A method and system for inverting tidal bore propagation in estuaries

Abstract

The present invention relates to the intersection of intelligent water conservancy and marine information technology, and particularly to a method and system for inverting estuary tidal bore propagation. The method comprises the following steps: using an image acquisition device to acquire river surface tidal head line images within an observation area at an observation location; organizing and summarizing the river surface tidal head line images to obtain a river surface tidal head line image dataset; building a feature extraction network and training the feature extraction network using the river surface tidal head line image dataset; selecting a target tidal head line and using the image acquisition device to obtain a target tidal head line image containing the target tidal head line; using the trained feature extraction network to extract positional features of the target tidal head line in the target tidal head line image; and obtaining a tidal bore propagation velocity field in the observation area based on the positional features of the target tidal head line. The present invention extracts the positional features of the river surface tidal head line image through the trained feature extraction network, thereby inverting the morphology and velocity field of tidal bore propagation with high real-time, high accuracy, and high reliability.

Description

Technical Field

[0001] The present invention relates to the intersection of intelligent water conservancy and marine information technology, and in particular to a method and system for inverting estuary tidal bore propagation. Background Art

[0002] Estuarine tidal bores are the high currents caused by ocean tides entering specific estuaries. Due to their complexity and unpredictability, real-time quantitative observation has always been a challenge, especially in large estuaries. Traditional mathematical models have been used to predict tidal bore development, but these models often fail to effectively validate large-scale tidal bore propagation patterns. Therefore, the development of a real-time and accurate inversion method for estuarine tidal bore propagation is crucial.

[0003] At present, traditional methods for simulating tidal bore propagation mainly rely on data collected from tidal stations and ships, and use these data for inference and model calculations. However, this method requires a lot of manpower and material resources, and is limited by the collection equipment and environment, and there are major problems with real-time and accuracy. Therefore, scientists began to seek new technical means to achieve real-time monitoring and quantitative observation of tidal bore propagation. As a new type of remote sensing technology, ocean radar technology can provide high-temporal and spatial resolution sea surface motion field information, thus providing a new approach to the study of tidal bore propagation. However, current ocean radar technology still faces many challenges, such as how to effectively invert the velocity field and morphology of tidal bore propagation, and how to quickly extract tidal wave information from radar image fast images. Summary of the Invention

[0004] In order to solve the limitations of traditional observation methods and meet the needs of practical applications, in the first aspect, the present invention provides a method for inverting tidal bore propagation in estuaries, which aims to provide a new simulation method for the study of tidal bore propagation. The method comprises the following steps: selecting an observation position and an observation area; setting an image acquisition device at the observation position; using the image acquisition device to acquire river surface tidal head line images within the observation area; arranging and summarizing the river surface tidal head line images to obtain a river surface tidal head line image dataset; building a feature extraction network and training the feature extraction network using the river surface tidal head line image dataset; selecting a target tidal head line and using the image acquisition device to obtain a target tidal head line image containing the target tidal head line; using the trained feature extraction network to extract the position features of the target tidal head line in the target tidal head line image; obtaining the velocity vector of the target tidal head line based on the position features and time intervals of the target tidal head line images that are adjacent in time sequence; summarizing the velocity vectors of the target tidal head lines in the observation area to obtain the tidal bore propagation velocity field of the observation area. The present invention inverts the morphology and velocity field of tidal bore propagation with high real-time, high accuracy, and high reliability by selecting an observation location and observation area, collecting river surface tidal crest line images, training a feature extraction network, and calculating the tidal bore propagation velocity field. The image acquisition device can be further selected based on ocean radar technology. Specifically, the present invention applies ocean radar technology to the estuary tidal bore propagation inversion method. The radar reflection image captured by ocean radar technology serves as the river surface tidal crest line image of the present invention. Combined with the feature extraction network, the method enables rapid extraction of tidal wave information from the radar reflection image. Furthermore, the method utilizes the complete and rich tidal wave information contained in the radar reflection image to invert the morphology and velocity field of the tidal bore propagation with greater accuracy and effectiveness.

[0005] Optionally, obtaining the velocity vector of the target tidal head line further includes the following steps: building a tidal head line prediction network and training the tidal head line prediction network using the river surface tidal head line image dataset; for the position features of the target tidal head line in any target tidal head line image, using the trained tidal head line prediction network to predict the predicted position features of the target tidal head line after one time step, wherein the time step is less than the time interval; inserting the predicted position features into the position features of the target tidal head line images that are adjacent in time sequence; obtaining the velocity vector of the target tidal head line based on the position features, predicted position features, time interval, and time sequence step of the target tidal head line images that are adjacent in time sequence. The present invention predicts the position features of the target tidal head line by adding a tidal head line prediction network, thereby obtaining the velocity vector of the target tidal head line. This not only solves the problem of insufficient acquisition of the target tidal head line position features within the observation area, but also greatly improves the accuracy of the inverted tidal bore propagation velocity field, while increasing the ability to predict future tidal bore propagation conditions, and improving the accuracy and reliability of the simulation results.

