Underground water distribution characterization method, system and equipment based on multi-source data fusion

Abstract

The invention provides an underground water distribution characterization method, system and equipment based on multi-source data fusion, and belongs to the technical field of underground engineering monitoring. The method comprises the following steps: acquiring multi-source acquisition data corresponding to a tunnel construction site; determining an underground water distribution prediction parameter matrix according to the multi-source collection data and a pre-built underground water distribution prediction model, and performing inversion on the underground water distribution prediction parameter matrix based on a pre-built physical constraint inversion model to obtain dynamic field parameters. According to the dynamic field parameters, rock mass stress and construction risk indexes corresponding to the tunnel construction site are calculated to construct a rock mass stress distribution diagram and a risk thermodynamic diagram, and the rock mass stress distribution diagram and the risk thermodynamic diagram are sent to a user terminal for visual display; and based on the construction risk index in the risk thermodynamic diagram, determining whether the tunnel construction site meets a preset construction risk condition, so as to generate early warning processing prompt information under the condition that the preset construction risk condition is met, and sending the early warning processing prompt information to a user terminal.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of underground engineering monitoring, and in particular to a groundwater distribution characterization method, system and equipment based on multi-source data fusion. BACKGROUND

[0002] At present, groundwater monitoring in tunnel engineering mainly relies on drilling pumping test, hydrogeological profile and empirical formula estimation, which has problems such as single data dimension and insufficient model accuracy. The traditional finite element method (FEM) has large simulation error for heterogeneous medium, and mainly depends on static geological data, which is difficult to reflect the dynamic change of groundwater; although the existing sensor network can realize data collection, it lacks the ability of spatio-temporal alignment and nonlinear relationship modeling of multi-source heterogeneous data, which limits the accuracy of groundwater distribution characterization.

[0003] Moreover, the prior art is also insufficient in real-time and dynamic response, and most systems rely on manual inspection and offline analysis, which cannot meet the rapid decision-making needs in the construction process. At the same time, the water inrush risk warning is mainly based on static threshold, and lacks dynamic correlation analysis of construction progress and groundwater pressure-displacement coupling relationship, resulting in late warning or high false alarm rate. The above defects limit the intelligent level of groundwater risk prevention and control in tunnel construction. SUMMARY

[0004] The embodiments of the present application provide a groundwater distribution characterization method, system and equipment based on multi-source data fusion, which is used to solve the technical problem of how to fuse multi-source data and physical modeling to achieve accurate and efficient groundwater risk prevention and control, and improve the intelligent level of groundwater risk prevention and control in tunnel construction.

[0005] In one aspect, the embodiments of the present application provide a groundwater distribution characterization method based on multi-source data fusion, which comprises: acquiring multi-source collected data corresponding to a tunnel construction site; determining a groundwater distribution prediction parameter matrix according to the multi-source collected data and a pre-constructed groundwater distribution prediction model, and inverting the groundwater distribution prediction parameter matrix based on a pre-established physical constraint inversion model to obtain dynamic field parameters; calculating rock mass stress and construction risk index corresponding to the tunnel construction site according to the dynamic field parameters, to construct a rock mass stress distribution map and a risk heat map, and sending the rock mass stress distribution map and the risk heat map to a user terminal for visual display; determining whether the tunnel construction site meets a preset construction risk condition based on the construction risk index in the risk heat map, to generate an early warning disposal prompt information and send it to the user terminal in the case of meeting the preset construction risk condition.

[0006] In an implementation form of the application, the multi-source collected data at least includes: a shallow aquifer spatial distribution matrix output by a preset ground penetrating radar, time-series monitoring data output by a preset fiber bragg grating sensor, a parameter vector constructed based on hydrological borehole data, and construction data output by a construction system; wherein the time-series monitoring data at least includes the following parameters: groundwater pressure, rock mass strain, and temperature; the parameter vector at least includes a permeability coefficient, an aquifer thickness, and a water level dynamic change value; and the construction data at least includes a tunneling speed and a support parameter.

[0007] In an implementation form of the application, before determining the groundwater distribution prediction parameter matrix according to the multi-source collected data and the pre-constructed groundwater distribution prediction model, the method further includes: timestamp alignment processing of the shallow aquifer spatial distribution matrix, the time-series monitoring data, and the parameter vector; spatial interpolation processing of the shallow aquifer spatial distribution matrix and the time-series monitoring data; normalization processing of each parameter in the parameter vector; determining multi-source collected data after data preprocessing according to the shallow aquifer spatial distribution matrix, the time-series monitoring data, the parameter vector, and the construction data after processing.

