BiLSTM-based intelligent simulation method and system for groundwater solute transport under strong heterogeneous condition
The BiLSTM algorithm is used to build a two-domain benchmark model, which solves the problem of computational complexity caused by the traditional simulation method without considering spatial heterogeneity when dealing with groundwater pollution, and realizes second-level computing and efficient pollutant migration simulation, which improves the source of groundwater pollution The accuracy of identification and parameter inversion.
Patent Information
- Application Number
- CN202411182197.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-12-03
AI Technical Summary
Traditional numerical simulation methods do not fully consider spatial heterogeneity when dealing with groundwater pollution, resulting in significant differences in simulation results from actual conditions, and the calculation complexity and calculation amount are large, resulting in high manpower and material cost and long-term calculation load.
Using the intelligent simulation method of groundwater solute migration under strong heterogeneous conditions based on BiLSTM, an alternative model is established by constructing a two-domain benchmark model and deep learning algorithm, and nonlinear dynamic modeling is performed using multi-layer neuronal structures and activation functions to adapt to Irregular data intervals to realize second-level operation.
It greatly reduces the calculation time and cost, improves the accuracy of groundwater pollution source identification and aquifer parameter inversion, improves the calculation efficiency of pollutant migration simulation prediction, and can handle highly nonlinearized groundwater parameter sequence sorting and numerical simulation.
Smart Images

Figure CN119066969A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of groundwater solute transport simulation, and in particular to a method and system for intelligent simulation of groundwater solute transport under strong heterogeneity conditions based on BiLSTM. Background Art
[0002] Groundwater pollution is difficult to observe directly due to its hidden nature. It can only be revealed through professional monitoring and investigation methods, but this is often accompanied by high costs in manpower and material resources. It is particularly important to accurately identify the source of groundwater pollution. It can not only effectively save resources, but also accurately track the location of the pollution source and its release history, laying a solid foundation for groundwater pollution risk assessment and responsibility determination.
[0003] On the other hand, the groundwater system is complex and the physical heterogeneity of the aquifer medium is significant. This characteristic directly or indirectly exacerbates the manifestation of chemical heterogeneity, which is specifically reflected in the dramatic changes in adsorption characteristics and reaction rate distribution over time and space during the migration of pollutants. However, traditional numerical simulation methods often fail to fully consider the impact of spatial heterogeneity when dealing with such problems, resulting in significant differences between simulation results and actual conditions.
[0004] In the process of identifying groundwater pollution sources and inverting aquifer parameters (permeability, porosity), especially when considering spatial heterogeneity, it is necessary to repeatedly call numerical simulation models for trial calculations. The complexity and amount of calculations in the models are significantly increased, which will generate a huge computing load and consume a lot of computing time. Summary of the invention
[0005] The purpose of the present invention is to propose a method and system for intelligent simulation of groundwater solute migration under strong heterogeneous conditions based on BiLSTM, to achieve second-level calculations, thereby greatly reducing calculation time and cost, and improving the accuracy of groundwater pollution source identification and aquifer parameter (permeability coefficient, porosity) inversion in actual production.
[0006] According to a first aspect of an embodiment of the present disclosure, a method for intelligent simulation of groundwater solute transport under strong heterogeneity conditions based on BiLSTM is provided, comprising the following steps:
[0007] The hydrogeological conditions were generalized and a preliminary hydrogeological conceptual model of groundwater solute transport was constructed to reflect the law of groundwater solute transport.
[0008] Determine the pollution source characteristics and aquifer parameters of pollutants in the hydrogeological conceptual model of groundwater solute transport; randomly sample several groups of working conditions according to the value range of pollution source characteristics;
[0009] Constructing a dual-domain benchmark model for groundwater solute transport under strong heterogeneity;
[0010] Substituting several groups of operating conditions into the groundwater solute transport-dual domain benchmark model one by one, obtaining the model response of each group of pollution source emission intensity samples, the model response corresponding to the pollutant concentration at the monitoring well location, and forming a training sample set and a test sample set consisting of "model input-model response" sample pairs;
[0011] According to the training sample set, the deep learning-BiLSTM algorithm was used to construct an alternative model of the groundwater solute transport-dual domain benchmark model, and the pollutant concentrations in the observation wells were predicted using the alternative model.
[0012] Furthermore, the Latin hypercube sampling algorithm SA-LHS is optimized using simulated annealing to randomly sample several groups of working conditions.
[0013] Furthermore, Modflow and MT3DMS programs were used to construct a dual-domain benchmark model for groundwater solute transport.
