A method, device, terminal and storage medium for predicting stratified flow velocity in tidal river sections

By constructing the average flow rate and stratified ratio convolution prediction model of the tidal-induced river section, combined with historical data and topographic information, the problem of insufficient prediction ability of the tidal-induced river section in the existing technology is solved, and the construction ship is realized safety improvement and flow rate warning.

CN118917218BActive Publication Date: 2024-12-03TIANJIN PORT ENG INST LTD OF CCCC FIRST HARBOR ENG +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411404494.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-10-10
Publication Date
2024-12-03
Estimated Expiration
2044-10-10

AI Technical Summary

Technical Problem

The existing hydrological data prediction model of tide-induced rivers lacks the ability to predict stratified flow velocity in the tide-induced river section, which affects the construction safety of construction ships.

Method used

By obtaining the flow rate data, water level data and topographic data of the actual measured engineering office over the years, the average flow rate results are calculated using the two-dimensional river digital model, and an average flow rate forecast model and a hierarchical ratio convolution prediction model are constructed, combining the three-dimensional river mathematical model to invert scores Laminar flow rate realizes multi-point forecast of flow rate during the preset forecast period.

Benefits of technology

Accurate prediction of the stratified flow rate of the tide-induced river section is achieved, the safety and forecast speed of construction ships are improved, real-time flow rate warning is provided, and safety during construction is enhanced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118917218B_ABST
    Figure CN118917218B_ABST
Patent Text Reader

Abstract

The present invention provides a method, device, terminal and storage medium for predicting stratified flow velocity in a tidal river section, wherein the method comprises obtaining the flow velocity data, water level data and upstream and downstream river topography data of the engineering site measured in previous years, and using a two-dimensional river digital model to estimate the flow process of the engineering site based on the flow velocity data, water level data and upstream and downstream river topography data of the engineering site measured in previous years, and obtaining the average flow velocity results of the engineering site measured in previous years. The method, device, terminal and storage medium for predicting stratified flow velocity in a tidal river section described in the present invention, in the process of estimation, a total of 3 types of machine learning models are nested and run, different parameters are solved at different stages, and finally the flow field distribution of the engineering site is predicted, which has the advantage of high accuracy, can accurately predict the stratified flow velocity in the tidal river section, and is conducive to improving the safety of construction ships.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of computing models, and in particular relates to a method, a device, a terminal and a storage medium for predicting layered flow velocity in a tidal river section. Background Art

[0002] During the construction of immersed tubes in tidal river sections, it is necessary to calculate the flow velocity process of the project over the years based on the existing historical data, and select the appropriate construction period based on the flow velocity process. It takes more than ten years of flow velocity data to calculate the construction period. During the construction process, it is necessary to predict the flow velocity in the next few days, provide flow velocity warnings for construction ships, and give the flow velocity distribution range in the river channel to provide risk avoidance area divisions for ships.

[0003] The existing Chinese invention patent with application number 202410018108.7 discloses a method and system for processing hydrological data of tidal rivers. The method comprises the following steps: collecting real-time hydrological data of tidal rivers to obtain real-time hydrological data of tidal rivers, wherein the real-time hydrological data of tidal rivers include tidal data, tidal flow data and river turbidity data, wherein the river turbidity data includes multi-vertical river turbidity data of corresponding sections of each section of the upper, middle and lower reaches of the tidal river; performing rationality check on the tidal data to obtain the first tidal process line data; constructing an adaptive Kalman prediction model, and using the adaptive Kalman prediction model to perform difference supplement on the first tidal process line data to obtain the second tidal process line data. The present invention provides accurate hydrological data of tidal rivers and generates a hydrological data analysis report, providing important support for the management and protection of tidal rivers.

[0004] However, existing hydrological data prediction models for tidal rivers generally lack the ability to predict stratified flow velocity in tidal river sections. The impact of stratified flow velocity on the safety of construction vessels is relatively important. In order to improve the safety of construction vessels or provide safe haven areas for ships, a prediction model is needed to predict the flow velocity distribution range in the river. Summary of the invention

[0005] In view of this, the present invention aims to propose a method, device, terminal and storage medium for predicting stratified flow velocity in tidal river sections, so as to solve the problem that the existing tidal river hydrological data prediction model lacks the ability to predict the stratified flow velocity in tidal river sections, thereby affecting the construction safety of construction vessels.

[0006] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0007] In a first aspect, the present invention provides a method for predicting stratified flow velocity in a tidal river section, comprising:

[0008] Obtain the measured flow velocity data, water level data, and upstream and downstream river topography data of the project in previous years, and use a two-dimensional river digital model to calculate the flow process of the project based on the measured flow velocity data, water level data, and upstream and downstream river topography data of the project in previous years, and obtain the average flow velocity results of the project in previous years;

[0009] Constructing an average flow velocity prediction model, and using the water level data of the engineering site measured over the years and the average flow velocity results of the engineering site measured over the years to train the average flow velocity prediction model, to obtain a trained average flow velocity prediction model;

[0010] The velocity stratification ratio of the three-dimensional river channel mathematical model is combined with the average velocity results of the engineering site measured in previous years to infer the long-sequence stratified velocity, and the stratified velocity results of the engineering site in previous years are obtained;

[0011] Constructing a layered ratio convolution prediction model, and using the layered flow velocity results of the engineering site in previous years and the water level data of the engineering site measured in previous years to perform multi-channel data training on the layered ratio convolution prediction model, to obtain a trained layered ratio convolution prediction model;

[0012] Using the trained water level forecast model, the water level process data of the project within the preset forecast period is predicted;

[0013] Input the water level process data of the project within the preset forecast period into the trained average flow velocity prediction model, predict the flow velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period, and solve the water depth data of each discrete point in the digital-analog space in combination with the terrain data of the project;

[0014] Input the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period into the trained stratification ratio convolution prediction model, and calculate the velocity stratification ratio prediction result within the preset forecast period;

[0015] Based on the velocity stratification ratio prediction results within the preset forecast period, the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period are scaled to obtain multi-point velocity forecast results of the project within the preset forecast period.

