Visual management application method and device for dike project dangerous case data

Abstract

The invention discloses a dike project dangerous case data visual management application method and device, and the method comprises the steps: constructing a dangerous case database and a drilling database based on an SQL database management system according to dike project dangerous case standardized data and drilling standardized data; based on MATLAB, constructing an embankment project dangerous case data visual management application system, and realizing dangerous case query, profile map / planar graph generation and CAD / GIS format export functions; based on the dangerous case database and the drilling database, constructing a dangerous case neural network prediction model and training the dangerous case neural network prediction model; and inputting set input data into the trained dangerous case neural network prediction model for prediction processing to obtain a dangerous case prediction result, and generating a predicted dangerous case profile map and a predicted dangerous case plane map. Visual management application of dike project dangerous cases can be achieved, dike project dangerous case design, drawing and decision making efficiency is improved, meanwhile, drill hole stratum differentiation information and drill hole multi-dimensional geological information are considered, and dangerous case prediction and risk assessment of dike projects are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of dike engineering hazard management application technology, and in particular to a dike engineering hazard data visualization management application method and device. Background Art

[0002] In levee construction, hazard data is a crucial basis for engineering design, decision-making, safety analysis, and evaluation. For proposed levee projects, understanding historical hazard information along the planned route helps accurately assess the region's geological conditions and engineering characteristics, providing data support for optimizing route site selection and enabling the preemptive development of preventative and reinforcement measures for potentially hazardous sections. For existing levee projects, hazard data collected during operation can directly reveal weaknesses in the structure, uncover potential safety hazards in existing projects, and provide a scientific basis for subsequent maintenance, reinforcement, and renovation.

[0003] Currently, due to the long construction and operation cycles of most levee projects, the wide scope of project areas involved, and the multiple phases of construction, hazard data comes from a wide range of sources and is complex. Furthermore, for older levee projects, historical hazard data is often stored in paper or scattered electronic files, lacking systematic organization. This leads to fragmented data and delayed updates. Traditional paper-based record keeping or offline systems lack flexibility and scalability, making them difficult to effectively apply in engineering projects. While some existing hazard data application systems can store and query hazard data, most remain at the static query stage, suffering from issues such as insufficient visualization capabilities and poor data interaction. Specifically, most existing hazard data management systems lack data visualization capabilities, making it impossible to intuitively view hazard distribution. In most cases, there is a lack of effective data interaction interfaces between hazard data management systems and other engineering application software (such as CAD and GIS). This necessitates repeated data writing when operating across multiple systems, impacting the efficiency of hazard data management and engineering drawing.

[0004] In addition to visualizing historical hazard information, effective hazard prediction is also a key component of levee engineering hazard management. Among the various factors influencing levee hazard occurrence, stratigraphic structure is a significant factor, and borehole data is the most direct source of data for understanding the subsurface strata. Due to the diversity and complexity of borehole data, it is difficult to consider the characteristics of the strata revealed from the perspective of a single borehole in engineering applications. Statistical analysis is typically performed on all borehole data from the same stratum to develop recommended parameter values ​​that represent the average characteristics of the stratum and serve as the basis for calculating levee cross-sectional stability and permeability. This statistical analysis method ensures uniformity and standardization in engineering design and improves design and calculation efficiency. However, this method fails to account for the complex variations in stratigraphic characteristics along the entire levee line. The differentiated stratigraphic information revealed by each borehole is not effectively exploited. Furthermore, the range of borehole parameters used in design calculations is limited, and the multidimensional geological information from the boreholes is underutilized. Among existing data analysis methods, with the continuous advancement of artificial intelligence technology, neural network methods have rapidly developed and been applied in fields such as geological hazard prediction and environmental risk identification. They are an effective method for analyzing complex data.

[0005] Therefore, how to invent an application method for visual management of embankment engineering hazard data, build a embankment engineering hazard database, and provide it with a visual hazard display function, support data interaction with other engineering application systems (CAD, GIS), and at the same time consider the differentiated stratigraphic information revealed by each borehole along the entire embankment and the multi-dimensional geological information of the boreholes to realize embankment engineering hazard prediction and risk assessment, has become an urgent problem to be solved. Summary of the Invention

[0006] To this end, the present invention provides a method and device for visual management of levee engineering hazard data. By constructing a levee engineering hazard database and providing a visual hazard display function, the method realizes the systematic and visual management application of historical hazard data, supports data interaction with other engineering application systems (CAD, GIS), and combines drilling data to propose a levee hazard prediction method based on a convolutional neural network + multi-layer perceptron (CNN+MLP) dual-branch hazard neural network, which realizes the systematic and visual management application of predicted hazard data, improves the efficiency of levee engineering project design, drawing, and decision-making, and provides a scientific basis for levee safety management and risk assessment.

[0007] To achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for visual management of embankment engineering hazard data, comprising:

[0008] By standardizing the hazard data information table structure and the borehole data information table structure, the embankment hazard data and the embankment drilling data are standardized and sorted respectively to obtain the standardized embankment hazard data and the embankment drilling data; based on the standardized embankment hazard data and the standardized embankment drilling data, a hazard database and a drilling database are constructed based on the SQL database management system;

[0009] Based on the hazard database and the borehole database, a levee engineering hazard data visualization management application system is constructed based on MATLAB; the levee engineering hazard data visualization management application system realizes hazard query, cross-section / plan view generation and CAD / GIS format export functions;

[0010] Based on the hazard database and the borehole database, a hazard neural network prediction model is constructed; the hazard neural network prediction model is trained by setting a neural network data set to obtain the trained hazard neural network prediction model;

[0011] The set input data is input into the trained dangerous situation neural network prediction model for prediction processing to obtain dangerous situation prediction results, and the dangerous situation prediction results are used to generate a predicted dangerous situation profile and a predicted dangerous situation plan.

[0012] As a preferred solution for the application method of visual management of embankment engineering hazard data, the standardized hazard data information table structure includes: occurrence time, starting pile number, ending pile number, hazard type, description information, on-site photos and water level outside the embankment when the hazard occurs;

[0013] The on-site photo is the storage path of the photo or the binary data of the photo;

[0014] The standardized drilling data information table structure includes: drilling name, pile number location, formation lithology, in-situ test data and sampling test data;

[0015] Among them, the in-situ test data include: standard penetration test number, dynamic penetration test number and water injection test data; the sampling test data include: sample permeability coefficient, particle composition, porosity, compression coefficient and strength index.