[0006] Optionally, the image acquisition device includes an X-band radar telemetry device.

[0007] Optionally, the collating and summarizing of the river surface tidal line images to obtain a river surface tidal line image dataset includes the following steps: cleaning the river surface tidal line images and unifying the format of the cleaned river surface tidal line images; marking feature information of the cleaned and formatted river surface tidal line images to obtain a river surface tidal line image dataset; and storing the river surface tidal line image dataset.

[0008] Optionally, building a feature extraction network and using the river surface tidal line image dataset to train the feature extraction network includes the following steps: building a feature extraction network based on a convolutional neural network; setting a division ratio, and dividing the river surface tidal line image dataset into a river surface tidal line image training set, a river surface tidal line image verification set, and a river surface tidal line image test set according to the division ratio; using the river surface tidal line image training set to train the feature extraction network, using the river surface tidal line image verification set to adjust the hyperparameters of the feature extraction network after training, and using the river surface tidal line image test set to evaluate the feature extraction network after the hyperparameters are adjusted.

[0009] Optionally, the feature extraction network includes: 4 convolutional layers, 4 pooling layers and 2 fully connected layers, the output of the convolutional layer is the pooling layer, and the fully connected layers are the first fully connected layer and the second fully connected layer; the convolutional layer includes a 3×3 convolution kernel and a ReLU activation function; the stride of the pooling layer is 2×2; the first fully connected layer contains 128 neurons and a ReLU activation function respectively; the second fully connected layer contains 64 neurons and a ReLU activation function respectively; the loss function of the feature extraction network is a cross entropy loss function.

[0010] Optionally, the method of obtaining the velocity vector of the target tidal line according to the position characteristics and time interval of the target tidal line images adjacent in time sequence comprises the following steps: obtaining the position characteristics of the target tidal line image that is earlier in time sequence; in, n represents the total number of position features, Represents the i-th position feature in the target tide head line image at time T0, which is earlier in the time sequence Obtain the position characteristics of the target tide head line image at a later time sequence in, n represents the total number of position features, Represents the i-th position feature in the target tide line image at time T1, which is later in the time sequence Extract the corresponding position features of the earlier and later time series respectively, and obtain the unidirectional velocity vector V between the two position features i, the unidirectional velocity vector V i Satisfies the following formula: in, The horizontal coordinate of the i-th position feature at the later time T0, The horizontal coordinate of the i-th position feature at the earlier time T1; The vertical coordinate of the i-th position feature at the later time T0, The vertical coordinate of the i-th position feature at time T1, which is earlier in the time sequence; the unidirectional velocity vector V corresponding to any position feature in the target tide line image adjacent to the time sequence i , obtain the velocity vector V of the target tide head line, which satisfies the following formula: V={V1,V2,...,V i ,...,V n}.

[0011] Optionally, the method of obtaining the velocity vector of the target tidal line according to the position features, predicted position features, time interval and time step of the target tidal line images adjacent in time sequence comprises the following steps: obtaining the position features of the target tidal line image that is earlier in time sequence; in, n represents the total number of position features, Represents the i-th position feature in the target tide head line image at time T0, which is earlier in the time sequence Obtain the predicted position features of the target tide head line image that is earlier in the time series in, n represents the total number of position features, Represents the i-th predicted position feature of the predicted target tide head line at the prediction time t0 Extract the position features corresponding to the earlier time series and the predicted time respectively, and obtain the unidirectional velocity vector V between the position features of the earlier time series and the position features of the predicted time i , the unidirectional velocity vector V i Satisfies the following formula: T1-T0=ΔT, t0-T0=Δt, T1-t0≤Δt, where T1 represents the later moment, ΔT represents the time interval, and Δt represents the time step. Represents the horizontal coordinate of the i-th position feature at the prediction time t0, The horizontal coordinate of the i-th position feature at the time T0, which is earlier in the time series, The ordinate of the i-th position feature at the prediction time t0; The vertical coordinate of the i-th position feature at the time T0 that is ahead in the time sequence; the one-way velocity vector V corresponding to any position feature in the target tidal head line image ahead in the time sequence and the predicted target tidal head line is summarized. i, obtain the velocity vector V of the target tide head line, which satisfies the following formula: V={V1,V2,...,V i ,...,V n}.

[0012] Optionally, the method of summarizing the velocity vectors of the target tide head line in the observation area to obtain the tidal bore propagation velocity field of the observation area includes the following steps: summarizing the velocity vectors of multiple frames of target tide head line images in the observation area into the same observation area image; obtaining a tidal bore propagation image of the observation area according to the direction of the velocity vector; obtaining a tidal bore flow velocity image of the observation area according to the size of the velocity vector; and combining the tidal bore propagation image and the tidal bore flow velocity image to obtain the tidal bore propagation velocity field of the observation area.