[0016] Furthermore, the flow velocity data, water level data, and upstream and downstream river topography data of the project measured over the years are obtained, and the flow process of the project is estimated using a two-dimensional river channel digital model based on the flow velocity data, water level data, and upstream and downstream river topography data of the project measured over the years, to obtain the average flow velocity results of the project measured over the years, including:

[0017] Obtain at least 10,000 measured flow velocity data, water level data, and upstream and downstream river topography data of the project over the years, and select the flow velocity data, water level data, and upstream and downstream river topography data of the project corresponding to the extremely high and extremely low water level processes in characteristic years to obtain preliminary data;

[0018] The preliminary data are checked, and the variance of the preliminary data of adjacent 1, 2, and 3 periods is solved based on the periodicity of the data, and the preliminary data with variance exceeding a preset threshold is eliminated to obtain the flow velocity data, water level data, and river topography data of the upstream and downstream of the project measured in previous years;

[0019] Based on the flow velocity data, water level data and upstream and downstream river topography data of the project measured in previous years, the flow process of the project was estimated using a two-dimensional river digital model to obtain the average flow velocity results of the project measured in previous years.

[0020] Furthermore, the spatial discrete scale of the project in the two-dimensional river channel digital model is not less than the scale of terrain change, and the upstream water level boundary and the downstream water level boundary in the two-dimensional river channel digital model are both kept away from the forecast area; among them, the upstream water level boundary needs to use the measured flow velocity data and water level data of the project, combined with the calculated calibration coefficient k The downstream water level boundary needs to use the measured water level data of the project as the downstream water level boundary; the calibration coefficient k The calculation formula is as follows: Q=kVA , Q For the rated flow rate, V is the measured average flow velocity, A is the cross-sectional area.

[0021] Furthermore, the average flow velocity prediction model is constructed, and the average flow velocity prediction model is trained using the water level data of the engineering site measured over the years and the average flow velocity results of the engineering site measured over the years to obtain a trained average flow velocity prediction model, including:

[0022] Based on the machine learning model, the average flow velocity prediction model is constructed;

[0023] The water level and water level growth rate in the water level data of the engineering site measured in previous years are used as features, and the average flow velocity results of the engineering site measured in previous years are used as labels to perform machine learning training on the average flow velocity prediction model to obtain a trained average flow velocity prediction model.

[0024] Furthermore, the layered ratio convolution prediction model is constructed, and multi-channel data training is performed on the layered ratio convolution prediction model using the layered flow velocity results of the engineering department in previous years and the water level data of the engineering department measured in previous years to obtain a trained layered ratio convolution prediction model, including:

[0025] Based on the DenseNet structure, a hierarchical ratio convolution prediction model covering multiple typical working conditions is constructed; wherein the hierarchical ratio convolution prediction model includes an input layer, an intermediate layer, a transition layer, and an output layer, wherein the input layer adopts a Conv+BN+ReLU layer structure, the intermediate layer adopts a Dense Block layer structure, and the number of convolution layers in the Dense Block layer structure is set to different values, the transition layer adopts a Conv+Strided Conv layer structure, and the output layer adopts a GlobalMaxPool layer structure;

[0026] The stratified flow velocity results of the engineering site in previous years and the water level data of the engineering site measured in previous years are used to perform multi-channel data training on the stratified ratio convolution prediction model to obtain a trained stratified ratio convolution prediction model.

[0027] Furthermore, the layered flow velocity results of the engineering department in previous years and the water level data of the engineering department measured in previous years are used to perform multi-channel data training on the layered ratio convolution prediction model to obtain a trained layered ratio convolution prediction model, including:

[0028] Based on the stratified flow velocity results of the engineering department in previous years and the measured water level data of the engineering department in previous years, the data of four channels, namely, the lateral flow velocity data, the longitudinal flow velocity data, the water depth data, and the water surface elevation data, are obtained;

[0029] The data of the four channels are normalized so that their value range falls between [0, 1] or [-1, 1] to obtain a data set;

[0030] Dividing the data set into a training set, a validation set, and a test set according to a preset ratio;

[0031] The layered ratio convolution prediction model is trained using the training set, validation set, and test set to obtain a trained layered ratio convolution prediction model.

[0032] Furthermore, after scaling the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period based on the velocity stratification ratio prediction result within the preset forecast period to obtain the multi-point velocity forecast result of the project within the preset forecast period, the method further includes:

[0033] Based on the multi-point velocity forecast results of the engineering department within the preset forecast period, the velocity distribution equipotential lines are drawn.

[0034] In a second aspect, the present invention further provides a tidal river section layered flow velocity prediction device, comprising:

[0035] The acquisition module is used to obtain the flow velocity data, water level data, and upstream and downstream river terrain data of the project measured in previous years, and to use the two-dimensional river digital model to calculate the flow process of the project based on the flow velocity data, water level data, and upstream and downstream river terrain data of the project measured in previous years, so as to obtain the average flow velocity result of the project measured in previous years;

[0036] The first construction module is used to construct an average flow velocity prediction model, and train the average flow velocity prediction model using the water level data of the engineering site measured over the years and the average flow velocity results of the engineering site measured over the years to obtain a trained average flow velocity prediction model;

[0037] A reverse calculation module is used to reversely calculate the long-sequence stratified flow velocity by using the velocity stratification ratio of the three-dimensional river channel mathematical model combined with the average flow velocity results of the engineering site measured in previous years, so as to obtain the stratified flow velocity results of the engineering site in previous years;

[0038] The second construction module is used to construct a layered ratio convolution prediction model, and use the layered flow velocity results of the engineering site in previous years and the water level data of the engineering site measured in previous years to perform multi-channel data training on the layered ratio convolution prediction model to obtain a trained layered ratio convolution prediction model;

[0039] The calculation module is used to use the trained water level forecast model to predict the water level process data of the project within the preset forecast period;

[0040] The first input module is used to input the water level process data of the project within the preset forecast period into the trained average flow velocity prediction model, predict the flow velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period, and solve the water depth data of each discrete point in the digital-analog space in combination with the terrain data of the project;

[0041] The second input module is used to input the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period into the trained stratification ratio convolution prediction model, and calculate the velocity stratification ratio prediction result within the preset forecast period;

[0042] The forecast module is used to scale the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period based on the velocity stratification ratio prediction result within the preset forecast period to obtain the multi-point velocity forecast result of the project within the preset forecast period.