[0016] As a preferred solution of the application method for visualization management of embankment engineering hazard data, during the application processing of the visualization management application system for embankment engineering hazard data, hazard data of a specified project, a specified occurrence time, a specified pile number range, and a specified hazard type are retrieved from the hazard database through SQL query;

[0017] Generate a historical hazard profile based on the retrieved hazard data and a stake coordinate system;

[0018] Reading the centerline data of the embankment project in the CAD design drawing of the embankment project planning, generating the centerline, and generating a historical hazard plan map based on the queried hazard data and the centerline coordinate system;

[0019] Converting the elements in the historical hazard cross-section and the historical hazard plan into a CAD importable format and importing the format into a CAD platform to generate a complete historical hazard cross-section and a complete historical hazard plan;

[0020] The hazard elements in the historical hazard plan are converted into a GIS importable format and imported into the GIS platform to generate complete historical hazard line data.

[0021] As a preferred solution for the application method of visualization management of embankment engineering hazard data, the hazard neural network prediction model is a convolutional neural network + multi-layer perceptron dual-branch hazard neural network model;

[0022] The input layer of the convolutional neural network branch is matrix data describing the borehole information; the row data of the matrix data corresponds to the stratigraphic layers from the top to the bottom of the borehole; the column data corresponds to the stratigraphic characteristics of the mapped stratigraphic layers; the matrix data includes at least two columns, the first column is the stratigraphic classification value, the second column is the stratigraphic thickness, and the other columns are data reflecting the stratigraphic characteristics, including in-situ test data and sampling test data, which are arranged in a unified order;

[0023] The minimum value of the stratum classification value is 0 and the maximum value is 1. It is used to reflect the permeability of the stratum. The greater the permeability, the greater the stratum classification value, and the smaller the permeability, the smaller the stratum classification value. For strata belonging to the same category in the engineering classification, differentiated values ​​are determined based on the different characteristics of the same stratum revealed by each borehole.

[0024] The input layer of the multi-layer perceptron branch is scalar data describing the hydrological environment and structural characteristics of the levee within the drilling control range.

[0025] As a preferred solution for the application method of visualization management of embankment engineering hazard data, in the process of training the hazard neural network prediction model through the set neural network data set, the set neural network data set includes: hazard data of each drilling control range segment extracted from the hazard database; drilling data of each drilling hole extracted from the drilling database; and embankment structure feature data extracted from the embankment engineering planning CAD design drawing.

[0026] The present invention further provides a device for visual management of embankment engineering hazard data, based on the above-mentioned method for visual management of embankment engineering hazard data, comprising:

[0027] A database construction module is used to standardize and organize the embankment engineering hazard data and the embankment engineering drilling data respectively through a standardized hazard data information table structure and a standardized drilling data information table structure, thereby obtaining standardized embankment engineering hazard data and standardized embankment engineering drilling data; and construct a hazard database and a drilling database based on the SQL database management system according to the standardized embankment engineering hazard data and the standardized embankment engineering drilling data;

[0028] A module for constructing a levee engineering hazard data visualization management application system is used to construct a levee engineering hazard data visualization management application system based on MATLAB according to the hazard database and the borehole database; the levee engineering hazard data visualization management application system realizes hazard query, cross-section / plan view generation, and CAD / GIS format export functions;

[0029] A dangerous situation neural network prediction model construction and training module is used to construct a dangerous situation neural network prediction model based on the dangerous situation database and the drilling database; train the dangerous situation neural network prediction model by setting a neural network data set to obtain the trained dangerous situation neural network prediction model;

[0030] The hazard prediction result output module is used to input the set input data into the trained hazard neural network prediction model for prediction processing, obtain the hazard prediction result, and generate the predicted hazard profile and predicted hazard plan view from the hazard prediction result.

[0031] As a preferred solution for a device for visual management of embankment engineering hazard data, in the database construction module, the standardized hazard data information table structure includes: occurrence time, starting pile number, ending pile number, hazard type, description information, on-site photos, and water level outside the embankment when the hazard occurs;

[0032] The on-site photo is the storage path of the photo or the binary data of the photo;

[0033] The standardized drilling data information table structure includes: drilling name, pile number location, formation lithology, in-situ test data and sampling test data;

[0034] Among them, the in-situ test data include: standard penetration test number, dynamic penetration test number and water injection test data; the sampling test data include: sample permeability coefficient, particle composition, porosity, compression coefficient and strength index.

[0035] As a preferred solution of a levee engineering hazard data visualization management application device, in the levee engineering hazard data visualization management application system construction module, during the application processing of the levee engineering hazard data visualization management application system, hazard data of a specified project, a specified occurrence time, a specified pile number range, and a specified hazard type are retrieved from the hazard database through SQL query;

[0036] Generate a historical hazard profile based on the retrieved hazard data and a stake coordinate system;

[0037] Reading the centerline data of the embankment project in the CAD design drawing of the embankment project planning, generating the centerline, and generating a historical hazard plan map based on the queried hazard data and the centerline coordinate system;

[0038] Converting the elements in the historical hazard cross-section and the historical hazard plan into a CAD importable format and importing the format into a CAD platform to generate a complete historical hazard cross-section and a complete historical hazard plan;

[0039] The hazard elements in the historical hazard plan are converted into a GIS importable format and imported into the GIS platform to generate complete historical hazard line data.

[0040] As a preferred solution for the visualization management application device of embankment engineering hazard data, in the hazard neural network prediction model construction and training module, the hazard neural network prediction model is a convolutional neural network + multi-layer perceptron dual-branch hazard neural network model;

[0041] The input layer of the convolutional neural network branch is matrix data describing the borehole information; the row data of the matrix data corresponds to the stratigraphic layers from the top to the bottom of the borehole; the column data corresponds to the stratigraphic characteristics of the mapped stratigraphic layers; the matrix data includes at least two columns, the first column is the stratigraphic classification value, the second column is the stratigraphic thickness, and the other columns are data reflecting the stratigraphic characteristics, including in-situ test data and sampling test data, which are arranged in a unified order;

[0042] The minimum value of the stratum classification value is 0 and the maximum value is 1. It is used to reflect the permeability of the stratum. The greater the permeability, the greater the stratum classification value, and the smaller the permeability, the smaller the stratum classification value. For strata belonging to the same category in the engineering classification, differentiated values ​​are determined based on the different characteristics of the same stratum revealed by each borehole.

[0043] The input layer of the multi-layer perceptron branch is scalar data describing the hydrological environment and structural characteristics of the levee within the drilling control range.