[0008] In an implementation form of the application, determining the groundwater distribution prediction parameter matrix according to the multi-source collected data and the pre-constructed groundwater distribution prediction model specifically includes: performing feature extraction on the multi-source collected data by a preset feature extraction hybrid network to obtain a spatio-temporal feature tensor after fusion of the multi-source collected data; wherein the preset feature extraction hybrid network includes a preset Transformer encoder, a preset GRU network, and a multi-head attention mechanism; inputting the spatio-temporal feature tensor into the groundwater distribution prediction model to adjust each parameter in the spatio-temporal feature tensor and output the groundwater distribution prediction parameter matrix.

[0009] In an implementation form of the application, performing inversion on the groundwater distribution prediction parameter matrix based on a pre-established physical constraint inversion model to obtain a dynamic field parameter, specifically including: inputting the groundwater distribution prediction parameter matrix into the physical constraint inversion model, taking Darcy's law and mass conservation equation as physical constraint terms, and dynamically adjusting a physical constraint weight by a Lagrange multiplier method to output the dynamic field parameter.

[0010] In an implementation form of the present application, according to the dynamic field parameters, the rock mass stress and the construction risk index corresponding to the tunnel construction site are calculated, specifically including: According to the groundwater pressure, rock mass strain and preset material weight group in the dynamic field parameters, the rock mass stress is calculated; According to the groundwater pressure distribution, preset critical pressure threshold, current tunneling length and preset tunneling length threshold in the dynamic field parameters, the construction risk index is calculated.

[0011] In an implementation form of the present application, based on the construction risk index in the risk thermodynamic map, it is determined whether the tunnel construction site meets the preset construction risk condition, specifically including: The construction risk index is matched with a plurality of preset risk index classification intervals; In the case that the construction risk index is in any of the preset risk index classification intervals, it is determined that the tunnel construction site meets the preset construction risk condition.

[0012] In an implementation form of the present application, in the case that the preset construction risk condition is met, a warning disposal prompt information is generated, specifically including: According to the preset risk index classification interval in which the construction risk index is located, a construction risk level is determined; The dynamic field parameters and the construction risk level are input into a pre-trained warning disposal model to determine a warning disposal control instruction corresponding to the construction risk level; wherein the warning disposal control instruction is used to control the corresponding site construction device in the construction system; The construction risk level and the warning disposal control instruction are taken as the warning disposal prompt information to be sent to the user terminal so that the user can dispose the construction risk.

[0013] In a second aspect, the embodiments of the present application provide a groundwater distribution representation system based on multi-source data fusion, the system comprising: An acquisition module is configured to acquire multi-source acquisition data corresponding to a tunnel construction site; A first determination module is configured to determine a groundwater distribution prediction parameter matrix according to the multi-source acquisition data and a pre-constructed groundwater distribution prediction model, and to perform inversion on the groundwater distribution prediction parameter matrix based on a pre-established physical constraint inversion model to obtain dynamic field parameters; A calculation module is configured to calculate the rock mass stress and the construction risk index corresponding to the tunnel construction site according to the dynamic field parameters, to construct a rock mass stress distribution map and a risk thermodynamic map, and to send the rock mass stress distribution map and the risk thermodynamic map to a user terminal for visual display; The second determining module is configured to determine whether the tunnel construction site meets a preset construction risk condition based on the construction risk index in the risk thermal map, to generate a warning disposal prompt information and send the warning disposal prompt information to the user terminal when the preset construction risk condition is met.

[0014] In a third aspect, an underground water distribution characterization device based on multi-source data fusion is provided, and the device comprises: at least one processor; and a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the above-mentioned underground water distribution characterization method based on multi-source data fusion.