[0043] In a third aspect, an embodiment of the present invention further provides a terminal, including:

[0044] one or more processors;

[0045] A storage device for storing one or more programs;

[0046] When the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned tidal river section stratified flow velocity prediction method.

[0047] In a fourth aspect, an embodiment of the present invention further provides a storage medium comprising computer executable instructions, which, when executed by a computer processor, are used to execute the above-mentioned method for predicting stratified flow velocity in a tidal river section.

[0048] Compared with the prior art, the method, device, terminal and storage medium for predicting stratified flow velocity in a tidal river section described in the present invention have the following advantages:

[0049] (1) The method, device, terminal and storage medium for predicting stratified flow velocity in a tidal river section described in the present invention nest and run three types of machine learning models in the prediction and estimation process, solve different parameters at different stages, and finally predict the flow field distribution at the engineering site. It has the advantage of high accuracy and can accurately predict the stratified flow velocity in a tidal river section, which is beneficial to improving the safety of construction ships.

[0050] (2) The method, device, terminal and storage medium for predicting layered flow velocity in tidal river sections described in the present invention can quickly extrapolate the results of a two-dimensional model to three dimensions through a layered ratio convolution prediction model, thereby saving the time for long sequence calculations. In particular, it has good prediction speed and effect for multi-year construction periods where it is necessary to estimate the specified flow velocity of a river project.

[0051] (3) The method, device, terminal and storage medium for predicting stratified flow velocity in a tidal river section described in the present invention can provide a real-time warning of the safe flow velocity for ships by setting an average flow velocity prediction model. The average flow velocity prediction model can calculate the real-time flow velocity of the project through the real-time water level process. Combined with the stratified ratio convolution prediction model, it can provide a real-time warning of the safe flow velocity for ships, which is conducive to further improving the safety of ships during the construction process. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The accompanying drawings constituting a part of the present invention are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0053] Figure 1 This is a schematic flow chart of a method for predicting stratified flow velocity in a tidal river section according to the first embodiment of the present invention;

[0054] Figure 2 This is a schematic diagram of a data processing flow in a method for predicting stratified flow velocity in a tidal river section according to the first embodiment of the present invention;

[0055] Figure 3 This is a schematic structural diagram of a device for predicting stratified flow velocity in a tidal river section according to a second embodiment of the present invention;

[0056] Figure 4 A schematic diagram of the structure of a tidal river section stratified flow velocity forecast terminal provided in Embodiment 3 of the present invention. DETAILED DESCRIPTION

[0057] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0058] Embodiment 1

[0059] Figure 1 Schematic diagram of a method for predicting stratified flow velocity in a tidal river section according to the first embodiment of the present invention. Figure 1 , this forecasting method specifically includes the following steps:

[0060] Step 101, obtain the flow velocity data, water level data, and upstream and downstream river terrain data of the project site measured in previous years, and use a two-dimensional river digital model to estimate the flow process of the project site based on the flow velocity data, water level data, and upstream and downstream river terrain data of the project site measured in previous years, and obtain the average flow velocity results of the project site measured in previous years.

[0061] The characteristics of two- and three-dimensional river channel mathematical models in hydrodynamics are mainly reflected in their high analytical ability for complex river channel geometry and fluid dynamics processes. These models can accurately describe the three-dimensional flow field distribution, flow velocity, flow direction and turbulence structure in the river channel, taking into account the complex morphology of the river channel such as bends, depth changes and dynamic changes in riverbed morphology. Two- and three-dimensional river channel mathematical models use a combination of physical equations such as the Navier-Stokes equations and the Reynolds averaged Navier-Stokes equations, combined with numerical methods such as the finite volume method or the finite element method, to simulate complex flow phenomena in the river channel.

[0062] In addition, since the construction of an accurate average flow velocity prediction model requires the measurement of the flow velocity, water level, and river topography data upstream and downstream of the project for many years. The average flow velocity at a single point in the river project is estimated using a two-dimensional hydrodynamic model. In order to obtain an accurate flow process, flow estimation is required. The construction of a two-dimensional river flow velocity machine learning model has certain requirements for the amount of training data and model generalization.

[0063] Therefore, the flow calculation in this embodiment uses a two-dimensional river mathematical model to calculate the flow process at the project site by measuring the flow velocity, water level and river topography at the project site. In fact, due to the unevenness of the river section and the uneven distribution of river flow velocity, the simple calculation result is not accurate. This section gives a calibration coefficient k, calibrate the average flow velocity at the measuring point of the engineering department.

[0064] Flow estimation requires the use of a long sequence of measured processes to perform digital model calibration to avoid good local fit and poor overall fit. To build a good flow rate data training set, the following tasks need to be performed:

[0065] 1. To enrich the relationship between flow velocity and water level characteristics, a large number of training sets (at least 10,000 training data) need to be constructed.

[0066] 2. To ensure the generalization of the training set, it is necessary to select extremely high and extremely low water level processes in characteristic years.

[0067] 3. To ensure the accuracy of the training set, the data is checked. The measured data is time-based, and the variance of the data in 1, 2, or 3 adjacent time periods can be solved, and the data with larger variance values ​​can be eliminated.