[0044] As a preferred solution for an application device for visualization management of embankment engineering hazard data, in the hazard neural network prediction model construction and training module, in the process of training the hazard neural network prediction model through the set neural network data set, the set neural network data set includes: hazard data of each drilling control range segment extracted from the hazard database; drilling data of each drilling hole extracted from the drilling database; and embankment structure feature data extracted from the embankment engineering planning CAD design drawing.

[0045] The present invention has the following advantages: Based on the SQL database management system, the present invention establishes a hazard database and a borehole database, which can integrate the hazard information and borehole information of the embankment project, form a unified data standard, and avoid information loss and errors. At the same time, based on the MATLAB platform, a visualization management application system for embankment project hazard data is developed, a graphical management interface is created, and data interaction and management are realized. The required hazard data can be quickly located according to time, space, hazard type and other conditions, thereby improving the efficiency of classified acquisition of hazard information. Based on the hazard information of the hazard database and the borehole information of the borehole database, the present invention constructs a convolutional neural network + multi-layer perceptron (CNN+MLP) dual-branch hazard neural network model, which can effectively utilize the differentiated information of the borehole strata and the multi-dimensional geological information of the borehole to simulate the occurrence of various types of hazards under different conditions, realize the segmented prediction of embankment hazards, and effectively serve the safety management and disaster warning work of the embankment project. The present invention, combined with the plot analysis function of the MATLAB platform, can generate historical hazard profiles and historical hazard plan views under different query conditions from the hazard data retrieved from the hazard database, and generate predicted hazard profiles and predicted hazard plan views under different simulation conditions from the hazard data predicted in the hazard neural network model, thereby achieving a visual display of hazard data and reflecting the spatial distribution characteristics of hazards in an intuitive and graphical manner. The present invention creates a data interaction interface between the levee engineering hazard data visualization management application system and other engineering application software (CAD, GIS), allowing historical hazard information and predicted hazard information to be directly generated on CAD design drawings and GIS platforms without the need to repeatedly write hazard data, effectively improving the efficiency of levee engineering hazard design, drawing, and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and those skilled in the art can, without inventive effort, derive other implementation drawings based on the provided drawings.

[0047] The structures, proportions, sizes, etc. illustrated in this specification are intended solely to complement the contents disclosed herein and to facilitate understanding and reading by persons skilled in the art. They are not intended to limit the conditions under which the present invention may be implemented and therefore have no substantive technical significance. Any structural modifications, changes in proportions, or adjustments in sizes, without affecting the efficacy and objectives of the present invention, shall remain within the scope of the technical contents disclosed herein.

[0048] Figure 1 This is a flow chart of an application method for visual management of embankment engineering hazard data provided in Example 1 of the present invention;

[0049] Figure 2 This is a schematic diagram of a specific implementation process framework of a method for visual management of embankment engineering hazard data provided in Example 1 of the present invention;

[0050] Figure 3 This is a schematic diagram of the module framework of the dike engineering hazard data visualization management application system in the dike engineering hazard data visualization management application method provided in Example 1 of the present invention;

[0051] Figure 4 This is a schematic diagram of the operation interface of the levee engineering hazard data visualization management application system in the levee engineering hazard data visualization management application method provided in Example 1 of the present invention;

[0052] Figure 5 This is a schematic diagram of the structure of a dangerous situation neural network prediction model in an application method for visual management of dangerous situation data of embankment projects provided in Example 1 of the present invention;

[0053] Figure 6 A schematic cross-sectional view of a series of dike projects in dangerous conditions in 2024 in a possible embodiment provided in Example 1 of the present invention;

[0054] Figure 7 A schematic plan view of the dangerous situation of a series of embankment projects in 2024 in a possible embodiment provided in Example 1 of the present invention;

[0055] Figure 8 This is a schematic diagram of the architecture of an application device for visual management of embankment engineering hazard data provided in Example 2 of the present invention. DETAILED DESCRIPTION

[0056] The following describes the implementation of the present invention using specific embodiments. Those skilled in the art will readily understand the other advantages and benefits of the present invention from the disclosure herein. Obviously, the embodiments described are only a portion of the present invention, not all of it. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are intended to fall within the scope of protection of the present invention.

[0057] Example 1

[0058] See also Figure 1 and Figure 2 Embodiment 1 of the present invention provides a method for visual management of embankment engineering hazard data, comprising the following steps:

[0059] S1. Standardizing and arranging embankment hazard data and embankment drilling data by using a standardized hazard data information table structure and a standardized drilling data information table structure, respectively, to obtain standardized embankment hazard data and standardized embankment drilling data; and constructing a hazard database and a drilling database based on the SQL database management system according to the standardized embankment hazard data and the standardized embankment drilling data.

[0060] S2. Based on the hazard database and the borehole database, construct a levee engineering hazard data visualization management application system based on MATLAB; and implement hazard query, cross-section / plan view generation, and CAD / GIS format export functions through the levee engineering hazard data visualization management application system;

[0061] S3. Constructing a dangerous situation neural network prediction model based on the dangerous situation database and the drilling database; training the dangerous situation neural network prediction model by setting a neural network data set to obtain the trained dangerous situation neural network prediction model;

[0062] S4. Input the set input data into the trained dangerous situation neural network prediction model for prediction processing to obtain dangerous situation prediction results, and generate a predicted dangerous situation profile and a predicted dangerous situation plan view from the dangerous situation prediction results.

[0063] In this embodiment, in step S1, the embankment engineering hazard data and the embankment engineering drilling data are standardized and sorted respectively by using the standardized hazard data information table structure and the standardized drilling data information table structure to obtain the embankment engineering hazard data and the embankment engineering drilling data; based on the embankment engineering hazard data and the embankment engineering drilling data, a hazard database and a drilling database are constructed based on the SQL database management system;

[0064] The standardized hazard data information table structure includes at least the following hazard data categories: occurrence time, starting pile number, ending pile number, hazard type, description information, on-site photos, and water level outside the dike when the hazard occurs. The on-site photos are the storage path of the photos or the binary data of the photos.

[0065] The standardized drilling data information table structure contains at least the following drilling data categories: drilling name, pile number location, stratum lithology, in-situ test data, sampling test data, of which the in-situ test data at least includes standard penetration test number, dynamic penetration test number, and water injection test data, and the sampling test data at least includes the sample permeability coefficient, particle composition, porosity, compressibility, and strength index;

[0066] Based on the SQLite relational database management system, a hazard database and a drilling database are established according to the standardized hazard and drilling data information table structure.