[0015] Compared with the prior art, the present application has the following significant effects: Through the above technical solution, the underground water distribution can be accurately and effectively predicted by using multi-source heterogeneous data and physical constraint inversion, and the prediction result can be ensured to comply with the physical law. Moreover, the underground water distribution is displayed by using the visualization technology, which reduces the manual inspection investment. In addition, the construction risk index calculated by the dynamic field parameter is used for risk early warning evaluation, which improves the early warning accuracy. The above scheme effectively improves the intelligent level of underground water risk prevention and control in tunnel construction, and realizes accurate and efficient underground water risk prevention and control by fusing multi-source data and physical modeling. BRIEF DESCRIPTION OF DRAWINGS

[0016] The accompanying drawings, which are included to provide a further understanding of the present application, constitute a part of the present application and illustrate embodiments of the present application and its description, which serve to explain the present application, and do not constitute improper limitations on the present application. In the drawings: Figure 1 A flowchart of the underground water distribution characterization method based on multi-source data fusion in the embodiments of the present application; Figure 2 A device structure diagram corresponding to the underground water distribution characterization method based on multi-source data fusion in the embodiments of the present application; Figure 3 A structure diagram of the underground water distribution characterization system based on multi-source data fusion in the embodiments of the present application; Figure 4 A structure diagram of the underground water distribution characterization device based on multi-source data fusion in the embodiments of the present application. DETAILED DESCRIPTION

[0017] In order to make the purposes, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described below in connection with specific embodiments of the present application and corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0018] The simulation error of the traditional finite element method (FEM) for heterogeneous media is large, and it depends on static geological data, and it is difficult to reflect the dynamic change of underground water; although the existing sensor network can realize data collection, it lacks the ability of spatio-temporal alignment and nonlinear relationship modeling of multi-source heterogeneous data, which limits the accuracy of underground water distribution representation. Moreover, the existing technology is also insufficient in real-time and dynamic response, and most systems rely on manual inspection and offline analysis, which cannot meet the rapid decision-making needs in the construction process. At the same time, the water inrush risk early warning is mostly based on static threshold, lacking dynamic correlation analysis with construction progress and groundwater pressure-displacement coupling relationship, resulting in late warning or high false alarm rate. The above defects limit the intelligent level of groundwater risk prevention and control in tunnel construction.

[0019] Based on this, the embodiments of the present application provide a groundwater distribution representation method, system and equipment based on multi-source data fusion, to solve the technical problem of how to fuse multi-source data and physical modeling to achieve accurate and efficient groundwater risk prevention and control, and improve the intelligent level of groundwater risk prevention and control in tunnel construction.

[0020] The various embodiments of the present application will be described in detail below in connection with the drawings.

[0021] The embodiments of the present application provide a groundwater distribution representation method based on multi-source data fusion, as shown in Figure 1 The method can include steps S101-S104: S101, acquiring multi-source collected data corresponding to the tunnel construction site.

[0022] The execution subject of the groundwater distribution representation method based on multi-source data fusion of the present application can be an edge computing device in tunnel engineering, or a server, a cloud server, etc., which is not specifically limited by the present application.

[0023] The edge computing device can be in communication connection with the multi-source heterogeneous data collection device, so as to collect multi-source collection data during tunnel construction. The hybrid model in the application can be compressed into a lightweight model using the knowledge distillation technology, and can be deployed on the edge computing device using the field programmable logic gate array (FPGA). The compressed model is deployed on the edge computing node (such as Xilinx Zynq UltraScale+ MPSoC), and the hardware acceleration is used to improve the computing efficiency, so that the model inference time is less than 1 second. In addition, the 5G-MEC (multi-access edge computing) platform is used to realize high-speed data transmission, reduce the delay to milliseconds, and meet the real-time monitoring requirements.

[0024] In the embodiment of the application, the multi-source collection data at least includes: a shallow aquifer spatial distribution matrix output by a preset ground penetrating radar, time series monitoring data output by a preset fiber Bragg grating sensor, a parameter vector constructed based on hydrological borehole data, and construction data output by a construction system. The time series monitoring data at least includes the following parameters: groundwater pressure, rock mass strain, and temperature. The parameter vector at least includes the permeability coefficient, the aquifer thickness, and the water level dynamic change value. The construction data at least includes the tunneling speed and the support parameter.