[0068] Therefore, in the actual application process, the flow velocity data, water level data, and upstream and downstream river terrain data of the engineering department measured in previous years are obtained, and the flow process of the engineering department is estimated using a two-dimensional river channel digital model based on the flow velocity data, water level data, and upstream and downstream river terrain data of the engineering department measured in previous years, and the average flow velocity result of the engineering department measured in previous years is obtained. The specific steps can be referred to as follows:

[0069] Step 1011, obtain at least 10,000 measured flow velocity data, water level data, and upstream and downstream river topography data of the project over the years, and select the flow velocity data, water level data, and upstream and downstream river topography data of the project corresponding to the extremely high and extremely low water level processes in characteristic years to obtain preliminary data.

[0070] Step 1012, verify the preliminary data, solve the variance of the preliminary data of 1, 2, and 3 adjacent time periods based on the time period of the data, eliminate the preliminary data with variance exceeding the preset threshold, and obtain the flow velocity data, water level data, and river topography data of the upstream and downstream of the project measured in previous years.

[0071] Specifically, the variance values ​​can be sorted from large to small, and the preliminary data corresponding to the first 10%, 20% or 30% variance values ​​can be eliminated to eliminate the influence of abnormal data on the accuracy of the preliminary data. For example, at flood time, the first 30% of the data can be eliminated, at non-flood time, the first 10% of the data can be eliminated, and at the time between the two, the first 20% of the data can be eliminated. Those skilled in the art can also select other appropriate preset thresholds according to actual needs to eliminate the variance values ​​that are abnormal to improve the accuracy of the data, which will not be repeated here.

[0072] In actual application, the specific flow velocity values ​​of each river may be different, and can be judged according to the flow state of the river. For example, assuming the average flow velocity of the river is 1.2m / s, it is a flood moment, and the measured flow velocity values ​​are violently jittering, so the preliminary data corresponding to the top 30% of the variance values ​​after sorting from large to small can be eliminated; if the flow velocity is below 0.7m / s and the river flow is gentle, the preliminary data corresponding to the top 10% of the variance values ​​can be eliminated; in between, the top 20% of the data can be eliminated.

[0073] Step 1013, based on the flow velocity data, water level data and upstream and downstream river topography data of the project measured in previous years, the flow process of the project is estimated using a two-dimensional river digital model to obtain the average flow velocity results of the project measured in previous years.

[0074] Specifically, the spatial discrete scale of the project in the two-dimensional river channel digital model is not less than the scale of terrain changes, and the upstream water level boundary and downstream water level boundary in the two-dimensional river channel digital model are kept away from the forecast area; among them, the upstream water level boundary needs to use the measured flow velocity data and water level data of the project, combined with the calculated calibration coefficient k The downstream water level boundary needs to use the measured water level data of the project as the downstream water level boundary; the calibration coefficient k The calculation formula is as follows: Q=kVA , Q For the rated flow rate, V is the measured average flow velocity, A is the cross-sectional area.

[0075] In actual application, when the cross-sectional area of ​​a natural river remains unchanged, the flow rate of the river is correlated with the water level. Therefore, it is necessary to calibrate multiple times before low water level conditions and high water level conditions to obtain the flow rate under different water levels. k value.

[0076] In addition, the spatial discretization, modeling scope and boundary conditions of the river channel in the two-dimensional river channel digital model can refer to the following settings: Spatial discretization of the river channel: The spatial discretization scale of the river channel is determined according to needs. For projects that need to accurately predict the flow velocity of the project in the river channel, the spatial discretization scale of the project should not be less than the scale of terrain changes. Modeling scope: Determined according to actual needs. To ensure the accuracy of the model, the upstream and downstream boundaries of the model are kept at a certain distance from the forecast area, generally more than one kilometer away from the forecast area. Boundary conditions: Use the measured long sequence flow velocity and water level data, combined with the calculated calibration coefficient k , calculate the upstream flow boundary of the long sequence. Use the measured water level as the downstream water level boundary.

[0077] Step 102: construct an average flow velocity prediction model, and train the average flow velocity prediction model using the water level data of the engineering site measured in previous years and the average flow velocity results of the engineering site measured in previous years to obtain a trained average flow velocity prediction model.

[0078] Machine Learning (ML) is a technology that enables computer systems to automatically learn from data and improve their performance. It is one of the key areas of artificial intelligence. Current machine learning techniques are mainly divided into three categories: supervised learning, unsupervised learning, and reinforcement learning. Supervised learning uses labeled data sets for model training. The goal is to learn the mapping relationship from input to output. It is often used for classification and regression tasks, such as image recognition and predictive analysis.

[0079] Specifically, the above-mentioned construction of the average flow velocity prediction model, and the use of the water level data of the engineering site measured in previous years and the average flow velocity results of the engineering site measured in previous years to train the average flow velocity prediction model, to obtain the trained average flow velocity prediction model, specifically includes the following steps:

[0080] Step 1021: construct an average flow velocity prediction model based on the machine learning model.

[0081] Step 1022: using the water level and water level growth rate in the water level data of the engineering site measured in previous years as features, and using the average flow velocity results of the engineering site measured in previous years as labels, perform machine learning training on the average flow velocity prediction model to obtain a trained average flow velocity prediction model.

[0082] In actual application, the average flow velocity prediction model can use water level, water level growth rate, etc. as features, and the flow velocity results calculated by the two-dimensional river mathematical model as labels, and perform machine learning to obtain a trained average flow velocity prediction model with high prediction accuracy.

[0083] Step 103, using the velocity stratification ratio of the three-dimensional river channel mathematical model combined with the average velocity results of the engineering site measured in previous years to reversely infer the long-sequence stratified velocity, and obtain the stratified velocity results of the engineering site in previous years.

[0084] Since the calculation time for long-sequence flow velocity estimation using a three-dimensional river mathematical model is too long, a two-dimensional river mathematical model can be used to calculate the average flow velocity. The long-sequence stratified flow velocity can then be inferred by combining the velocity stratification ratio of the three-dimensional river mathematical model with the average flow velocity, thus shortening the calculation time and improving the calculation efficiency.