[0067] Among them, the necessary information in the hazard database includes at least: starting pile number, ending pile number and hazard type. Hazard data lacking any of the necessary information cannot be entered into the database as a valid hazard data;

[0068] The necessary information in the drilling database includes at least: drilling name, pile number location and stratum lithology. Drilling data that lacks any of the necessary information cannot be entered into the database as a valid drilling data.

[0069] In this embodiment, in step S2, a levee engineering hazard data visualization management application system is constructed based on MATLAB according to the hazard database and the borehole database; the levee engineering hazard data visualization management application system realizes hazard query, cross-section / plan view generation, and CAD / GIS format export functions;

[0070] Specifically, such as Figure 3 As shown in the figure, MATLAB App Designer is used to develop a visualization management application system for levee engineering hazard data, and a graphical management interface for hazard data and borehole data is created. The interface includes a variety of controls for data management, modeling processing, visualization display, and interactive operations. The MATLAB database toolbox is used to connect to the hazard database and borehole database. System functions are created through MATLAB's callback function mechanism. The callback function responds to operation events in the system graphical management interface and triggers corresponding database operations, thereby completing data interaction, management, and update. Among them, the interface window of the levee engineering hazard data visualization management application system is as follows Figure 4 shown.

[0071] In this embodiment, during the application processing of the levee engineering hazard data visualization management application system, the hazard data of the specified project, specified occurrence time, specified pile number range and specified hazard type are retrieved from the hazard database through SQL query;

[0072] Generate a historical hazard profile based on the retrieved hazard data and a stake coordinate system;

[0073] Specifically, the pile number coordinate system is a coordinate system with the pile number as the horizontal coordinate. The method for generating the historical hazard profile is to use the MATLAB drawing tool to generate the pile number coordinate system, generate hazard line segments at corresponding horizontal coordinate positions according to the starting pile number and the ending pile number of the queried hazard data, and mark the corresponding occurrence time and hazard type at each hazard line segment position;

[0074] Reading the centerline data of the embankment project in the CAD design drawing of the embankment project planning, generating the centerline, and generating a historical hazard plan map based on the queried hazard data and the centerline coordinate system;

[0075] Specifically, the centerline coordinate system is consistent with the coordinate system of the embankment engineering planning CAD design drawing. The method for generating the historical hazard plan is to use MATLAB drawing tools to generate the centerline coordinate system, generate the centerline according to the centerline data, and the starting pile number and the ending pile number of the queried hazard data correspond to the starting point and the end point of the cumulative length of the centerline respectively. A hazard curve is generated between the corresponding starting and end points on the centerline, and the corresponding occurrence time and hazard type are marked at each hazard curve position.

[0076] Converting the elements in the historical hazard cross-section and the historical hazard plan into a CAD importable format and importing the format into a CAD platform to generate a complete historical hazard cross-section and a complete historical hazard plan;

[0077] Specifically, the method for converting each element in the drawn historical hazard profile and historical hazard plan into a format that can be imported into CAD is to read all the graphic element data of the historical hazard profile and historical hazard plan drawn in MATLAB, including coordinate points, line segments, curves, and annotated graphic objects. The embankment engineering hazard data visualization management application system creates corresponding LISP commands one by one according to the coordinate position, type, and color characteristics of each graphic element based on MATLAB functions, and saves the generated complete LISP command string as a .lsp file;

[0078] The hazard elements in the historical hazard plan are converted into a GIS importable format and imported into the GIS platform to generate complete historical hazard line data.

[0079] Specifically, the method for converting the hazard elements in the drawn historical hazard plan map into a format that can be imported into GIS is to read the hazard element data in the historical hazard plan map drawn by MATLAB. The embankment project hazard data visualization management application system converts the hazard element data into a string that conforms to the GeoJSON format specification based on the MATLAB function, describes the spatial position and attribute information of the element, and saves the generated complete GeoJSON format string as a .geojson file.

[0080] In this embodiment, in step S3, a hazard neural network prediction model is constructed based on the hazard database and the drilling database; the hazard neural network prediction model is trained by setting a neural network data set to obtain the trained hazard neural network prediction model;

[0081] Specifically, such as Figure 5 As shown in the figure, based on the hazard information and drilling information, a corresponding convolutional neural network + multi-layer perceptron (CNN+MLP) dual-branch hazard neural network model is constructed and trained for each hazard type. Specifically, the center line of the embankment project is first segmented, and the center positions of adjacent drilling holes are segmentation points. The pile number range contained in the adjacent segmentation points is the control range of the drilling holes within the segment. A set of neural network data sets can be constructed for each drilling control range. Each set of data sets includes the hazard data of each drilling control range segment extracted from the hazard database through SQL query, the drilling data of each drilling hole extracted from the drilling database, and the embankment structure extracted from the CAD design drawing of the embankment project planning. Characteristic data; the structure of the CNN+MLP dual-branch hazard neural network model is as follows: the CNN branch input layer is matrix data describing the borehole information, each row of the matrix data corresponds to the stratigraphic layer from the top to the bottom of the borehole, and each column corresponds to the stratigraphic characteristics of the mapped stratigraphic layer; the matrix data includes at least two columns, the first column is the stratigraphic classification value, the second column is the stratigraphic thickness, and the other columns are data reflecting the stratigraphic characteristics, such as in-situ test data and sampling test data, which are arranged in a unified order. By adjusting the dimensional structure of the matrix data, only the influence of a single stratigraphic information factor on the embankment hazard can be considered, or the influence of multi-dimensional geological information on the embankment hazard can be considered simultaneously by adding columns and combining them;

[0082] The minimum value of the stratum classification value is 0, and the maximum value is 1. It is used to reflect the permeability of the stratum. The greater the permeability, the greater the stratum classification value, and the smaller the permeability, the smaller the stratum classification value. The value is flexibly determined based on the type of embankment stratum and the permeability characteristics contained in the model training, testing, and prediction data sets. For strata belonging to the same category in the engineering classification, differentiated values ​​are determined based on the different characteristics of the same stratum revealed by each borehole.

[0083] The input layer of the MLP branch contains scalar data describing the hydrological environment and structural characteristics of the levee within the corresponding drilling control range, including but not limited to the water level outside the levee, levee height, and levee angle. The number of intermediate layers and nodes is determined by the size and complexity of the dataset. The output layer is used to predict whether a dangerous situation has occurred in the corresponding drilling control range, which is a binary classification result.

