the sea area to be measured, and calculate the mean value of the background field of the flow field accordingly; calculate the real-time deviation between the real-time sampled value in the spatiotemporal velocity tensor and the mean value of the background field of the flow field; determine whether the real-time deviation falls within the range of the background field variance envelope obtained based on the statistical data of the same historical period; and remove outlier sampling points whose real-time deviation is outside the background field variance envelope.

7. The method for calculating total tidal power based on multi-source fusion data according to claim 1, characterized in that, After calculating the total global tidal energy, the land area voxel nodes are addressed and masked using the filled spatiotemporal velocity tensor to construct a closed loop for data quality control.

8. The method for calculating total tidal power based on multi-source fusion data according to claim 1, characterized in that, The process of calculating the total global tidal energy is as follows: extract the velocity component from the filled spatiotemporal velocity tensor; and calculate the energy density within each three-dimensional voxel grid by combining the seawater density parameters of the sea area to be measured. Numerical integration was performed on the global three-dimensional voxel mesh in both the time and spatial dimensions; the seawater density parameter was set to 1025 kg / m³. 3 .

9. The method for calculating total tidal power based on multi-source fusion data according to claim 1, characterized in that, Geographic topology data includes seabed topographic elevation data and coastline vector data. By mapping the coastline vector data to the zero potential energy boundary of a three-dimensional voxel grid, numerical diffusion in terrestrial areas is blocked.

10. The method for calculating total tidal power based on multi-source fusion data according to claim 1, characterized in that, The total global tidal energy is used for the site selection decision of tidal power generators. By comparing the energy distribution of different candidate grid areas within a set time period, the layout coordinates of the tidal power generators in the three-dimensional voxel grid are determined.

Images