[0085] Step 104: construct a stratified ratio convolution prediction model, and use the stratified flow velocity results of the engineering site in previous years and the water level data of the engineering site measured in previous years to perform multi-channel data training on the stratified ratio convolution prediction model to obtain a trained stratified ratio convolution prediction model.

[0086] Specifically, the above-mentioned construction of the layered ratio convolution prediction model, and the use of the layered flow velocity results of the engineering department in previous years and the water level data of the engineering department measured in previous years to perform multi-channel data training on the layered ratio convolution prediction model to obtain the trained layered ratio convolution prediction model, specifically includes the following steps:

[0087] Step 1041. Based on the DenseNet structure, a hierarchical ratio convolution prediction model covering multiple typical working conditions is constructed; wherein the hierarchical ratio convolution prediction model includes an input layer, an intermediate layer, a transition layer, and an output layer, the input layer adopts a Conv+BN+ReLU layer structure, the intermediate layer adopts a Dense Block layer structure, and the number of convolution layers in the Dense Block layer structure is set to different values, the transition layer adopts a Conv+Strided Conv layer structure, and the output layer adopts a GlobalMaxPool layer structure.

[0088] DenseNet is a convolutional neural network architecture that features a dense connection mode, where each layer is directly connected to all subsequent layers to achieve cross-layer reuse of features. It consists of dense blocks and transition layers, where each layer in the dense block is fully connected, and the transition layer is responsible for dimensionality reduction. This structure can effectively alleviate gradient vanishing, enhance feature propagation and reuse, and is suitable for tasks such as image recognition.

[0089] This embodiment makes relevant changes to the DenseNet model architecture. The specific changes are as follows: (1) In the input layer, the "Conv+BN+ReLU" structure is used, and MaxPool is not used, because more detailed information may need to be retained in the initial stage. (2) In the intermediate layer, the number of convolutional layers in the Dense Block is set to different values. The first Dense Block contains 6 convolutional layers, the second contains 8, and the third contains 10. The output of each convolutional layer will be used as the input of all subsequent convolutional layers to achieve dense connection and reuse of features. (3) In the transition layer, "Conv+Strided Conv" is used instead of DenseNet's "Conv+AvgPool". By refining features with a convolution with a step size greater than 1, the data dimension can be efficiently reduced and the condensed expression of spatial information can be enhanced. (4) In the output layer, the "GlobalMaxPool" layer is used instead of the "GAPool" layer to capture the maximum value information in the feature map, which helps the model focus on the most critical part, making the stratification ratio prediction larger, and the final flow rate prediction more conservative.

[0090] In addition, the velocity stratification ratio is correlated with the flow direction, velocity, water depth and other characteristics in the river channel. The water level in the tidal river section changes periodically, and the velocity stratification ratio also changes. The flood process weakens the tidal characteristics of the tidal river section, and the velocity stratification ratio is also affected. The velocity stratification ratio under multiple typical working conditions can be calculated, and a machine learning model can be constructed. Through given hydrological characteristics, the velocity stratification ratio can be quickly inferred to solve the flow velocity of each layer.

[0091] Therefore, in the actual application process, the layered ratio convolution prediction model is constructed, and the typical working conditions that can be included are as follows: 1. Multiple high, medium and low tides in the dry season. 2. Multiple high, medium and low tides in the flood season. 3. Multiple typical flood processes. 4. High and low water level processes recorded in history. In addition, those skilled in the art can also select other typical working conditions according to actual needs, which will not be repeated here.

[0092] In the actual application process, when the data is processed and calculated under typical working conditions, the estimated data such as flow direction, flow velocity, water depth, etc. are obtained. The specific steps for constructing the layered ratio convolution prediction model are as follows (taking the surface layer as an example, the other layers are not repeated):

[0093] First, in order to consider the flow direction in the model, the flow velocities in the horizontal and vertical directions (i.e., the X-axis direction and the Y-axis direction) are used for training the model.

[0094] Secondly, the model feature layer mainly includes the lateral flow velocity, longitudinal flow velocity, water depth, and water surface elevation characteristics.

[0095] Again, the model data is rasterized. Mathematical models mostly use unstructured triangles to better adapt to terrain changes. Convolutional neural networks use raster data and interpolation methods to convert digital model data into raster data.

[0096] Finally, data labeling, unlike traditional manual labeling, can use the equipotential line method to label data for training model stratification.

[0097] Step 1042: Use the stratified flow velocity results of the engineering site in previous years and the water level data of the engineering site measured in previous years to perform multi-channel data training on the stratified ratio convolution prediction model to obtain a trained stratified ratio convolution prediction model.

[0098] Unlike traditional visual models, layered convolutional prediction models require multi-channel data for training. In multi-channel convolutional model training, the use of DenseNet (dense convolutional network) structure can significantly improve the efficiency and performance of the model.

[0099] In the actual application process, the layered flow velocity results of the engineering department in previous years and the water level data of the engineering department measured in previous years are used to perform multi-channel data training on the layered ratio convolution prediction model to obtain a trained layered ratio convolution prediction model, which specifically includes the following steps:

[0100] Firstly, based on the stratified flow velocity results of the engineering site in previous years and the measured water level data of the engineering site in previous years, data of four channels are obtained, namely, lateral flow velocity data, longitudinal flow velocity data, water depth data, and water surface elevation data.

[0101] Secondly, the data of the four channels are normalized so that their value range falls between [0, 1] or [-1, 1] to obtain a data set. By normalizing so that their value range falls between [0, 1] or [-1, 1], model training can be accelerated and stability can be improved.

[0102] Again, the data set is divided into a training set, a validation set, and a test set according to a preset ratio. Specifically, the ratio of the training set, the validation set, and the test set can be 6:2:2. Those skilled in the art can also adjust the ratio according to actual needs, which will not be repeated here.