[0084] In this embodiment, after the model is constructed, each set of neural network data sets for each drilling control range segment can be used to train and test different types of dangerous situation neural network models;

[0085] Specifically, the hazard data of each drilling control range section extracted from the hazard database are classified according to the hazard type, and the classified hazard data are respectively corresponded to various types of hazard neural network models. Each type of hazard model is trained and tested in turn. For the drilling control range section where the target hazard occurs, the corresponding output layer value is "1", and for the drilling control range section where the target hazard does not occur, the output layer value is "0". All the obtained neural network data sets are randomly sampled and divided according to the preset ratio to construct training sets and test sets respectively. The training set data is input into the neural network model of the corresponding hazard type for supervised learning training, and the test set data is used to evaluate the prediction accuracy of the model. When the prediction accuracy meets the requirements of the engineering application, the model training process is completed.

[0086] In this embodiment, in step S4, the set input data is input into the trained hazard neural network prediction model for prediction processing to obtain hazard prediction results, and the hazard prediction results are used to generate a predicted hazard profile and a predicted hazard plan.

[0087] Specifically, based on the trained CNN+MLP dual-branch hazard neural network model for different hazard types, the input data in the model input layer is changed, the occurrence of various types of hazards under different conditions is simulated, and the drilling control range where hazards may occur and the corresponding hazard types are predicted. Specifically, the water level data outside the embankment in the MLP branch input layer of the model is changed, and the occurrence of hazards within the control range of each drilling hole under different water level conditions is simulated and predicted. The drilling information data in the CNN branch input layer of the model is changed, and the occurrence of hazards within the drilling control range under the stratum structure exposed by the new drilling hole is simulated and predicted; according to the hazard data predicted by the model, the predicted hazard profile and predicted hazard plan under different conditions are generated according to step S2, and converted into a format that can be imported into CAD and GIS.

[0088] Among them, the water level data outside the dike can be set based on the maximum water level value estimated each year by the hydrological and meteorological departments, and the borehole control range sections and corresponding hazard types that may occur under the annual estimated maximum water level value are predicted to obtain the annual estimated water level value-predicted hazard data set; on this basis, the step size is set to gradually increase the water level input value, and a simulation prediction is performed for each incremental water level value, and the minimum water level value outside the dike and the corresponding hazard type for the first occurrence of a hazard in each borehole control range section are recorded to obtain the predicted hazard response-water level value threshold data set.

[0089] In a possible embodiment, an example of a visualization management application of dangerous situation data of a series of embankment projects in Dongting Lake is provided as follows:

[0090] Based on the developed levee project hazard data visualization management application system, the hazard data of a series of levee projects in Dongting Lake in 2024 was retrieved from the hazard database through SQL query. The pile number coordinate system was generated using MATLAB drawing tools. The 2024 hazard data was generated according to the starting pile number and the ending pile number at the corresponding horizontal coordinate position to generate hazard line segments. The corresponding occurrence time and hazard type were marked at each hazard line segment position to generate the hazard profile of a series of levee projects in Dongting Lake in 2024, as shown in the figure below. Figure 6 As shown. Read the centerline data of the embankment project in the CAD design drawing of the embankment project planning, including the centerline number, polyline number, polyline start point X, polyline start point Y, polyline end point X, polyline end point Y, and polyline length information, and export the read data as a structured table. Import the structured table data into MATLAB, use the drawing tool to generate the centerline coordinate system, generate the centerline according to the centerline data, and the starting pile number and ending pile number of the 2024 hazard data queried correspond to the starting point and end point of the cumulative length of the centerline respectively. Generate a hazard curve between the corresponding starting and end points on the centerline, and mark the corresponding occurrence time and hazard type at each hazard curve position to generate the 2024 hazard plan of a series of embankment projects in Dongting Lake, as shown below. Figure 7 shown.

[0091] Read all the graphic element data from the 2024 hazard profile and 2024 hazard plan of a series of Dongting Lake embankment projects drawn by the system, including coordinate points, line segments, curves, and annotated graphic objects. Based on the MATLAB function, create corresponding LISP commands one by one according to the coordinate position, type, and color characteristics of each graphic element. Save the generated complete LISP command string as a .lsp file, import it into the CAD platform to automatically generate the 2024 hazard profile and 2024 hazard plan. Read the hazard element data from the 2024 hazard plan of a series of Dongting Lake embankment projects drawn by the system. Based on the MATLAB function, convert the hazard element data into a string that conforms to the GeoJSON format specification, describing the spatial position and attribute information of the element. Save the generated complete GeoJSON format string as a .geojson file, and import it into the GIS platform to automatically generate the 2024 hazard line data.

[0092] Based on the above data, a prediction model was constructed. Corresponding CNN+MLP dual-branch hazard neural network models were constructed and trained for the three types of hazard: piping, seepage, and bank collapse. The center line of a series of embankment projects in Dongting Lake was segmented, and the center positions of adjacent boreholes were segmentation points. The pile number range contained in the adjacent segmentation points was the control range of the boreholes within the segment. A set of neural network data sets can be constructed for each borehole control range. Each data set includes the hazard data of each borehole control range segment extracted from the hazard database through SQL query, the drilling data of each borehole extracted from the drilling database, and the embankment structure feature data extracted from the embankment project planning CAD design drawing. The structure of the CNN+MLP dual-branch hazard neural network model is as follows: the CNN branch input layer is the matrix data describing the borehole stratum information. Each row of the matrix data corresponds to the stratum stratum from the top to the bottom of the borehole. The first column of the matrix data is the stratum classification value, the second column is the stratum thickness, the third column is the standard penetration number, and the fourth column is the compression coefficient. The stratigraphic classification values ​​for gravel range from 0.9 to 1, silt and fine sand from 0.75 to 0.85, sandy loam from 0.6 to 0.7, loam from 0.45 to 0.55, silty loam from 0.3 to 0.45, silty clay from 0.2 to 0.3, and clay from 0 to 0.1. Strata within the same category are assigned different values ​​within this range based on the different characteristics of the same strata revealed by each borehole. The input layer of the MLP branch contains scalar data describing the hydrological environment and structural characteristics of the levee within the corresponding borehole control range, including the water level outside the levee, levee height, and levee angle. The number of intermediate layers and nodes is determined by the size and complexity of the dataset and is adjusted flexibly during data updates. The output layer is used to predict whether a dangerous situation has occurred within the corresponding borehole control range.