[0025] The preset ground penetrating radar (GPR) can be arranged at the tunnel working face, such as the GSSISIR-3000 system, using a 200MHz antenna frequency to obtain the distribution information of the shallow aquifer at the working face and generate a shallow aquifer spatial distribution matrix. The GPR data collection frequency can be once every 2 hours, which can be set by the user and is not limited here. The coverage range is the area 5 meters in front and 5 meters behind the tunneling direction, to ensure continuous monitoring of the groundwater permeation path. The fiber Bragg grating sensor (FBG) can be arranged on the surface of the tunnel surrounding rock, and the sensor spacing is set to 1 meter, covering the tunnel working face and the surrounding area. The fiber Bragg grating sensor can monitor the groundwater pressure (accuracy ±0.01MPa), temperature (±0.1℃) and rock mass strain (±1 ) and the like, the data acquisition frequency can be once per second, and the data is transmitted to the edge computing device through the optical fiber communication module. The data acquisition frequency can be set by the user, and is not specifically limited herein. The hydrological borehole data can be obtained by arranging hydrological boreholes around the tunnel to obtain historical permeability coefficients, aquifer thicknesses, and water level dynamic change values, and the like. The data acquisition period of the borehole data is once per week, and the data acquisition period can be set by the user, and is not specifically limited herein. The construction data can be collected by a construction management system, such as a pre-set building information modeling (BIM) platform, to collect construction data such as tunneling speed, support parameters, and geological structure. The construction data acquisition frequency can be once per hour, and the data acquisition frequency can be set by the user, and is not specifically limited herein.

[0026] In S102, a groundwater distribution prediction parameter matrix is determined according to the multi-source collected data and a pre-constructed groundwater distribution prediction model, and the groundwater distribution prediction parameter matrix is inverted based on a pre-established physical constraint inversion model to obtain dynamic field parameters.

[0027] In the embodiments of the present application, before the groundwater distribution prediction parameter matrix is determined according to the multi-source collected data and the pre-constructed groundwater distribution prediction model, the method further includes: The shallow aquifer spatial distribution matrix, the time series monitoring data, and the parameter vector are time stamped and aligned. The shallow aquifer spatial distribution matrix and the time series monitoring data are spatially interpolated. The parameters in the parameter vector are normalized. The multi-source collected data after data preprocessing is determined according to the shallow aquifer spatial distribution matrix, the time series monitoring data, the parameter vector, and the construction data after processing.

[0028] In other words, before the multi-source collected data is used for groundwater analysis, the multi-source collected data is preprocessed, including time and space alignment, feature standardization, and the like. In the present application, the timestamps of the GPR, FBG, and hydrological borehole data can be aligned to millisecond accuracy using a timestamp synchronization algorithm. For data with inconsistent spatial resolution (such as 0.1 m resolution of GPR and 1 m resolution of FBG), Kriging interpolation is used for spatial interpolation to generate unified spatial grid groundwater distribution data. The groundwater pressure (in the present application, the groundwater pressure refers to osmotic pressure), temperature, strain, and the like in the time series detection data are subjected to Z-score normalization processing, and the formula is: wherein, is the original data, is the mean value, The standardized data range is [-1, 1] for the standard deviation, facilitating subsequent model training. Subsequently, the application can also concatenate GPR, FBG, drilling and construction record data to realize multi-modal data fusion (Concatenate) and construct multi-source heterogeneous fusion data for model processing.

[0029] In the embodiment of the application, the groundwater distribution prediction parameter matrix is determined according to the multi-source collected data and the pre-constructed groundwater distribution prediction model, and specifically includes: The multi-source collected data is subjected to feature extraction by a preset feature extraction hybrid network to obtain a spatio-temporal feature tensor after fusion of the multi-source collected data. The preset feature extraction hybrid network includes a preset Transformer encoder, a preset Gated Recurrent Unit (GRU) network and a multi-head attention mechanism. The spatio-temporal feature tensor is input into the groundwater distribution prediction model to adjust each parameter in the spatio-temporal feature tensor and output the groundwater distribution prediction parameter matrix.

[0030] That is, the application constructs a Transformer-GRU hybrid network as a feature extraction layer, which can use the Transformer encoder to adopt a multi-head self-attention mechanism to capture long-range spatio-temporal dependencies in GPR data. For example, the correlation between different GPR profiles is calculated by the self-attention mechanism to extract the layered distribution characteristics of the aquifer; the GRU network is used to process the time series data (such as osmotic pressure changes) of the FBG sensor, and the hidden state is dynamically adjusted by the GRU to capture the short-term fluctuation characteristics of the groundwater pressure; the fusion weight of the fused GPR, FBG, drilling and construction record data is obtained by the multi-head attention mechanism, and the formula is: wherein is the query, key and value matrix, is the dimension. The spatio-temporal feature tensor is obtained by using the fusion weight and the multi-source collected data.