[0014] Furthermore, the alternative model of the groundwater solute transport-dual-domain benchmark model includes a forward LSTM unit and a backward LSTM unit; the model learns and retains long-term dependencies based on memory units and gating mechanisms; performs nonlinear dynamic modeling through a multi-level neuron structure and activation function; and uses a recursive time step to adapt to irregular data intervals.
[0015] Furthermore, for given pollution sources S1 and S2, the pollutant emission intensity during the four stress periods SP1, SP2, SP3, and SP4 is x t , t represents the time step, and the output of the forward LSTM unit is as follows:
[0016]
[0017] In the formula, Represents the output of the forward input gate at time t, which is used to control the state of the candidate unit at the current moment How much information is updated to the unit state Middle; W xi , W hi is the weight matrix of the forward input gate, is the forward input gate bias term, is the input at the current moment, is the groundwater pollutant concentration of observation wells Obs1, Obs2, Obs3, Obs4, Obs5, Obs6, and Obs7 at the previous moment; σ represents the sigmoid function, whose output value is between 0 and 1, and is used to determine the degree of information retention;
[0018]
[0019] In the formula, Represents the output of the forward forget gate at time t, which is used to control the unit state at the previous moment How much information has been forgotten? xf , W hf is the weight matrix of the forward forget gate, is the forward forget gate bias;
[0020]
[0021] In the formula, Represents the output of the forward output gate at time t, which is used to control the cell state How much information is output to the outside, W xo , W ho is the weight matrix of the forward output gate, is the forward output gate bias term;
[0022]
[0023] In the formula, Represents the forward candidate unit state at time t, used to update the forward unit state W xg , W hg is the weight matrix of the forward candidate unit state, It is the forward candidate unit state bias term, tanh hyperbolic tangent function, whose output value is between -1 and 1, and is used to generate the value of the forward candidate unit state;
[0024]
[0025] In the formula, Represents the output of the forward LSTM unit at time t, that is, the groundwater pollutant concentrations of observation wells Obs1, Obs2, Obs3, Obs4, Obs5, Obs6, and Obs7 at time t.
[0026] Furthermore, the output of the backward LSTM unit is as follows:
[0027]
[0028] In the formula, Represents the output of the backward input gate at time t, which is used to control the state of the candidate unit at the current time How much information is updated to the cell state Middle; W xi1 , W hi1 is the weight matrix of the backward input gate, is the backward input gate bias term, is the input at the current moment, is the groundwater pollutant concentration of observation wells Obs1, Obs2, Obs3, Obs4, Obs5, Obs6, and Obs7 at the next moment; σ represents the sigmoid function, whose output value is between 0 and 1, and is used to determine the degree of information retention;
[0029]
[0030] In the formula, Represents the output of the backward forget gate at time t, which is used to control the unit state at the next moment How much information has been forgotten? xf1 , W hf1 is the weight matrix of the backward forget gate, is the backward forget gate bias;
[0031]
[0032] In the formula, Represents the output of the backward output gate at time t, which is used to control the cell state How much information is output to the outside, W xo1 , W ho1 is the weight matrix of the backward output gate, is the backward output gate bias term;
[0033]
[0034] In the formula, Represents the state of the backward candidate unit at time t, which is used to update the backward unit state W xg1 , W hg1 is the weight matrix of the backward candidate unit state, It is the backward candidate unit state bias term, tanh hyperbolic tangent function, whose output value is between -1 and 1, used to generate the value of the backward candidate unit state;
[0035]
[0036] In the formula, Represents the output of the backward LSTM unit at time t, that is, the groundwater pollutant concentrations of observation wells Obs1, Obs2, Obs3, Obs4, Obs5, Obs6, and Obs7 at time t.
[0037] Furthermore, the hidden states of the forward LSTM unit and the backward LSTM unit at each time step t are Combine them to get the groundwater pollutant concentration h corresponding to the observation wells Obs1, Obs2, Obs3, Obs4, Obs5, Obs6, and Obs7 t :
[0038]
[0039] According to a second aspect of an embodiment of the present disclosure, a BiLSTM-based intelligent simulation system for groundwater solute transport under strong heterogeneity conditions is provided, comprising:
[0040] The conceptual model building module generalizes the hydrogeological conditions and preliminarily builds a hydrogeological conceptual model of groundwater solute transport to reflect the law of groundwater solute transport;
[0041] The working condition acquisition module determines the pollution source characteristics and aquifer parameters of pollutants in the hydrogeological conceptual model of groundwater solute transport; according to the value range of the pollution source characteristics, several groups of working conditions are randomly sampled;
[0042] Benchmark model building module, building a dual-domain benchmark model of groundwater solute transport under strong heterogeneity conditions;
[0043] The sample set acquisition module is used to input several groups of working conditions into the groundwater solute transport-dual domain benchmark model one by one, and obtain the model response of each group of pollution source emission intensity samples, which corresponds to the pollutant concentration at the monitoring well location, to form a training sample set and a test sample set consisting of "model input-model response" sample pairs;
[0044] The alternative model building module uses the deep learning-BiLSTM algorithm to build an alternative model of the groundwater solute transport-dual domain benchmark model based on the training sample set, and uses the alternative model to predict the pollutant concentration in the observation well.