[0093] The constructed prediction model was trained by extracting hazard data from the hazard database for each borehole control range segment and classifying them according to hazard type: piping, seepage, and bank collapse. The classified hazard data were then mapped to a neural network model for each hazard type, and each hazard model was trained and tested sequentially. For borehole control range segments where the target hazard occurred, the corresponding output layer value was "1," while for borehole control range segments where the target hazard did not occur, the corresponding output layer value was "0." The entire neural network dataset was randomly sampled and divided into a training set and a test set in a 7:3 ratio. The training set data was input into the neural network model for the corresponding hazard type for supervised learning, and the test set data was used to evaluate the model's prediction accuracy. Data suitable for training and testing the hazard neural network model was continuously accumulated from the hazard database and the borehole database until the prediction accuracy met the engineering application requirements, completing the initial model training process. The model was then iteratively trained as data accumulated to continuously improve its prediction accuracy.

[0094] After training, predictions are made using the trained prediction model. Based on the trained CNN+MLP dual-branch hazard neural network model, the water level data outside the dike in the MLP branch input layer is changed according to the maximum water level value estimated by the hydrological and meteorological departments for that year. The matrix data describing the borehole stratum information in the CNN branch input layer is maintained, and the dike height and dike angle data describing the structural characteristics in the MLP branch input layer are updated in a timely manner. The drilling control range segments and corresponding hazard types of the three hazard types of piping, seepage, and bank collapse that may occur in a series of dike projects in Dongting Lake under the estimated maximum water level value of that year are predicted, and the annual estimated water level value-predicted hazard data set is obtained.

[0095] The predicted hazard data was visualized. Based on the levee project hazard data visualization management application system, MATLAB drawing tools were used to generate a pile number coordinate system. Each hazard data in the annual estimated water level value-predicted hazard data set was used to generate hazard line segments at the corresponding horizontal coordinate positions based on the starting and ending pile numbers. The corresponding occurrence time and hazard type were annotated at each hazard line segment position to generate a hazard profile diagram for a series of levee projects in Dongting Lake that year. The levee project centerline data was read from the levee project planning CAD design drawing and imported into MATLAB. The centerline coordinate system was generated using drawing tools. A centerline was generated based on the centerline data. The starting and ending pile numbers of each hazard data in the annual estimated water level value-predicted hazard data set corresponded to the starting and ending points of the centerline's cumulative length, respectively. A hazard curve was generated between the corresponding starting and ending points on the centerline. The corresponding occurrence time and hazard type were annotated at each hazard curve position to generate a hazard plan diagram for a series of levee projects in Dongting Lake that year.

[0096] Interactive visualization of predicted hazard data; reads all graphic element data from the predicted hazard profile and plan view of a series of levee projects in Dongting Lake drawn by the system, including coordinate points, line segments, curves, and annotated graphic objects; creates corresponding LISP commands one by one based on the coordinate position, type, and color characteristics of each graphic element using MATLAB functions; saves the generated complete LISP command string as a .lsp file, and imports it into the CAD platform to automatically generate the predicted hazard profile and plan view of the same year. Reads the hazard element data from the predicted hazard plan view of a series of levee projects in Dongting Lake drawn by the system; converts the hazard element data into a string that conforms to the GeoJSON format specification based on the MATLAB function, describing the spatial position and attribute information of the element; saves the generated complete GeoJSON format string as a .geojson file, and imports it into the GIS platform to automatically generate the predicted hazard line data of the same year.

[0097] In summary, the present invention has the following advantages: the present invention is based on the SQL database management system, establishes a hazard database and a borehole database, can integrate the hazard information and borehole information of the levee project, form a unified data standard, and avoid information loss and errors. At the same time, based on the MATLAB platform, a visualization management application system for levee project hazard data is developed, a graphical management interface is created, and data interaction and management are realized. The required hazard data can be quickly located according to time, space, hazard type and other conditions, thereby improving the efficiency of classified acquisition of hazard information. Based on the hazard information of the hazard database and the borehole information of the borehole database, the present invention constructs a convolutional neural network + multi-layer perceptron (CNN+MLP) dual-branch hazard neural network model, which can effectively utilize the differentiated information of the borehole strata and the multi-dimensional geological information of the borehole to simulate the occurrence of various types of hazards under different conditions, realize the segmented prediction of levee hazards, and effectively serve the safety management and disaster warning work of the levee project. The present invention, combined with the plot analysis function of the MATLAB platform, can generate historical hazard profiles and historical hazard plan views under different query conditions from the hazard data retrieved from the hazard database, and generate predicted hazard profiles and predicted hazard plan views under different simulation conditions from the hazard data predicted in the hazard neural network model, thereby achieving a visual display of hazard data and reflecting the spatial distribution characteristics of hazards in an intuitive and graphical manner. The present invention creates a data interaction interface between the levee engineering hazard data visualization management application system and other engineering application software (CAD, GIS), allowing historical hazard information and predicted hazard information to be directly generated on CAD design drawings and GIS platforms without the need to repeatedly write hazard data, effectively improving the efficiency of levee engineering hazard design, drawing, and decision-making.

[0098] It should be noted that the method of the embodiments of the present disclosure can be performed by a single device, such as a computer or server. The method of the embodiments of the present disclosure can also be applied in a distributed scenario, where multiple devices cooperate to perform the method. In such a distributed scenario, one of the multiple devices may only perform one or more steps of the method of the embodiments of the present disclosure, and the multiple devices will interact with each other to complete the method.

[0099] It should be noted that the above description is limited to some embodiments of the present disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in an order different from that described in the above embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0100] Example 2

[0101] See also Figure 8 Embodiment 2 of the present invention further provides an application device for visual management of embankment engineering hazard data, comprising:

[0102] Database construction module 001 is used to standardize and organize embankment hazard data and embankment drilling data respectively by using a standardized hazard data information table structure and a standardized drilling data information table structure to obtain standardized embankment hazard data and standardized embankment drilling data; and construct a hazard database and a drilling database based on the SQL database management system according to the standardized embankment hazard data and the standardized embankment drilling data.

[0103] The dike engineering hazard data visualization management application system construction module 002 is used to construct the dike engineering hazard data visualization management application system based on MATLAB according to the hazard database and the borehole database; the dike engineering hazard data visualization management application system realizes hazard query, cross-section / plan view generation and CAD / GIS format export functions;

[0104] Dangerous situation neural network prediction model construction and training module 003 is used to construct a dangerous situation neural network prediction model based on the dangerous situation database and the drilling database; train the dangerous situation neural network prediction model by setting a neural network data set to obtain the trained dangerous situation neural network prediction model;

[0105] The hazard prediction result output module 004 is used to input the set input data into the trained hazard neural network prediction model for prediction processing, obtain the hazard prediction result, and generate the predicted hazard profile and predicted hazard plan from the hazard prediction result.