[0103] Finally, the layered ratio convolution prediction model is trained using the training set, validation set, and test set to obtain a trained layered ratio convolution prediction model.

[0104] Step 105: Use the trained water level forecast model to predict the water level process data of the project within a preset forecast period. The preset forecast period can be set according to forecast needs, such as one day, two days, one week or other time, which will not be repeated here.

[0105] Unlike the average flow rate forecast model and the stratification ratio forecast model, the water level forecast of the project office needs to find the temporal correlation between the water level of the project office and the water level of the upstream hydrological station. A variety of deep learning models such as transformer, autoformer, informer, and reformer can be used. By learning the water level data of multiple hydrological stations upstream of the project and the tidal data of the project office, a deep learning model with better water level expression effect is comprehensively selected. Those skilled in the art can select a suitable existing water level forecast model according to actual needs to realize the forecast of the water level, which will not be repeated here.

[0106] However, it is necessary to pay attention to the following requirements when training the water level forecast model: 1. Select hydrological stations without dams. 2. Select evenly along the main river. 3. In order to increase the forecast period, it is necessary to select hydrological stations far away from the project along the river. 4. Check whether each station has long-series observation data of the same period in time and the accuracy of the data.

[0107] Step 106: input the water level process data of the project site during the preset forecast period into the trained average flow velocity prediction model, predict the flow velocity process data of all discrete points in the digital-analog space of the project site during the preset forecast period, and solve the water depth data of each discrete point in the digital-analog space in combination with the terrain data of the project site.

[0108] Step 107, input the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period into the trained stratification ratio convolution prediction model, and calculate the velocity stratification ratio prediction result within the preset forecast period.

[0109] Step 108: Based on the velocity stratification ratio prediction result within the preset forecast period, the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period are scaled to obtain multi-point velocity forecast results of the project within the preset forecast period.

[0110] Specifically, after scaling the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period based on the velocity stratification ratio prediction result within the preset forecast period to obtain the multi-point velocity forecast result of the project within the preset forecast period, the method further includes:

[0111] Step 109: Draw velocity distribution equipotential lines according to the multi-point velocity forecast results of the project within the preset forecast period.

[0112] Figure 2 The figure is a schematic diagram of the data processing flow in a method for predicting stratified flow velocity in a tidal river section according to the first embodiment of the present invention. In the actual application process, the stratified equipotential lines of the project can be obtained by inputting the four-channel data of the stratified ratio convolution prediction model to calculate the characteristics (lateral flow velocity, longitudinal flow velocity, water depth, and water surface elevation). Then, the coordinates of the discrete unstructured triangle nodes in the digital-analog space are obtained, and the stratified ratio is interpolated to a specific spatial point using an interpolation algorithm.

[0113] This embodiment obtains the flow velocity data, water level data, and upstream and downstream river terrain data of the engineering site measured in previous years, and uses a two-dimensional river digital model to estimate the flow process of the engineering site based on the flow velocity data, water level data, and upstream and downstream river terrain data of the engineering site measured in previous years, so as to obtain the average flow velocity results of the engineering site measured in previous years; constructs an average flow velocity prediction model, and uses the water level data of the engineering site measured in previous years and the average flow velocity results of the engineering site measured in previous years to train the average flow velocity prediction model, so as to obtain the trained average flow velocity prediction model; uses the flow velocity stratification ratio of the three-dimensional river mathematical model combined with the average flow velocity results of the engineering site measured in previous years to infer the long sequence stratified flow velocity, so as to obtain the stratified flow velocity results of the engineering site in previous years; constructs a stratified ratio convolution prediction model, and uses the stratified flow velocity results of the engineering site measured in previous years and the water level data of the engineering site measured in previous years to train the stratified ratio convolution prediction model The multi-channel data training is carried out on the model to obtain a trained stratified ratio convolution prediction model; the trained water level forecast model is used to predict the water level process data of the project within the preset forecast period; the water level process data of the project within the preset forecast period is input into the trained average flow velocity forecast model to predict the flow velocity process data of all digital-analog space discrete points of the project within the preset forecast period, and the water depth data of each digital-analog space discrete point is solved in combination with the terrain data of the project; the flow velocity process data of all digital-analog space discrete points of the project within the preset forecast period are input into the trained stratified ratio convolution prediction model to calculate the flow velocity stratification ratio prediction result within the preset forecast period; based on the flow velocity stratification ratio prediction result within the preset forecast period, the flow velocity process data of all digital-analog space discrete points of the project within the preset forecast period are scaled to obtain the multi-point flow velocity forecast result of the project within the preset forecast period.

[0114] Compared with the prior art, this embodiment nested and ran three types of machine learning models during the calculation process, solved different parameters at different stages, and finally predicted the flow field distribution at the engineering site. It has the advantage of high accuracy and can accurately predict the stratified flow velocity in tidal river sections, which is beneficial to improving the safety of construction ships.

[0115] Embodiment 2

[0116] Figure 3 Schematic diagram of the structure of a tidal river section layered flow velocity forecasting device according to the second embodiment of the present invention. Figure 3 The device for predicting stratified flow velocity in tidal river sections comprises:

[0117] The acquisition module 201 is used to obtain the flow velocity data, water level data, and upstream and downstream river terrain data of the project measured in previous years, and use the two-dimensional river digital model to calculate the flow process of the project based on the flow velocity data, water level data, and upstream and downstream river terrain data of the project measured in previous years, and obtain the average flow velocity result of the project measured in previous years;

[0118] The first construction module 202 is used to construct an average flow velocity prediction model, and train the average flow velocity prediction model using the water level data of the engineering site measured over the years and the average flow velocity results of the engineering site measured over the years to obtain a trained average flow velocity prediction model;

[0119] The reverse calculation module 203 is used to reversely calculate the long-sequence stratified flow velocity by using the velocity stratification ratio of the three-dimensional river channel mathematical model and the average flow velocity results of the engineering site measured in previous years, so as to obtain the stratified flow velocity results of the engineering site in previous years;