[0045] According to the third aspect of an embodiment of the present disclosure, an electronic device is provided, including a memory, a processor, and a computer program stored and running on the memory, wherein when the processor executes the program, the intelligent simulation method for groundwater solute migration under strong heterogeneous conditions based on BiLSTM is implemented.
[0046] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the BiLSTM-based intelligent simulation method for groundwater solute migration under strong heterogeneity conditions is implemented.
[0047] Compared with the prior art, the above technical solution adopted by the present invention has the following advantages: 1) The present invention establishes an alternative model of groundwater solute transport-dual-domain benchmark model. Compared with the traditional deep learning model, it has stronger learning and generalization capabilities for the input-output relationship of the model, and is more adept at handling highly nonlinear groundwater parameter sequence arrangement and numerical simulation.
[0048] 2) The present invention solves the problem of high-precision replacement of the hydrogeological conceptual model of groundwater solute transport, which can greatly improve the computational efficiency of pollutant transport simulation and prediction, and provide an efficient solution to the problem of tedious calculations that require thousands of calls to the conceptual model for inversion identification of groundwater pollution source characteristics or pollutant transport parameters. BRIEF DESCRIPTION OF THE DRAWINGS
[0049] The drawings in the specification, which constitute a part of the present application, are used to provide further understanding of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute improper limitations on the present application.
[0050] Figure 1 This is an overview map of the example area;
[0051] Figure 2 The schematic diagram of the dual-domain model applicable to strong heterogeneity conditions;
[0052] Figure 3 This is the simulation result diagram of the dual-domain model;
[0053] Figure 4 This is the schematic diagram of the BiLSTM algorithm;
[0054] Figure 5 The scatter point fitting diagram of the BiLSTM alternative model corresponding to 7 observation wells. Specific implementation methods
[0055] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.
[0056] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.
[0057] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.
[0058] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the methods and systems according to various embodiments of the present disclosure. It should be noted that each box in the flowchart or block diagram can represent a module, a program segment, or a part of a code, and the module, program segment, or a part of a code may include one or more executable instructions for implementing the logical functions specified in each embodiment. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, or they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the flowchart and / or block diagram, and the combination of boxes in the flowchart and / or block diagram can be implemented using a dedicated hardware-based system that performs a specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.
[0059] Embodiment 1:
[0060] This embodiment provides a method for intelligent simulation of groundwater solute transport under strong heterogeneity conditions based on BiLSTM, including the following steps:
[0061] S1. Generalize the hydrogeological conditions and preliminarily construct a hydrogeological conceptual model of groundwater solute transport to reflect the law of groundwater solute transport;
[0062] S2. Determine the pollution source characteristics and aquifer parameters of pollutants in the hydrogeological conceptual model of groundwater solute transport; according to the value range of pollution source characteristics, as shown in Table 1, randomly sample several groups of working conditions;
[0063] like Figure 1 As shown in the figure, in this model, the blue lines AB and CD of the aquifer are constant head boundary conditions, while the other two sides (green lines AC and BD) are impermeable boundaries. The aquifer is simulated by sequential Gaussian simulation to generate a random permeability field, assuming that the permeability value is constant within its corresponding grid.
[0064] Table 1 Prior intervals of pollution source parameters
[0065]
[0066] S3. Figure 2 As shown in Figure 2, a dual-domain benchmark model of groundwater solute transport under strong heterogeneity is constructed. The physicochemical parameters of the benchmark model are shown in Table 2:
[0067] Table 2 Physicochemical parameters
[0068]
[0069] Based on the above information, the model was constructed using Modflow and MT3DMS.
[0070] S4. Substitute several groups of working conditions into the groundwater solute transport-dual domain benchmark model one by one to obtain the model response of each group of pollution source emission intensity samples, the model response corresponding to the pollutant concentration at the monitoring well location, and form a training sample set and a test sample set consisting of "model input-model response" sample pairs;
[0071] Specifically, the training sample set may be 350 groups, and the test sample set may be 150 groups.