[0106] In this embodiment, in the database construction module 001, the standardized hazard data information table structure includes: occurrence time, starting pile number, ending pile number, hazard type, description information, on-site photos and water level outside the dike when the hazard occurs;

[0107] The on-site photo is the storage path of the photo or the binary data of the photo;

[0108] The standardized drilling data information table structure includes: drilling name, pile number location, formation lithology, in-situ test data and sampling test data;

[0109] Among them, the in-situ test data include: standard penetration test number, dynamic penetration test number and water injection test data; the sampling test data include: sample permeability coefficient, particle composition, porosity, compression coefficient and strength index.

[0110] In this embodiment, in the dike engineering hazard data visualization management application system construction module 002, during the application processing of the dike engineering hazard data visualization management application system, hazard data of a specified project, a specified occurrence time, a specified pile number range, and a specified hazard type are retrieved from the hazard database through SQL query;

[0111] Generate a historical hazard profile based on the retrieved hazard data and a stake coordinate system;

[0112] Reading the centerline data of the embankment project in the CAD design drawing of the embankment project planning, generating the centerline, and generating a historical hazard plan map based on the queried hazard data and the centerline coordinate system;

[0113] Converting the elements in the historical hazard cross-section and the historical hazard plan into a CAD importable format and importing the format into a CAD platform to generate a complete historical hazard cross-section and a complete historical hazard plan;

[0114] The hazard elements in the historical hazard plan are converted into a GIS importable format and imported into the GIS platform to generate complete historical hazard line data.

[0115] In this embodiment, in the dangerous situation neural network prediction model construction and training module 003, the dangerous situation neural network prediction model is a convolutional neural network + multi-layer perceptron dual-branch dangerous situation neural network model;

[0116] The input layer of the convolutional neural network branch is matrix data describing the borehole information; the row data of the matrix data corresponds to the stratum layering from the top to the bottom of the borehole; the column data corresponds to the stratum characteristics of the mapped stratum; the matrix data includes at least two columns, the first column is the stratum classification value, the second column is the stratum thickness, and the other columns are data reflecting the stratum characteristics, such as in-situ test data and sampling test data, which are arranged in a unified order; the minimum value of the stratum classification value is 0 and the maximum value is 1, which is used to reflect the permeability of the stratum. The greater the permeability, the greater the stratum classification value, and the smaller the permeability, the smaller the stratum classification value. The value is flexibly determined according to the type of embankment stratum and the permeability characteristics contained in the data set of model training, testing and prediction. For strata belonging to the same category in the engineering classification, differentiated values ​​are performed according to the different characteristics of the same stratum revealed by each borehole when taking the value;

[0117] The input layer of the multi-layer perceptron branch is scalar data describing the hydrological environment and structural characteristics of the levee within the drilling control range.

[0118] In this embodiment, in the hazard neural network prediction model construction and training module 003, in the process of training the hazard neural network prediction model through the set neural network data set, the set neural network data set includes: hazard data of each drilling control range segment extracted from the hazard database; drilling data of each drilling hole extracted from the drilling database; and embankment structure feature data extracted from the embankment engineering planning CAD design drawing.

[0119] It should be noted that the information interaction, execution process, etc. between the modules of the above-mentioned system are based on the same concept as the method embodiment in Example 1 of the present application, and the technical effects they bring are the same as those of the method embodiment of the present application. For specific contents, please refer to the description in the method embodiment shown above in the present application, and no further details will be given here.

[0120] Example 3

[0121] Example 3 of the present invention provides a non-transitory computer-readable storage medium, in which a program code for an application method for visual management of embankment engineering hazard data is stored. The program code includes instructions for executing an application method for visual management of embankment engineering hazard data of Example 1 or any possible implementation thereof.

[0122] Computer-readable storage media can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

[0123] Example 4

[0124] Embodiment 4 of the present invention provides an electronic device, including: a memory and a processor;

[0125] The processor and the memory communicate with each other through a bus; the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute an application method for visual management of embankment engineering hazard data in embodiment 1 or any possible implementation thereof.

[0126] Specifically, the processor can be implemented by hardware or by software. When implemented by hardware, the processor can be a logic circuit, an integrated circuit, etc.; when implemented by software, the processor can be a general-purpose processor, which is implemented by reading software code stored in a memory. The memory can be integrated into the processor or located outside the processor and exist independently.

[0127] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable systems. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode.

[0128] Obviously, those skilled in the art will appreciate that the various modules or steps of the present invention described above can be implemented using a general-purpose computing system. They can be centralized on a single computing system or distributed across a network of multiple computing systems. Alternatively, they can be implemented using program code executable by a computing system, and thus, they can be stored in a storage system and executed by the computing system. In some cases, the steps shown or described herein can be performed in a different order than that shown, or they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0129] Although the present invention has been described in detail above using general descriptions and specific embodiments, it will be apparent to those skilled in the art that modifications and improvements may be made thereto. Therefore, such modifications and improvements, without departing from the spirit of the present invention, are intended to be within the scope of protection claimed herein.

Claims

1. A method for visual management of embankment engineering hazard data, characterized in that: include: By standardizing the hazard data information table structure and the borehole data information table structure, the embankment hazard data and the embankment drilling data are standardized and sorted respectively to obtain the standardized embankment hazard data and the embankment drilling data; based on the standardized embankment hazard data and the standardized embankment drilling data, a hazard database and a drilling database are constructed based on the SQL database management system; Based on the hazard database and the borehole database, a levee engineering hazard data visualization management application system is constructed based on MATLAB; the levee engineering hazard data visualization management application system realizes hazard query, cross-section / plan view generation and CAD / GIS format export functions; Based on the hazard database and the borehole database, a hazard neural network prediction model is constructed; the hazard neural network prediction model is trained by setting a neural network data set to obtain the trained hazard neural network prediction model; The set input data is input into the trained dangerous situation neural network prediction model for prediction processing to obtain dangerous situation prediction results, and the dangerous situation prediction results are used to generate a predicted dangerous situation profile and a predicted dangerous situation plan.

2. A method for visual management of embankment engineering hazard data according to claim 1, characterized in that: The standardized hazard data information table structure includes: occurrence time, starting pile number, ending pile number, hazard type, description information, on-site photos and water level outside the dike when the hazard occurs; The on-site photo is the storage path of the photo or the binary data of the photo; The standardized drilling data information table structure includes: drilling name, pile number location, formation lithology, in-situ test data and sampling test data; The in-situ test data include: standard penetration test number, dynamic penetration test number and water injection test data; the sampling test data include: sample permeability coefficient, particle composition, porosity, compression coefficient and strength index.