[0031] Subsequently, the spatio-temporal feature tensor is input into a groundwater distribution prediction model for processing, and finally a groundwater distribution prediction parameter matrix output by the model is obtained. The groundwater distribution prediction model can be a model based on a deep Q network (DQN), and the present application can be constructed by the following steps: (1) state space design: the state space is composed of feature vectors such as normalized osmotic pressure, temperature, strain, construction parameters, and the dimension is n. For example, if 10 sensor data (osmotic pressure, temperature, strain, driving speed, support parameters, etc.) are collected, the state space dimension is 10; (2) action space design: the action space includes adjusting the parameters of the groundwater distribution model (such as the permeability coefficient K and the water conductivity coefficient T) and the construction parameters (such as the driving speed V and the support parameter P); (3) reward function design: the reward function is defined as: , wherein are respectively a first weight coefficient and a second weight coefficient, which are set by a user or an expert, and are not specifically limited here, is a predicted water level, is a measured water level, is a model calculation time consumption; (4) target network update: store state-action-reward-next state (s, a, r, s') data through experience replay, and regularly synchronize the parameters of the main network and the target network to stabilize the training process.

[0032] Through the constructed groundwater distribution prediction model, the groundwater distribution can be predicted in the state space, and the parameters in the spatio-temporal feature tensor are optimized and adjusted through the action space to obtain the groundwater distribution prediction parameter matrix, so as to realize the DQN reinforcement learning execution of the decision layer.

[0033] Further, in an embodiment of the present application, the groundwater distribution prediction parameter matrix is inverted based on a pre-established physical constraint inversion model to obtain dynamic field parameters, specifically including: The groundwater distribution prediction parameter matrix is input into the physical constraint inversion model, Darcy's law and the mass conservation equation are taken as physical constraint terms, and the physical constraint weight is dynamically adjusted by the Lagrange multiplier method, and the dynamic field parameters are output.

[0034] The present application also introduces Darcy's law and the mass conservation equation as physical constraint terms, which are added to the loss function, and the formula is: , is a final loss function value, is a loss function value of the groundwater distribution prediction model, which fuses data fitting and physical constraints, wherein are respectively a third weight coefficient and a fourth weight coefficient, i.e., a physical constraint weight, and respectively, are residual terms of Darcy's law and mass conservation. The application dynamically adjusts the constraint weight by the Lagrange multiplier method to ensure that the model training meets the physical law. Finally, according to determines that the dynamic field parameter is output if it is less than a predetermined value, and if determines that it is not less than the predetermined value, it means that further model training is needed, and an alarm prompt information can be generated. The predetermined value can be set by the user according to the actual use scene, which is not specifically limited here.

[0035] Through the above scheme, the application fuses multi-source heterogeneous data of various sensors, and establishes a physical constraint model to invert the dynamic groundwater distribution parameter, so as to ensure that the result meets the physical law.

[0036] S103, according to the dynamic field parameter, the rock mass stress and construction risk index corresponding to the tunnel construction site are calculated to construct the rock mass stress distribution map and risk thermal map, and the rock mass stress distribution map and risk thermal map are sent to the user terminal for visual display.

[0037] The user terminal can be understood as a terminal device of the relevant personnel of the tunnel construction, such as a mobile phone, a computer, etc., which is not specifically limited by the application.

[0038] In the embodiment of the application, according to the dynamic field parameter, the rock mass stress and construction risk index corresponding to the tunnel construction site are calculated, which specifically includes: According to the groundwater pressure, rock mass strain and preset material weight group in the dynamic field parameter, the rock mass stress is calculated. According to the groundwater pressure distribution, the preset critical pressure threshold, the current tunneling length and the preset tunneling length threshold in the dynamic field parameter, the construction risk index is calculated.

[0039] Specifically, the application inputs the groundwater pressure, rock mass strain and preset material weight group in the dynamic field parameter into the pre-set groundwater pressure-displacement coupling equation wherein, is the rock mass stress, p is the groundwater pressure, is the strain, are the first material parameter and the second material parameter respectively, so as to calculate the rock mass stress. which can be obtained by expert experience, which is not specifically limited by the application. The application also inputs the dynamic field parameter groundwater pressure distribution, preset critical pressure threshold, current tunneling length and preset tunneling length threshold into the preset construction risk index calculation formula to calculate the construction risk index, and the preset construction risk index calculation formula is as follows:

[0040] wherein, indicates the construction risk index, indicates the groundwater pressure distribution, indicates a preset critical pressure threshold, indicates the current tunneling length, indicates a preset tunneling length threshold, which can be understood as a maximum tunneling length.