[0072] S5. Based on the training sample set, the deep learning-BiLSTM algorithm is used to construct an alternative model of the groundwater solute transport-dual domain benchmark model, and the alternative model is used to predict the pollutant concentrations in the observation wells.
[0073] Specifically, forward modeling and prediction calculations were performed based on the training samples and test samples to obtain the corresponding spatial and temporal distribution of pollutant concentrations, that is, the pollutant concentrations at the seven monitoring wells. Figure 3 shown.
[0074] The alternative models constructed by BiLSTM have shown high accuracy in predicting the concentration of pollutants at specific monitoring points. For the test results of the inspection samples, the outputs of these alternative models are almost perfectly distributed along a 1:1 straight line with the outputs of the original simulation model. Figure 5 The scatter fitting plot proves that the BiLSTM alternative model can effectively learn and accurately reconstruct the complex relationship between input and output in the simulation model under strong heterogeneous conditions.
[0075] In summary, this invention effectively solves the key problem of intelligent substitution in the numerical model of groundwater organic pollutant migration. The present invention not only greatly accelerates the originally time-consuming and lengthy calculation process, but also achieves accurate characterization of the temporal and spatial variation of groundwater solutes under strong heterogeneity conditions while ensuring high calculation efficiency. This innovative achievement not only provides strong technical support for research and practice in the field of groundwater pollution prevention and control, but also opens up a new path for promoting the deep integration of environmental science and information technology.
[0076] Embodiment 2:
[0077] This embodiment provides an intelligent simulation system for groundwater solute transport under strong heterogeneity conditions based on BiLSTM, including:
[0078] The conceptual model building module generalizes the hydrogeological conditions and preliminarily builds a hydrogeological conceptual model of groundwater solute transport to reflect the law of groundwater solute transport;
[0079] The working condition acquisition module determines the pollution source characteristics and aquifer parameters of pollutants in the hydrogeological conceptual model of groundwater solute transport; according to the value range of the pollution source characteristics, several groups of working conditions are randomly sampled;
[0080] Benchmark model building module, building a dual-domain benchmark model of groundwater solute transport under strong heterogeneity conditions;
[0081] The sample set acquisition module is used to input several groups of working conditions into the groundwater solute transport-dual domain benchmark model one by one, and obtain the model response of each group of pollution source emission intensity samples, which corresponds to the pollutant concentration at the monitoring well location, to form a training sample set and a test sample set consisting of "model input-model response" sample pairs;
[0082] The alternative model building module uses the deep learning-BiLSTM algorithm to build an alternative model of the groundwater solute transport-dual domain benchmark model based on the training sample set, and uses the alternative model to predict the pollutant concentration in the observation well.
[0083] Embodiment three:
[0084] An electronic device, comprising a memory, a processor, and a computer program stored and running on the memory, wherein the processor implements the above-mentioned intelligent simulation method of groundwater solute migration under strong heterogeneity conditions based on BiLSTM when executing the program, including:
[0085] The hydrogeological conditions were generalized and a preliminary hydrogeological conceptual model of groundwater solute transport was constructed to reflect the law of groundwater solute transport.
[0086] Determine the pollution source characteristics and aquifer parameters of pollutants in the hydrogeological conceptual model of groundwater solute transport; randomly sample several groups of working conditions according to the value range of pollution source characteristics;
[0087] Constructing a dual-domain benchmark model for groundwater solute transport under strong heterogeneity;
[0088] Substituting several groups of operating conditions into the groundwater solute transport-dual domain benchmark model one by one, obtaining the model response of each group of pollution source emission intensity samples, the model response corresponding to the pollutant concentration at the monitoring well location, and forming a training sample set and a test sample set consisting of "model input-model response" sample pairs;
[0089] According to the training sample set, the deep learning-BiLSTM algorithm was used to construct an alternative model of the groundwater solute transport-dual domain benchmark model, and the pollutant concentrations in the observation wells were predicted using the alternative model.
[0090] Embodiment 4:
[0091] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned intelligent simulation method for groundwater solute transport under strong heterogeneity conditions based on BiLSTM, including:
[0092] The hydrogeological conditions were generalized and a preliminary hydrogeological conceptual model of groundwater solute transport was constructed to reflect the law of groundwater solute transport.