[0013] In a second aspect, to better implement the above-mentioned estuarine tidal bore propagation inversion method, the present invention further provides an estuarine tidal bore propagation inversion system, the estuarine tidal bore propagation inversion system comprising one or more processors; one or more input devices, one or more output devices, and a memory, wherein the processor, the input device, the output device, and the memory are connected via a bus, the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the estuarine tidal bore propagation inversion method provided by the first aspect of the present invention. The estuarine tidal bore propagation inversion system provided by the present invention has a compact structure and stable performance, and can efficiently and accurately implement the estuarine tidal bore propagation inversion method. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A flow chart of a dynamic simulation method for estuarine tidal bore propagation provided by an embodiment of the present invention;

[0015] Figure 2 This is an image of the river tidal line collected by an X-band radar telemetry device according to an embodiment of the present invention;

[0016] Figure 3 A schematic diagram of the velocity vector of the target tidal head line obtained in an embodiment of the present invention;

[0017] Figure 4 A flow chart for improving the accuracy of the inverted tidal bore propagation velocity field provided by an embodiment of the present invention;

[0018] Figure 5 Schematic diagram of the velocity vector of the target tidal head line after inserting the predicted position feature in an embodiment of the present invention;

[0019] Figure 6 This is a structural diagram of the estuary tidal bore propagation inversion system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] Specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the present invention. In the following description, numerous specific details are set forth to provide a thorough understanding of the present invention. However, it will be apparent to one of ordinary skill in the art that these specific details are not necessarily required to practice the present invention. In other instances, well-known circuits, software, or methods are not specifically described to avoid obscuring the present invention.

[0021] Throughout this specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present invention. Therefore, appearances of the phrases "in one embodiment," "in an embodiment," "an example," or "an example" in various places throughout this specification are not necessarily all referring to the same embodiment or example. Furthermore, the particular features, structures, or characteristics may be combined in any suitable combinations and / or subcombinations in one or more embodiments or examples. Furthermore, those of ordinary skill in the art will appreciate that the figures provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0022] In an alternative embodiment, see Figure 1 , Figure 1 Flowchart of the dynamic simulation method for estuarine tidal bore propagation provided by the embodiment of the present invention. Figure 1 As shown, the estuarine tidal bore propagation inversion method includes the following steps:

[0023] S01. Select the observation location and observation area.

[0024] It should be understood that the selection of the observation area is related to actual needs, that is, the area of the observation area must include the actual estuary area to be observed. The selection of the observation position is related to the location of the observation area and the size of the field of view that the image acquisition device can capture, assuming the captured image has sufficient clarity. Furthermore, the number of observation positions can include one or more, wherein multiple observation positions are set for image acquisition devices with a small field of view.

[0025] Furthermore, since factors such as time, tides, and weather will affect the observation results, when selecting the observation location and observation area, it is also necessary to record the objective conditions when collecting the river tide line image. Specifically, the objective conditions include time conditions, tidal conditions, and meteorological conditions, wherein the time conditions refer to observations in different time periods, such as sunrise and sunset, day and night, etc.; the tidal conditions refer to tidal water levels at different heights, such as high tide and low tide, etc.; the meteorological conditions refer to meteorological parameters such as wind force, temperature, humidity, and different wind directions of different numerical values. By recording the objective conditions when collecting the river tide line image, the comprehensiveness and accuracy of the collected data are guaranteed, providing strong data support for simulating more accurate tidal propagation forms and velocity fields.

[0026] S02. Setting an image acquisition device at the observation position.

[0027] The image acquisition device described in step S02 includes traditional image acquisition devices such as cameras and video cameras, but this method has disadvantages such as limited shooting field of view and inability to observe continuously for a long time. Furthermore, it is possible to use the above-mentioned traditional image acquisition device on flying equipment such as drones to expand the shooting field of view, but this image acquisition method is easily affected by environmental factors. The image acquisition device described in step S02 also includes remote sensing satellites, but the clarity of the river surface tidal bore images obtained by remote sensing satellites is easily affected by weather to a certain extent, and the accuracy is not high. In order to obtain more comprehensive and high-resolution river surface tidal bore images, the image acquisition device described in step S02 also includes an X-band radar telemetry device, that is, the river surface tidal head line image is an X-band radar reflection image captured by the X-band radar telemetry device. The X-band radar telemetry device can obtain river surface tidal head line images with a wide area and high resolution in real time, and the X-band radar telemetry device also performs well under different weather conditions.

[0028] S03: Using the image acquisition device to acquire river surface tidal line images within the observation area.

[0029] In an optional embodiment, the image acquisition device is selected as an X-band radar telemetry device, and the step S03 described in using the image acquisition device to acquire the river surface tidal line image within the observation area is specifically as follows: the X-band radar telemetry device is mounted on a movable vehicle; the movable vehicle is driven to the observation position, and further, the observation position is the land closest to the observation area; objective conditions and changes in objective conditions are recorded; and the river surface tidal line image is acquired. Figure 2 As shown, Figure 2 This is an image of the river tidal line collected by an X-band radar telemetry device according to an embodiment of the present invention, where A represents the river bank line and B represents the tidal line.

[0030] S04. Arrange and summarize the river surface tidal line images to obtain a river surface tidal line image dataset.