[0041] It should be noted that the above groundwater pressure-displacement coupling equation and the preset construction risk index calculation formula can be independently trained for different tunnel projects. A federal learning framework can be used, and each tunnel project only shares model parameter gradients. At the same time, differential privacy technology is used to add noise to the gradient data: , to protect the privacy and security of multi-project data sharing, wherein, is the noisy gradient, is the noise intensity.

[0042] After the rock mass stress and the construction risk index are calculated, the rock mass stress distribution map and the risk thermal map are constructed according to the geographic position coordinate system of the corresponding tunnel construction site. At the same time, the rock mass stress distribution map and the risk thermal map calculated by combining multi-source heterogeneous data and physical constraint conditions are visualized and displayed, so as to represent the groundwater distribution, and facilitate the user to timely understand the groundwater distribution in the construction process.

[0043] In S104, based on the construction risk index in the risk thermal map, it is determined whether the tunnel construction site meets the preset construction risk condition, so as to generate a pre-warning disposal prompt information and send it to the user terminal in the case of meeting the preset construction risk condition.

[0044] In the embodiments of the present application, based on the construction risk index in the risk thermal map, it is determined whether the tunnel construction site meets the preset construction risk condition, which specifically includes: The construction risk index is matched with a plurality of preset risk index classification intervals. In the case that the construction risk index is in any preset risk index classification interval, it is determined that the tunnel construction site meets the preset construction risk condition.

[0045] In other words, the present application pre-sets a preset risk index classification interval for different levels of risk, and the preset risk index classification interval is, for example, three (a, b], (b, c], (c, d], at which time there can be three levels of warning signals, and different intervals correspond to one level. The number of preset risk index classification intervals is not specifically limited in the present application. Once the construction risk index is in any preset risk index classification interval, it means that the preset construction risk condition is met at this time, and risk warning is needed.

[0046] Further, in an embodiment of the present application, the pre-warning disposal prompt information is generated in the case of meeting the preset construction risk condition, which specifically includes: According to the preset risk index classification interval of the construction risk index, the construction risk level is determined. The dynamic field parameters and the construction risk level are input into the pre-trained early warning and disposal model to determine the early warning and disposal control instruction corresponding to the construction risk level. The early warning and disposal control instruction is used to control the corresponding field construction device in the construction system. The construction risk level and the early warning and disposal control instruction are used as early warning and disposal prompt information to be sent to the user terminal, so that the user can dispose the construction risk.

[0047] That is, the application can obtain the construction risk level according to the preset risk index classification interval of the construction risk index. Subsequently, the dynamic field parameters and the construction risk level are processed by using the early warning and disposal model. The early warning and disposal model can be a neural network model trained by a plurality of sample data corresponding to the construction site parameters and the risk level marked with the early warning and disposal mode. The output of the early warning and disposal model can determine the early warning and disposal mode corresponding to the above-mentioned dynamic field parameters and the above-mentioned construction risk level. Subsequently, the early warning and disposal control instruction corresponding to the early warning and disposal mode can be obtained by using a preset instruction comparison table. Finally, the application also sends the construction risk level and the early warning and disposal control instruction to the user for confirmation, so that the user can know the construction risk and choose whether to adopt the early warning and disposal control instruction for construction risk disposal. The early warning and disposal control instruction is, for example, excavation speed control, support reinforcement control, etc., which can be set by the actual use scene and is not limited here.

[0048] Through the above technical solution, the underground water distribution can be accurately and effectively predicted by using multi-source heterogeneous data and physical constraint inversion, and the prediction result can be ensured to comply with the physical law. Moreover, the application uses the visual technology to display the underground water distribution, reduces the artificial inspection investment, and uses the construction risk index calculated by the dynamic field parameters to perform risk early warning and evaluation, thereby improving the early warning accuracy. The above scheme effectively improves the intelligent level of underground water risk prevention and control in tunnel construction, and realizes the precise and efficient underground water risk prevention and control by fusing multi-source data and physical modeling.