3. A method for visual management and application of embankment engineering hazard data according to claim 2, characterized in that: During the application processing of the embankment engineering hazard data visualization management application system, the hazard data of the specified project, the specified occurrence time, the specified pile number range and the specified hazard type are retrieved from the hazard database through SQL query; Generate a historical hazard profile based on the retrieved hazard data and a stake coordinate system; Reading the centerline data of the embankment project in the CAD design drawing of the embankment project planning, generating the centerline, and generating a historical hazard plan map based on the queried hazard data and the centerline coordinate system; Converting the elements in the historical hazard cross-section and the historical hazard plan into a CAD importable format and importing the format into a CAD platform to generate a complete historical hazard cross-section and a complete historical hazard plan; The hazard elements in the historical hazard plan are converted into a GIS importable format and imported into the GIS platform to generate complete historical hazard line data.

4. The method for visual management of embankment engineering hazard data according to claim 3, characterized in that: The dangerous situation neural network prediction model is a convolutional neural network + multi-layer perceptron dual-branch dangerous situation neural network model; The input layer of the convolutional neural network branch is matrix data describing the borehole information; the row data of the matrix data corresponds to the stratum layer from the top to the bottom of the borehole; The column data corresponds to the formation characteristics of the mapped formation; the matrix data includes at least two columns, the first column is the formation classification value, the second column is the formation thickness, and the other columns are data reflecting the formation characteristics, including in-situ test data and sampling test data, which are arranged in a unified order; The minimum value of the stratum classification value is 0 and the maximum value is 1. It is used to reflect the permeability of the stratum. For strata belonging to the same category in the engineering classification, differentiated values ​​are determined based on the different characteristics of the same stratum revealed by each borehole. The input layer of the multi-layer perceptron branch is scalar data describing the hydrological environment and structural characteristics of the levee within the drilling control range.

5. A method for visual management and application of embankment engineering hazard data according to claim 4, characterized in that: In the process of training the hazard neural network prediction model through the set neural network data set, the set neural network data set includes: hazard data of each drilling control range segment extracted from the hazard database; drilling data of each drilling hole extracted from the drilling database; and embankment structure feature data extracted from the embankment engineering planning CAD design drawing.

6. A device for visual management of embankment engineering hazard data, using a method for visual management of embankment engineering hazard data according to any one of claims 1 to 5, characterized in that: include: A database construction module is used to standardize and organize the embankment engineering hazard data and the embankment engineering drilling data respectively through a standardized hazard data information table structure and a standardized drilling data information table structure, thereby obtaining standardized embankment engineering hazard data and standardized embankment engineering drilling data; and construct a hazard database and a drilling database based on the SQL database management system according to the standardized embankment engineering hazard data and the standardized embankment engineering drilling data; A module for constructing a levee engineering hazard data visualization management application system is used to construct a levee engineering hazard data visualization management application system based on MATLAB according to the hazard database and the borehole database; the levee engineering hazard data visualization management application system realizes hazard query, cross-section / plan view generation, and CAD / GIS format export functions; A dangerous situation neural network prediction model construction and training module is used to construct a dangerous situation neural network prediction model based on the dangerous situation database and the drilling database; train the dangerous situation neural network prediction model by setting a neural network data set to obtain the trained dangerous situation neural network prediction model; The hazard prediction result output module is used to input the set input data into the trained hazard neural network prediction model for prediction processing, obtain the hazard prediction result, and generate the predicted hazard profile and predicted hazard plan view from the hazard prediction result.

7. The device for visual management of embankment engineering hazard data according to claim 6, characterized in that: In the database construction module, the standardized hazard data information table structure includes: occurrence time, starting pile number, ending pile number, hazard type, description information, on-site photos and water level outside the dike when the hazard occurs; The on-site photo is the storage path of the photo or the binary data of the photo; The standardized drilling data information table structure includes: drilling name, pile number location, formation lithology, in-situ test data and sampling test data; The in-situ test data include: standard penetration test number, dynamic penetration test number and water injection test data; the sampling test data include: sample permeability coefficient, particle composition, porosity, compression coefficient and strength index.

8. The device for visual management of embankment engineering hazard data according to claim 7, characterized in that: In the construction module of the levee engineering hazard data visualization management application system, during the application processing of the levee engineering hazard data visualization management application system, hazard data of a specified project, a specified occurrence time, a specified pile number range, and a specified hazard type are retrieved from the hazard database through SQL query; Generate a historical hazard profile based on the retrieved hazard data and a stake coordinate system; Reading the centerline data of the embankment project in the CAD design drawing of the embankment project planning, generating the centerline, and generating a historical hazard plan map based on the queried hazard data and the centerline coordinate system; Converting the elements in the historical hazard cross-section and the historical hazard plan into a CAD importable format and importing the format into a CAD platform to generate a complete historical hazard cross-section and a complete historical hazard plan; The hazard elements in the historical hazard plan are converted into a GIS importable format and imported into the GIS platform to generate complete historical hazard line data.

9. The device for visual management of embankment engineering hazard data according to claim 8, characterized in that: In the dangerous situation neural network prediction model construction and training module, the dangerous situation neural network prediction model is a convolutional neural network + multi-layer perceptron dual-branch dangerous situation neural network model; The input layer of the convolutional neural network branch is matrix data describing the borehole information; the row data of the matrix data corresponds to the stratum layer from the top to the bottom of the borehole; The column data corresponds to the formation characteristics of the mapped formation; the matrix data includes at least two columns, the first column is the formation classification value, the second column is the formation thickness, and the other columns are data reflecting the formation characteristics, including in-situ test data and sampling test data, which are arranged in a unified order; The minimum value of the stratum classification value is 0 and the maximum value is 1. It is used to reflect the permeability of the stratum. For strata belonging to the same category in the engineering classification, differentiated values ​​are determined based on the different characteristics of the same stratum revealed by each borehole. The input layer of the multi-layer perceptron branch is scalar data describing the hydrological environment and structural characteristics of the levee within the drilling control range.

10. The device for visual management of embankment engineering hazard data according to claim 9, characterized in that: In the hazard neural network prediction model construction and training module, in the process of training the hazard neural network prediction model through the set neural network data set, the set neural network data set includes: hazard data of each drilling control range segment extracted from the hazard database; drilling data of each drilling hole extracted from the drilling database; and embankment structure feature data extracted from the embankment engineering planning CAD design drawing.

Images