[0031] It should be understood that step S02 is a foundational step for subsequent training of the feature extraction network and the tidal wave prediction network. Its primary task is to process and annotate the collected river surface tidal line images to ensure that the obtained river surface tidal line image dataset has good quality and usability. Furthermore, the process of collating and summarizing the river surface tidal line images in step S04 to obtain the river surface tidal line image dataset includes the following steps:

[0032] S041. Cleaning the river surface tidal line image and unifying the format of the cleaned river surface tidal line image. Furthermore, the cleaning process includes performing denoising, cropping, and correction on the river surface tidal line image to filter out non-tidal line noise and other interference factors in the river surface tidal bore image, thereby improving the quality and accuracy of the river surface tidal line image.

[0033] S042: Label the cleaned and formatted river tidal line images to obtain a river tidal line image dataset. The feature information further includes the location features of the tidal line and the time features of the acquisition of the river tidal line images.

[0034] S043, storing the river surface tidal line image dataset. Step S043 stores the river surface tidal line image dataset in a suitable location for subsequent training and application of a deep learning network.

[0035] Commonly used methods for simulating the dynamic propagation of tidal bores in estuaries typically use mathematical models to describe the propagation process, requiring extensive field data to validate and optimize the models. The method provided in step S02, however, uses collected river surface tidal crest line images to represent the measured data. This, combined with a subsequent deep learning network, allows for direct simulation and prediction of tidal wave propagation, demonstrating high accuracy and practicality.

[0036] S05. Building a feature extraction network, and using the river tidal line image dataset to train the feature extraction network.

[0037] In order to better and more quickly extract the positional features of the tidal crest line in a river tidal crest line image, the present invention constructs a feature extraction network based on a deep learning model and trains the feature extraction network using the river tidal crest image dataset. Ultimately, the trained feature extraction network is used to extract the positional features of the tidal crest line in any river tidal crest line image. In an optional embodiment, the construction of the feature extraction network described in step S05 and the training of the feature extraction network using the river tidal crest line image dataset include the following steps:

[0038] S051. Construct a feature extraction network based on a convolutional neural network. Convolutional neural network (CNN) is a commonly used, easy-to-master, and easy-to-build deep learning model. Furthermore, during the design process, it is necessary to determine hyperparameters such as the number of network layers, convolution kernel size, pooling method, and model parameters such as activation function. In another optional embodiment, a feature extraction network is constructed based on a convolutional neural network, including: 4 convolution layers, 4 pooling layers, and 2 fully connected layers, the output of the convolution layer is the pooling layer, and the fully connected layer is the first fully connected layer and the second fully connected layer; the convolution layer includes a 3×3 convolution kernel and a ReLU activation function; the stride of the pooling layer is 2×2; the first fully connected layer contains 128 neurons and a ReLU activation function respectively; the second fully connected layer contains 64 neurons and a ReLU activation function respectively; the loss function of the feature extraction network is a cross-entropy loss function. The feature extraction network constructed in this embodiment can effectively meet the requirements for extracting the position features of the tide line in the river tide line image.

[0039] S052. Set a division ratio, and divide the river surface tidal head line image dataset into a river surface tidal head line image training set, a river surface tidal head line image validation set, and a river surface tidal head line image test set according to the division ratio. The division ratio is set according to actual needs. In this embodiment, the river surface tidal head line image dataset is divided according to the ratio of the number of river surface tidal head line images of 7:2:1, that is, the river surface tidal head line image training set accounts for 70% of the total dataset, the river surface tidal head line image validation set accounts for 20% of the total dataset, and the river surface tidal head line image test set accounts for 10% of the total dataset.

[0040] S053, using the river surface tidal line image training set to train the feature extraction network, using the river surface tidal line image validation set to adjust the hyperparameters of the trained feature extraction network, and using the river surface tidal line image test set to evaluate the feature extraction network after the hyperparameter adjustment. Further, based on the feature extraction network built in step S051, step S053 uses the Adam optimization algorithm to train the feature extraction network during the training of the feature extraction network, sets the initial learning rate to 0.001, and the number of training rounds to 100; and performs regularization and early stopping on the feature extraction network during the training process to prevent overfitting; uses the river surface tidal line image validation set to adjust the hyperparameters of the feature extraction network; and uses the river surface tidal line image test set to evaluate the trained feature extraction network, for example, calculating the accuracy, recall rate, F1 value and other indicators of the feature extraction network to evaluate the performance of the feature extraction network. If the performance of the feature extraction network is not ideal, continue to adjust the network hyperparameters or change the network structure, and retrain and evaluate the corresponding feature extraction network.

[0041] Through steps S051 to S053, this embodiment establishes a feature extraction network for extracting positional features from river tidal line images, thereby providing powerful software support for inverting the tidal bore propagation velocity field. Furthermore, the feature extraction network can automatically learn features from river tidal line images, thus avoiding the difficulty and uncertainty associated with manually designing positional features. This embodiment also improves the performance of the feature extraction network by adjusting hyperparameters based on a validation set of river tidal line images, and can use a test set of river tidal line images to evaluate the network's generalization ability.