[0049] Figure 2 For the device structure diagram corresponding to the underground water distribution characterization method based on multi-source data fusion in the embodiment of the application, as shown in Figure 2 , it includes a data layer, a data preprocessing module, a feature extraction layer, a decision layer, and an output layer. The device can execute the above-mentioned underground water distribution characterization method based on multi-source data fusion.

[0050] Figure 3 For the structure diagram of the underground water distribution characterization system based on multi-source data fusion provided by the embodiment of the application, as shown in Figure 3 , the underground water distribution characterization system based on multi-source data fusion 300 includes: The acquisition module 301 is configured to acquire multi-source acquisition data corresponding to the tunnel construction site. The first determination module 302 is configured to determine a groundwater distribution prediction parameter matrix according to the multi-source acquisition data and a pre-constructed groundwater distribution prediction model, and perform inversion on the groundwater distribution prediction parameter matrix based on a pre-established physical constraint inversion model to obtain dynamic field parameters. The calculation module 303 is configured to calculate rock mass stress and construction risk indexes corresponding to the tunnel construction site according to the dynamic field parameters, to construct a rock mass stress distribution map and a risk heat map, and send the rock mass stress distribution map and the risk heat map to a user terminal for visual display. The second determination module 304 is configured to determine whether the tunnel construction site meets a preset construction risk condition based on the construction risk indexes in the risk heat map, to generate an early warning disposal prompt information in the case of meeting the preset construction risk condition, and send the early warning disposal prompt information to the user terminal.

[0051] Figure 4 A structural schematic diagram of an underground water distribution characterization device based on multi-source data fusion provided by an embodiment of the present application is shown in Figure 4 As shown in the figure, the device comprises: at least one processor, and a memory connected in communication with the at least one processor. The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: acquire multi-source acquisition data corresponding to the tunnel construction site, determine a groundwater distribution prediction parameter matrix according to the multi-source acquisition data and a pre-constructed groundwater distribution prediction model, and perform inversion on the groundwater distribution prediction parameter matrix based on a pre-established physical constraint inversion model to obtain dynamic field parameters, calculate rock mass stress and construction risk indexes corresponding to the tunnel construction site according to the dynamic field parameters, to construct a rock mass stress distribution map and a risk heat map, and send the rock mass stress distribution map and the risk heat map to a user terminal for visual display, and determine whether the tunnel construction site meets a preset construction risk condition based on the construction risk indexes in the risk heat map, to generate an early warning disposal prompt information in the case of meeting the preset construction risk condition, and send the early warning disposal prompt information to the user terminal.

[0052] Each of the embodiments in the present application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, for system and device embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the part of the method embodiment.

[0053] The system and device and the method provided by the embodiments of the present application are one-to-one correspondence, therefore, the system and device also have similar beneficial technical effects to the corresponding method. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the system and device will not be described here.

[0054] It also needs to be explained that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0055] The above only describes the embodiments of the present application and is not used to limit the present application. The present application can have various changes and variations for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the scope of claims of the present application.

Claims

1. A method for representing groundwater distribution based on multi-source data fusion, characterized in that, The method comprises: acquiring multi-source acquisition data corresponding to a tunnel construction site; determining a groundwater distribution prediction parameter matrix according to the multi-source acquisition data and a pre-constructed groundwater distribution prediction model, and performing inversion on the groundwater distribution prediction parameter matrix based on a pre-constructed physical constraint inversion model to obtain dynamic field parameters; calculating rock mass stress and construction risk indexes corresponding to the tunnel construction site according to the dynamic field parameters, to construct a rock mass stress distribution map and a risk thermal map, and send the rock mass stress distribution map and the risk thermal map to a user terminal for visual display; determining whether the tunnel construction site meets a preset construction risk condition based on the construction risk indexes in the risk thermal map, to generate an early warning disposal prompt information and send it to the user terminal in the case that the preset construction risk condition is met. 2.The groundwater distribution characterization method based on multi-source data fusion according to claim 1, characterized in that, The multi-source acquisition data at least includes: a shallow aquifer spatial distribution matrix output by a preset geological radar, time series monitoring data output by a preset fiber Bragg grating sensor, a parameter vector constructed based on hydrological drilling data, and construction data output by a construction system; wherein the time series monitoring data at least includes the following parameters: groundwater pressure, rock mass strain, and temperature; the parameter vector at least includes permeability coefficient, aquifer thickness, and water level dynamic change value; and the construction data at least includes driving speed and support parameters.