[0120] The second construction module 204 is used to construct a layered ratio convolution prediction model, and use the layered flow velocity results of the engineering site in previous years and the water level data of the engineering site measured in previous years to perform multi-channel data training on the layered ratio convolution prediction model to obtain a trained layered ratio convolution prediction model;

[0121] The calculation module 205 is used to use the trained water level forecasting model to predict the water level process data of the project within a preset forecast period;

[0122] The first input module 206 is used to input the water level process data of the project within the preset forecast period into the trained average flow velocity prediction model, predict the flow velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period, and solve the water depth data of each discrete point in the digital-analog space in combination with the terrain data of the project;

[0123] The second input module 207 is used to input the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period into the trained stratification ratio convolution prediction model, and calculate the velocity stratification ratio prediction result within the preset forecast period;

[0124] The forecast module 208 is used to scale the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period based on the velocity stratification ratio prediction results within the preset forecast period to obtain multi-point velocity forecast results of the project within the preset forecast period.

[0125] Optionally, the tidal river section layered flow velocity prediction device also includes:

[0126] The drawing module 209 is used to draw velocity distribution equipotential lines according to the multi-point velocity forecast results of the project within the preset forecast period.

[0127] The device for predicting stratified flow velocity in a tidal river section described in this embodiment can execute the method for predicting stratified flow velocity in a tidal river section provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0128] Embodiment 3

[0129] Figure 4 A schematic diagram of the structure of a tidal river section layered flow velocity forecast terminal provided in the third embodiment of the present invention; Figure 4 A block diagram of an exemplary terminal suitable for implementing embodiments of the present invention is shown. Figure 4 The terminal shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0130] like Figure 4 As shown, the terminal 12 is in the form of a general-purpose computing device. The components of the terminal 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 connecting different system components (including the system memory 28 and the processing unit 16).

[0131] Bus 18 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. By way of example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0132] Terminal 12 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by terminal 12, including volatile and non-volatile media, removable and non-removable media.

[0133] System memory 28 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 30 and / or cache memory 32. Terminal 12 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 34 may be used to read and write non-removable, non-volatile magnetic media ( Figure 4 not shown, usually called a "hard drive"). Although Figure 4Not shown in the figure, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 18 via one or more data medium interfaces. The memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present invention.

[0134] A program / utility 40 having a set (at least one) of program modules 42 may be stored, for example, in the memory 28, such program modules 42 including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 42 generally perform the functions and / or methods of the embodiments described herein.

[0135] The terminal 12 may also communicate with one or more external devices 14 (e.g., keyboards, pointing devices, displays 24, etc.), may communicate with one or more devices that enable a user to interact with the terminal 12, and / or may communicate with any device that enables the terminal 12 to communicate with one or more other computing devices (e.g., network cards, modems, etc.). Such communication may be performed via an input / output (I / O) interface 22. Furthermore, the terminal 12 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 20. As shown, the network adapter 20 communicates with other modules of the terminal 12 via a bus 18. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the terminal 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0136] The processing unit 16 executes various functional applications and data processing by running the programs stored in the system memory 28, such as implementing the stratified flow velocity prediction method for a tidal river section provided in an embodiment of the present invention.

[0137] Embodiment 4

[0138] Embodiment 4 of the present invention further provides a storage medium comprising computer executable instructions, wherein the computer executable instructions, when executed by a computer processor, are used to execute any of the tidal river section stratified flow velocity prediction methods provided in the above embodiments.

[0139] The computer storage medium of the embodiment of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, device or device.

[0140] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, which carry computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0141] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0142] Computer program code for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0143] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for predicting stratified flow velocity in a tidal river section, characterized in that: include: Obtain the measured flow velocity data, water level data, and upstream and downstream river topography data of the project in previous years, and use a two-dimensional river digital model to calculate the flow process of the project based on the measured flow velocity data, water level data, and upstream and downstream river topography data of the project in previous years, and obtain the average flow velocity results of the project in previous years; Constructing an average flow velocity prediction model, and using the water level data of the engineering site measured over the years and the average flow velocity results of the engineering site measured over the years to train the average flow velocity prediction model, to obtain a trained average flow velocity prediction model; The velocity stratification ratio of the three-dimensional river channel mathematical model is combined with the average velocity results of the engineering site measured in previous years to infer the long-sequence stratified velocity, and the stratified velocity results of the engineering site in previous years are obtained; Constructing a layered ratio convolution prediction model, and using the layered flow velocity results of the engineering site in previous years and the water level data of the engineering site measured in previous years to perform multi-channel data training on the layered ratio convolution prediction model, to obtain a trained layered ratio convolution prediction model; Using the trained water level forecast model, the water level process data of the project within the preset forecast period is predicted; Input the water level process data of the project within the preset forecast period into the trained average flow velocity prediction model, predict the flow velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period, and solve the water depth data of each discrete point in the digital-analog space in combination with the terrain data of the project; Input the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period into the trained stratification ratio convolution prediction model, and calculate the velocity stratification ratio prediction result within the preset forecast period; Based on the velocity stratification ratio prediction results within the preset forecast period, the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period are scaled to obtain multi-point velocity forecast results of the project within the preset forecast period.

2. The method according to claim 1, characterized in that The method of obtaining the measured flow velocity data, water level data, and upstream and downstream river terrain data of the project in previous years, and using the two-dimensional river digital model to calculate the flow process of the project based on the measured flow velocity data, water level data, and upstream and downstream river terrain data of the project in previous years, and obtaining the average flow velocity results of the project in previous years, includes: Obtain at least 10,000 measured flow velocity data, water level data, and upstream and downstream river topography data of the project over the years, and select the flow velocity data, water level data, and upstream and downstream river topography data of the project corresponding to the extremely high and extremely low water level processes in characteristic years to obtain preliminary data; The preliminary data are checked, and the variance of the preliminary data of adjacent 1, 2, and 3 periods is solved based on the periodicity of the data, and the preliminary data with variance exceeding a preset threshold is eliminated to obtain the flow velocity data, water level data, and river topography data of the upstream and downstream of the project measured in previous years; Based on the flow velocity data, water level data and upstream and downstream river topography data of the project measured in previous years, the flow process of the project was estimated using a two-dimensional river digital model to obtain the average flow velocity results of the project measured in previous years.