[0042] S06: Select a target tidal head line, and use the image acquisition device to obtain a target tidal head line image containing the target tidal head line.

[0043] It should be understood that the target tide head line can be any tide head that appears in the observation area of the river surface. In order to better simulate the morphology and velocity field of the global tidal bore propagation in the observation area, the target tide head line is further selected as the tide head that has just entered the observation area, or the tide head that has just been generated in the observation area. At the same time, the target tide head line image containing the target tide head line in step S06 represents multiple frames of continuous target tide head line images, for example, multiple frames of continuous target tide head line images from the time when the target tide head just enters the observation area until the target tide head disappears at the estuary, wherein the minimum recording time interval of two adjacent target tide head line images is determined by the performance of the image acquisition device.

[0044] Furthermore, in order to better utilize the feature extraction network to extract the position features in the target tidal head line image, the step S06 of selecting the target tidal head line and using the image acquisition device to obtain the target tidal head line image containing the target tidal head line also includes the following steps: cleaning the target tidal head line image and unifying the format of the cleaned target tidal head line image. Among them, the format unification is mainly aimed at the resolution of the target tidal head line image, that is, the target tidal head line image after format unification can be used as the input of the feature extraction network. Furthermore, the cleaning process includes denoising, cropping and correction of the river surface tidal head line image to filter out non-tidal head line noise and other interference factors in the river surface tidal surge image, thereby improving the quality and accuracy of the river surface tidal head line image.

[0045] S07. Utilizing the trained feature extraction network, extract the position features of the target tidal head line in the target tidal head line image.

[0046] Step S07 inputs the multiple frames of continuous target tide head line images obtained in step S06 into the trained feature extraction network, uses the trained feature extraction network to extract the position features of the target tide head line in the corresponding target tide head line image, and marks the position features in the corresponding target tide head image and outputs them to facilitate the visualization of the tidal propagation velocity field in the subsequent observation area.

[0047] S08. Obtaining a velocity vector of the target tidal head line according to positional features and time intervals of the target tidal head line images that are adjacent in time sequence.

[0048] It should be understood that the step S08 of obtaining the velocity vector of the target tidal head line according to the positional features and time intervals of the target tidal head line images that are adjacent in time sequence is a calculation operation based on the corresponding positional features in the two target tidal head line images that are adjacent in time sequence and the time intervals used for the two to change. In an optional embodiment, the step of obtaining the velocity vector of the target tidal head line according to the positional features and time intervals of the target tidal head line images that are adjacent in time sequence comprises the following steps: obtaining the positional features of the target tidal head line image that is earlier in time sequence; in, n represents the total number of position features, Represents the i-th position feature in the target tide head line image at time T0, which is earlier in the time series Obtain the position characteristics of the target tide head line image at a later time sequence in, n represents the total number of position features, Represents the i-th position feature in the target tide line image at time T1, which is later in the time sequence Extract the corresponding position features of the earlier and later time series respectively, and obtain the unidirectional velocity vector V between the two position features i , the unidirectional velocity vector V i Satisfies the following formula: in, The horizontal coordinate of the i-th position feature at the later time T0, The horizontal coordinate of the i-th position feature at the earlier time T1; The vertical coordinate of the i-th position feature at the later time T0, The vertical coordinate of the i-th position feature at time T1, which is earlier in the time sequence; the unidirectional velocity vector V corresponding to any position feature in the target tide line image adjacent to the time sequence i , obtain the velocity vector V of the target tide head line, which satisfies the following formula: V={V1,V2,...,V i ,...,V n}. See Figure 3 , Figure 3 Schematic diagram of the velocity vector of the target tidal head line obtained by an embodiment of the present invention, where the upper and lower dotted lines represent the river banks, the arrows represent the direction of the tidal head line, and the circles represent the position features at the front of the time series. The triangle represents the position feature at the end of the time series The line between the two represents a unidirectional velocity vector, and the overall irregular line represents the velocity vector of the target tide head line. At the same time, the overall irregular line also represents the propagation state of the tide surge.

[0049] S09: Summarize the velocity vectors of the target tide head lines in the observation area to obtain the tidal bore propagation velocity field in the observation area.

[0050] Step S09 of the present invention obtains the tidal bore propagation velocity field of the observation area by summarizing the velocity vectors corresponding to multiple frames of continuous target tidal head line images. The real-time estuarine tidal bore propagation velocity field information obtained in step S09 has important application value in the fields of studying estuarine tidal movement and estuarine environmental protection. Furthermore, the step S09 of summarizing the velocity vectors of the target tidal head line in the observation area to obtain the tidal bore propagation velocity field of the observation area includes the following steps:

[0051] S091: Aggregating the velocity vectors from multiple frames of target tidal head line images within the observation area into a single observation area image. Step S091 superimposes the velocity vector information from multiple frames of target tidal head line images, providing more comprehensive and accurate information about the tidal bore propagation velocity field within the observation area. This step also effectively removes isolated and inaccurate velocity vector information, improving the reliability and accuracy of the tidal bore propagation velocity field information.