3. The method according to claim 2, characterized in that, Before determining the groundwater distribution prediction parameter matrix according to the multi-source acquisition data and the pre-constructed groundwater distribution prediction model, the method further comprises: performing timestamp alignment processing on the shallow aquifer spatial distribution matrix, the time series monitoring data, and the parameter vector; performing spatial interpolation processing on the shallow aquifer spatial distribution matrix and the time series monitoring data; performing normalization processing on each parameter in the parameter vector; determining multi-source acquisition data after data preprocessing according to the shallow aquifer spatial distribution matrix, the time series monitoring data, the parameter vector, and the construction data after processing.

4. The method according to claim 3, characterized in that, Determining a groundwater distribution prediction parameter matrix according to the multi-source acquisition data and a pre-constructed groundwater distribution prediction model specifically comprises: extracting features of the multi-source acquisition data through a pre-set feature extraction hybrid network to obtain a spatio-temporal feature tensor after fusion of the multi-source acquisition data; wherein the pre-set feature extraction hybrid network comprises a pre-set Transformer encoder, a pre-set GRU network, and a multi-head attention mechanism; inputting the spatio-temporal feature tensor into the groundwater distribution prediction model to adjust each parameter in the spatio-temporal feature tensor and output the groundwater distribution prediction parameter matrix.

5. The method of claim 1, wherein, Performing inversion on the groundwater distribution prediction parameter matrix based on a pre-constructed physical constraint inversion model to obtain dynamic field parameters specifically comprises: inputting the groundwater distribution prediction parameter matrix into the physical constraint inversion model, taking Darcy's law and mass conservation equation as physical constraint terms, and dynamically adjusting physical constraint weights through the Lagrange multiplier method to output the dynamic field parameters.

6. The method of claim 1, wherein, According to the dynamic field parameters, the rock mass stress and the construction risk index corresponding to the tunnel construction site are calculated, specifically including: According to the groundwater pressure, rock mass strain and preset material weight group in the dynamic field parameters, the rock mass stress is calculated; According to the groundwater pressure distribution, preset critical pressure threshold, current tunneling length and preset tunneling length threshold in the dynamic field parameters, the construction risk index is calculated.

7. The method of claim 1, wherein, Based on the construction risk index in the risk thermodynamic map, it is determined whether the tunnel construction site meets the preset construction risk condition, specifically including: Matching the construction risk index with a plurality of preset risk index classification intervals; In the case that the construction risk index is in any of the preset risk index classification intervals, it is determined that the tunnel construction site meets the preset construction risk condition.

8. The method according to claim 7, characterized in that, In the case that the preset construction risk condition is met, a warning disposal prompt information is generated, specifically including: According to the preset risk index classification interval where the construction risk index is located, a construction risk level is determined; The dynamic field parameters and the construction risk level are input into a pre-trained warning disposal model to determine a warning disposal control instruction corresponding to the construction risk level; wherein the warning disposal control instruction is used to control the corresponding site construction device in the construction system; The construction risk level and the warning disposal control instruction are taken as the warning disposal prompt information to be sent to the user terminal so that the user can dispose the construction risk. 9.A system for characterizing groundwater distribution based on multi-source data fusion, characterized in that, The system includes: An acquisition module is configured to acquire multi-source acquisition data corresponding to a tunnel construction site; A first determination module is configured to determine a groundwater distribution prediction parameter matrix according to the multi-source acquisition data and a pre-constructed groundwater distribution prediction model, and to perform inversion on the groundwater distribution prediction parameter matrix based on a pre-established physical constraint inversion model to obtain dynamic field parameters; A calculation module is configured to calculate the rock mass stress and the construction risk index corresponding to the tunnel construction site according to the dynamic field parameters, to construct a rock mass stress distribution map and a risk thermodynamic map, and to send the rock mass stress distribution map and the risk thermodynamic map to a user terminal for visual display; A second determination module is configured to determine whether the tunnel construction site meets a preset construction risk condition based on the construction risk index in the risk thermodynamic map, to generate a warning disposal prompt information in the case that the preset construction risk condition is met, and to send the warning disposal prompt information to the user terminal.

10. A groundwater distribution characterization device based on multi-source data fusion, characterized in that, The device includes: At least one processor; and A memory connected in communication with the at least one processor; wherein The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method of characterizing groundwater distribution based on multi-source data fusion according to any one of claims 1-8.

Images