3. The method according to claim 2, characterized in that: The spatial discrete scale of the project in the two-dimensional river channel digital model is not less than the scale of terrain changes. The upstream water level boundary and the downstream water level boundary in the two-dimensional river channel digital model are both kept away from the forecast area; among them, the upstream water level boundary needs to use the measured flow velocity data and water level data of the project, combined with the calculated calibration coefficient k The downstream water level boundary needs to use the measured water level data of the project as the downstream water level boundary; the calibration coefficient k The calculation formula is as follows: Q=kVA , Q For the rated flow rate, V is the measured average flow velocity, A is the cross-sectional area.

4. The method according to claim 1, characterized in that: The average flow velocity prediction model is constructed, and the average flow velocity prediction model is trained using the water level data of the engineering site measured over the years and the average flow velocity results of the engineering site measured over the years to obtain a trained average flow velocity prediction model, including: Based on the machine learning model, the average flow velocity prediction model is constructed; The water level and water level growth rate in the water level data of the engineering site measured in previous years are used as features, and the average flow velocity results of the engineering site measured in previous years are used as labels to perform machine learning training on the average flow velocity prediction model to obtain a trained average flow velocity prediction model.

5. The method according to claim 1, characterized in that The layered ratio convolution prediction model is constructed, and multi-channel data training is performed on the layered ratio convolution prediction model using the layered flow velocity results of the engineering department in previous years and the water level data of the engineering department measured in previous years to obtain a trained layered ratio convolution prediction model, including: Based on the DenseNet structure, a hierarchical ratio convolution prediction model covering multiple typical working conditions is constructed; wherein the hierarchical ratio convolution prediction model includes an input layer, an intermediate layer, a transition layer, and an output layer, wherein the input layer adopts a Conv+BN+ReLU layer structure, the intermediate layer adopts a Dense Block layer structure, and the number of convolution layers in the Dense Block layer structure is set to different values, the transition layer adopts a Conv+Strided Conv layer structure, and the output layer adopts a GlobalMaxPool layer structure; The stratified flow velocity results of the engineering site in previous years and the water level data of the engineering site measured in previous years are used to perform multi-channel data training on the stratified ratio convolution prediction model to obtain a trained stratified ratio convolution prediction model.

6. The method according to claim 5, characterized in that The method uses the stratified flow velocity results of the engineering department in previous years and the water level data of the engineering department measured in previous years to perform multi-channel data training on the stratified ratio convolution prediction model to obtain a trained stratified ratio convolution prediction model, including: Based on the stratified flow velocity results of the engineering department in previous years and the measured water level data of the engineering department in previous years, the data of four channels, namely, the lateral flow velocity data, the longitudinal flow velocity data, the water depth data, and the water surface elevation data, are obtained; The data of the four channels are normalized so that their value range falls between [0, 1] or [-1, 1] to obtain a data set; Dividing the data set into a training set, a validation set, and a test set according to a preset ratio; The layered ratio convolution prediction model is trained using the training set, validation set, and test set to obtain a trained layered ratio convolution prediction model.

7. The method according to claim 1, characterized in that After scaling the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period based on the velocity stratification ratio prediction result within the preset forecast period to obtain the multi-point velocity forecast result of the project within the preset forecast period, the method further includes: Based on the multi-point velocity forecast results of the engineering department within the preset forecast period, the velocity distribution equipotential lines are drawn.

8. A device for predicting stratified flow velocity in a tidal river section, characterized in that: include: The acquisition module is used to obtain the flow velocity data, water level data, and upstream and downstream river terrain data of the project measured in previous years, and to use the two-dimensional river digital model to calculate the flow process of the project based on the flow velocity data, water level data, and upstream and downstream river terrain data of the project measured in previous years, so as to obtain the average flow velocity result of the project measured in previous years; The first construction module is used to construct an average flow velocity prediction model, and train the average flow velocity prediction model using the water level data of the engineering site measured over the years and the average flow velocity results of the engineering site measured over the years to obtain a trained average flow velocity prediction model; A reverse calculation module is used to reversely calculate the long-sequence stratified flow velocity by using the velocity stratification ratio of the three-dimensional river channel mathematical model combined with the average flow velocity results of the engineering site measured in previous years, so as to obtain the stratified flow velocity results of the engineering site in previous years; The second construction module is used to construct a layered ratio convolution prediction model, and use the layered flow velocity results of the engineering site in previous years and the water level data of the engineering site measured in previous years to perform multi-channel data training on the layered ratio convolution prediction model to obtain a trained layered ratio convolution prediction model; The calculation module is used to use the trained water level forecast model to predict the water level process data of the project within the preset forecast period; The first input module is used to input the water level process data of the project within the preset forecast period into the trained average flow velocity prediction model, predict the flow velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period, and solve the water depth data of each discrete point in the digital-analog space in combination with the terrain data of the project; The second input module is used to input the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period into the trained stratification ratio convolution prediction model, and calculate the velocity stratification ratio prediction result within the preset forecast period; The forecast module is used to scale the velocity process data of all discrete points in the digital-analog space of the project within the preset forecast period based on the velocity stratification ratio prediction result within the preset forecast period to obtain the multi-point velocity forecast result of the project within the preset forecast period.

9. A terminal, characterized in that: include: one or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the tidal river section stratified flow velocity prediction method as described in any one of claims 1-7.

10. A storage medium comprising a computer executable instruction program, wherein the computer executable instruction program is used to execute the stratified flow velocity prediction method for a tidal river section as claimed in any one of claims 1 to 7 when executed by a computer processor.