[0052] S092. Obtain a tidal bore propagation image of the observation area according to the direction of the velocity vector. Step S092 draws a tidal bore propagation image according to the direction of the velocity vector, which intuitively reflects the direction and range of tidal bore propagation.

[0053] S093, obtaining a tidal flow velocity image of the observation area according to the magnitude of the velocity vector. Step S093 draws a tidal flow velocity image according to the magnitude of the velocity vector, which intuitively reflects the magnitude differences of tidal flow velocities in different areas.

[0054] S094: Combine the tidal bore propagation image and the tidal bore velocity image to obtain a tidal bore propagation velocity field in the observation area. Step S094 combines the tidal bore propagation image and the tidal bore velocity image to obtain a tidal bore propagation velocity field that comprehensively reflects the direction, range, and velocity of tidal bore propagation. Furthermore, step S094 can also visualize the tidal bore propagation velocity field to facilitate intuitive observation and analysis by researchers.

[0055] In this embodiment, steps S091-S094 integrate and analyze the velocity vector information corresponding to multiple frames of continuous target tide head line images to obtain real-time tidal bore propagation velocity field information; at the same time, the obtained tidal bore propagation velocity field has strong real-time performance and comprehensive information, and has important application value for studying estuarine tidal movement, estuarine environmental protection and other fields.

[0056] The present invention inverts the morphology and velocity field of tidal bore propagation with high real-time performance, high accuracy and high reliability through the selection of observation positions and observation areas, the collection of river surface tidal head line images, the training of feature extraction networks and the calculation of tidal bore propagation velocity field.

[0057] To further improve the accuracy of the inverted tidal bore propagation velocity field, see Figure 4 , Figure 4 The flowchart of improving the accuracy of the inverted tidal bore propagation velocity field provided by the embodiment of the present invention is as follows. Figure 4 As shown, based on Figure 1 A flow chart of a dynamic simulation method for estuarine tidal bore propagation is provided, wherein the step of obtaining the velocity vector of the target tidal head line further includes the following steps:

[0058] S081. Build a tidal line prediction network, and use the river surface tidal line image dataset to train the tidal line prediction network. Furthermore, the method of building a tidal line prediction network in step S081, and the method of training the tidal line prediction network using the river surface tidal line image dataset, can be implemented after targeted adjustments with reference to the building and training methods of steps S051 to S053. For example, the tidal line prediction network mainly learns the time-varying characteristics of the tidal line position features of the river surface tidal line image in the river surface tidal line image dataset under corresponding objective conditions, and predicts the tidal line position features of the next time step based on the time-varying characteristics. Therefore, the tidal line prediction network can be built based on deep learning network models such as RNN convolutional neural network, LSTM (Long Short-Term Memory), and GRU (Gated Recurrent Unit).

[0059] S082. For the positional features of the target tidal head line in any target tidal head line image, use the trained tidal head line prediction network to predict the predicted positional features of the target tidal head line after one time step, wherein the time step is less than the time interval. It should be understood that due to factors such as the performance limitations of the image acquisition device, the performance of the feature extraction network, and the processing of the target tidal head line image, the positional features representing the target tidal head line image may be missing or insufficient to a certain extent. Therefore, step S082 is to obtain a more complete positional feature of the target tidal head line image to simulate a more comprehensive and higher-precision tidal propagation velocity field. The time step is the difference between the time point of the target tidal head line image predicted by the tidal head line prediction network and the time point of the input target tidal head line image, and the time interval is the time difference between two adjacent frames of the target tidal head line image captured by the image acquisition device.

[0060] S083: inserting the predicted position feature into the position feature of the target tidal head line image adjacent in time sequence. By inserting the predicted position feature corresponding to the target tidal head line into the position features corresponding to two target tidal head lines with a longer time interval, the deficiency of the position feature of the target tidal head line image is compensated.

[0061] S084. Obtain a velocity vector of the target tidal head line according to the position features, predicted position features, time intervals, and time sequence steps of the target tidal head line images that are adjacent in time sequence.

[0062] Similarly, the step S084 described in which the velocity vector of the target tidal line is obtained based on the position features, predicted position features, time intervals and time step lengths of the target tidal line images that are adjacent in time sequence is calculated based on the corresponding position features in the two target tidal line images that are adjacent in time sequence and the time step lengths used for the changes of the two. In an optional embodiment, the step S084 described in which the velocity vector of the target tidal line is obtained based on the position features, predicted position features, time intervals and time step lengths of the target tidal line images that are adjacent in time sequence comprises the following steps: obtaining the position features of the target tidal line image that is earlier in time sequence; in, n represents the total number of position features, Represents the i-th position feature in the target tide head line image at time T0, which is earlier in the time sequence Obtain the predicted position features of the target tide head line image that is earlier in the time series in, n represents the total number of position features, Represents the i-th predicted position feature of the predicted target tide head line at the prediction time t0 Extract the position features corresponding to the earlier time series and the predicted time respectively, and obtain the unidirectional velocity vector V between the position features of the earlier time series and the position features of the predicted time i, the unidirectional velocity vector V i Satisfies the following formula: T1-T0=ΔT, t0-T0=Δt, T1-t0≤Δt, where T1 represents the later moment, ΔT represents the time interval, and Δt represents the time step. Represents the horizontal coordinate of the i-th position feature at the prediction time t0, The horizontal coordinate of the i-th position feature at the time T0, which is earlier in the time series, The ordinate of the i-th position feature at the prediction time t0; The vertical coordinate of the i-th position feature at the time T0 that is ahead in the time sequence; the one-way velocity vector V corresponding to any position feature in the target tidal head line image ahead in the time sequence and the predicted target tidal head line is summarized. i , obtain the velocity vector V of the target tide head line, which satisfies the following formula: V={V1,V2,...,V i ,...,V n}. See Figure 5 , Figure 5 Schematic diagram of the velocity vector of the target tidal head line after inserting the predicted position feature in an embodiment of the present invention, where the upper and lower dotted lines represent the river bank, the arrows represent the direction of the tidal head line, and the circles represent the position features at the front of the time series. The triangle represents the position feature at the end of the time series Squares represent predicted position features The line between the two represents a unidirectional velocity vector, and the overall irregular line represents the velocity vector of the target tide head line. At the same time, the overall irregular line also represents the propagation state of the tide surge.

[0063] In order to further improve the accuracy of the inverted tidal propagation velocity field, multiple sets of predicted position features can be inserted into the target tide head line images adjacent to each other in two time series. Furthermore, among the multiple sets of predicted position features, the predicted time t m The time difference between the time step and the later time step T1 is less than or equal to the above time step.

[0064] The present invention incorporates a tidal head prediction network to predict the positional characteristics of the target tidal head line, thereby obtaining the velocity vector of the target tidal head line. This not only solves the problem of insufficient acquisition of the target tidal head line positional characteristics within the observation area, but also greatly improves the accuracy of the inverted tidal bore propagation velocity field, while also increasing the ability to predict future tidal bore propagation conditions, and improving the accuracy and reliability of the simulation results.

[0065] In order to better implement the above-mentioned estuarine tidal bore propagation inversion method, in an optional embodiment, see Figure 6 , Figure 6This is a structural diagram of the estuary tidal bore propagation inversion system provided by an embodiment of the present invention. Figure 6 As shown, the present invention also provides an estuarine tidal bore propagation inversion system, comprising one or more processors; one or more input devices; one or more output devices; and a memory. The processors, the input devices, the output devices, and the memory are connected via a bus. The memory is used to store a computer program, the computer program including program instructions. The processor is configured to call the program instructions to execute the estuarine tidal bore propagation inversion method provided by the first aspect of the present invention. The estuarine tidal bore propagation inversion system provided by the present invention has a compact structure and stable performance, and can efficiently and accurately implement the estuarine tidal bore propagation inversion method.

[0066] In an optional embodiment, the processor 601 can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or it can be any conventional processor. The input device 602 can be used to input the collected river surface tide head line image. The output device 603 can output the tidal propagation velocity field obtained by simulating the method of the present invention to the target terminal for display. The memory 604 can include a read-only memory and a random access memory, and provides instructions and data to the processor 601. A portion of the memory 604 can also include a non-volatile random access memory. For example, the memory 604 can also store information about the device type.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and description of the present invention.

Claims

1. A method for inverting estuarine tidal bore propagation, characterized in that: The estuary tidal bore propagation inversion method comprises the following steps: Select the observation location and observation area; An image acquisition device is provided at the observation position; Using the image acquisition device to acquire an image of the river tidal head line within the observation area; Arranging and summarizing the river surface tidal line images to obtain a river surface tidal line image dataset; Building a feature extraction network, and using the river tidal line image dataset to train the feature extraction network; Selecting a target tidal head line, and using the image acquisition device to obtain a target tidal head line image containing the target tidal head line; Using the trained feature extraction network, extracting the position features of the target tidal head line in the target tidal head line image; According to the position characteristics and time intervals of the target tidal line images adjacent to each other in time sequence, the velocity vector of the target tidal line is obtained; Summarizing the velocity vectors of the target tide head lines in the observation area to obtain the tidal bore propagation velocity field in the observation area; The method of obtaining the velocity vector of the target tidal line according to the position characteristics and time intervals of the target tidal line images adjacent in time sequence comprises the following steps: Obtain the position characteristics of the target tide head line image at the front of the time series ,in, , represents the total number of position features, Indicates the earlier time sequence The target tide head line image at the moment Location features ; Obtain the position characteristics of the target tide head line image at a later time sequence ,in, , represents the total number of position features, Indicates later timing The target tide head line image at the moment Location features ; Extract the corresponding position features of the earlier and later time series respectively, and obtain the unidirectional velocity vector between the two position features , the unidirectional velocity vector Satisfies the following formula: ,in, Indicates the earlier time sequence Moment The horizontal coordinate of the position feature, Indicates later timing Moment The horizontal coordinate of the position feature; Indicates the earlier time sequence Moment The vertical coordinate of the position feature, Indicates later timing Moment The vertical coordinate of the position feature; Summarize the unidirectional velocity vector corresponding to any position feature in the target tide head line image adjacent to the time sequence , obtain the velocity vector of the target tide head line , the velocity vector Satisfies the following formula: .

2. The estuarine tidal bore propagation inversion method according to claim 1, characterized in that: The method of obtaining the velocity vector of the target tidal line according to the position characteristics and time intervals of the target tidal line images adjacent in time sequence further includes the following steps: Building a tidal head line prediction network, and using the river surface tidal head line image dataset to train the tidal head line prediction network; For the position features of the target tidal head line in any target tidal head line image, using the trained tidal head line prediction network to predict the predicted position features of the target tidal head line after a time step, wherein the time step is less than the time interval; Inserting the predicted position feature into the position features of the target tide head line images adjacent in time sequence; The velocity vector of the target tidal head line is obtained according to the position characteristics, predicted position characteristics, time interval and time step of the target tidal head line images adjacent in time sequence.

3. The estuarine tidal bore propagation inversion method according to claim 1, characterized in that: The image acquisition device includes an X-band radar telemetry device.

4. The estuarine tidal bore propagation inversion method according to claim 1, characterized in that: The step of arranging and summarizing the river surface tidal line images to obtain a river surface tidal line image dataset comprises the following steps: cleaning the river surface tidal line image and unifying the format of the cleaned river surface tidal line image; Label the feature information of the river surface tidal line images after cleaning and formatting to obtain a river surface tidal line image dataset; The river surface tidal crest line image dataset is stored.

5. The estuarine tidal bore propagation inversion method according to claim 1, characterized in that: The step of building a feature extraction network and training the feature extraction network using the river tidal line image dataset comprises the following steps: Build a feature extraction network based on convolutional neural network; Setting a division ratio, and dividing the river surface tidal head line image dataset into a river surface tidal head line image training set, a river surface tidal head line image verification set, and a river surface tidal head line image test set according to the division ratio; The feature extraction network is trained using the river surface tidal line image training set, the hyperparameters of the trained feature extraction network are adjusted using the river surface tidal line image verification set, and the feature extraction network after hyperparameter adjustment is evaluated using the river surface tidal line image test set.

6. The estuarine tidal bore propagation inversion method according to claim 5, characterized in that: The feature extraction network includes: 4 convolutional layers, 4 pooling layers and 2 fully connected layers, the output of the convolutional layer is the pooling layer, and the fully connected layers are the first fully connected layer and the second fully connected layer; The convolution layer includes a 3×3 convolution kernel and a ReLU activation function; The stride of the pooling layer is 2×2; The first fully connected layer contains 128 neurons and a ReLU activation function; The second fully connected layer contains 64 neurons and a ReLU activation function; The loss function of the feature extraction network is a cross entropy loss function.

7. The estuarine tidal bore propagation inversion method according to claim 2, characterized in that: The method of obtaining the velocity vector of the target tidal head line according to the position characteristics, predicted position characteristics, time interval and time sequence step length of the target tidal head line images adjacent in time sequence comprises the following steps: Obtain the position characteristics of the target tide head line image at the front of the time series ,in, , represents the total number of position features, Indicates the earlier time sequence The target tide head line image at the moment Location features ; Obtain the predicted position features of the target tide head line image that is earlier in the time series ,in, , represents the total number of position features, Indicates the predicted time The predicted target tide head line Predicted location features ; Extract the position features corresponding to the earlier time series and the predicted time respectively, and obtain the unidirectional velocity vector between the position features of the earlier time series and the position features of the predicted time. , the unidirectional velocity vector Satisfies the following formula: , ,in, Indicates a later moment in time. Indicates the time interval, represents the time step, Indicates the predicted time No. The horizontal coordinate of the position feature, Indicates the earlier time sequence Moment The horizontal coordinate of the position feature, Indicates the predicted time No. The vertical coordinate of the position feature; Indicates the earlier time sequence Moment The vertical coordinate of the position feature; Summarize the one-way velocity vector corresponding to the target tidal head line image at the front of the time sequence and any position feature in the predicted target tidal head line , obtain the velocity vector of the target tide head line , the velocity vector Satisfies the following formula: .

8. The estuarine tidal bore propagation inversion method according to claim 7, characterized in that: The step of summarizing the velocity vectors of the target tide head lines in the observation area to obtain the tidal bore propagation velocity field in the observation area comprises the following steps: Aggregating velocity vectors of multiple frames of target tidal line images in the observation area into the same observation area image; obtaining a tidal propagation image of the observation area according to the direction of the velocity vector; obtaining a tidal flow velocity image of the observation area according to the magnitude of the velocity vector; The tidal surge propagation image and the tidal surge flow velocity image are combined to obtain a tidal surge propagation velocity field in the observation area.

9. An estuary tidal bore propagation inversion system, characterized in that: The estuarine tidal bore propagation inversion system includes one or more processors; one or more input devices, one or more output devices and a memory, the processor, the input device, the output device and the memory are connected via a bus, the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the estuarine tidal bore propagation inversion method according to any one of claims 1 to 8.

Images