[0093] Determine the pollution source characteristics and aquifer parameters of pollutants in the hydrogeological conceptual model of groundwater solute transport; randomly sample several groups of working conditions according to the value range of pollution source characteristics;
[0094] Constructing a dual-domain benchmark model for groundwater solute transport under strong heterogeneity;
[0095] Substituting several groups of operating conditions into the groundwater solute transport-dual domain benchmark model one by one, obtaining the model response of each group of pollution source emission intensity samples, the model response corresponding to the pollutant concentration at the monitoring well location, and forming a training sample set and a test sample set consisting of "model input-model response" sample pairs;
[0096] According to the training sample set, the deep learning-BiLSTM algorithm was used to construct an alternative model of the groundwater solute transport-dual domain benchmark model, and the pollutant concentrations in the observation wells were predicted using the alternative model.
[0097] Those skilled in the art should understand that the modules or steps of the present disclosure can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present disclosure is not limited to any specific combination of hardware and software.
[0098] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
[0099] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.
Claims
1. Intelligent simulation method of groundwater solute transport under strong heterogeneity conditions based on BiLSTM, It is characterized in that The following steps are involved: The hydrogeological conditions were generalized and a preliminary hydrogeological conceptual model of groundwater solute transport was constructed to reflect the law of groundwater solute transport. Determine the pollution source characteristics and aquifer parameters of pollutants in the hydrogeological conceptual model of groundwater solute transport; randomly sample several groups of working conditions according to the value range of pollution source characteristics; Constructing a dual-domain benchmark model for groundwater solute transport under strong heterogeneity; Substituting several groups of operating conditions into the groundwater solute transport-dual domain benchmark model one by one, obtaining the model response of each group of pollution source emission intensity samples, the model response corresponding to the pollutant concentration at the monitoring well location, and forming a training sample set and a test sample set consisting of "model input-model response" sample pairs; According to the training sample set, the deep learning-BiLSTM algorithm was used to construct an alternative model of the groundwater solute transport-dual domain benchmark model, and the pollutant concentrations in the observation wells were predicted using the alternative model.
2. According to claim 1, the intelligent simulation method of groundwater solute transport under strong heterogeneity conditions based on BiLSTM, It is characterized in that The Latin hypercube sampling algorithm SA-LHS is optimized using simulated annealing method to randomly sample several groups of working conditions.
3. According to claim 1, the intelligent simulation method of groundwater solute transport under strong heterogeneity conditions based on BiLSTM, It is characterized in that A dual-domain benchmark model of groundwater solute transport was constructed using Modflow and MT3DMS programs.
4. According to claim 1, the intelligent simulation method of groundwater solute transport under strong heterogeneity conditions based on BiLSTM, It is characterized in that An alternative model of the groundwater solute transport-dual domain benchmark model includes a forward LSTM unit and a backward LSTM unit; the model learns and retains long-term dependencies based on memory units and gating mechanisms; and performs nonlinear dynamic modeling through multi-level neuron structures and activation functions; Use a recursive approach to time stepping to accommodate irregular data spacing.
5. According to claim 4, the intelligent simulation method of groundwater solute transport under strong heterogeneity conditions based on BiLSTM, It is characterized in that For given pollution sources S1 and S2, the pollutant emission intensity during the four stress periods SP1, SP2, SP3, and SP4 is x t , t represents the time step, and the output of the forward LSTM unit is as follows: In the formula, Represents the output of the forward input gate at time t, which is used to control the state of the candidate unit at the current moment How much information is updated to the unit state Middle; W xi , W hi is the weight matrix of the forward input gate, is the forward input gate bias term, is the input at the current moment, is the groundwater pollutant concentration of observation wells Obs1, Obs2, Obs3, Obs4, Obs5, Obs6, and Obs7 at the previous moment; σ represents the sigmoid function, whose output value is between 0 and 1, and is used to determine the degree of information retention; In the formula, Represents the output of the forward forget gate at time t, which is used to control the unit state at the previous moment How much information has been forgotten? xf , W hf is the weight matrix of the forward forget gate, is the forward forget gate bias; In the formula, Represents the output of the forward output gate at time t, which is used to control the cell state How much information is output to the outside, W xo , W ho is the weight matrix of the forward output gate, is the forward output gate bias term; In the formula, Represents the forward candidate unit state at time t, used to update the forward unit state W xg , W hg is the weight matrix of the forward candidate unit state, It is the forward candidate unit state bias term, tanh hyperbolic tangent function, whose output value is between -1 and 1, and is used to generate the value of the forward candidate unit state; In the formula, Represents the output of the forward LSTM unit at time t, that is, the groundwater pollutant concentrations of observation wells Obs1, Obs2, Obs3, Obs4, Obs5, Obs6, and Obs7 at time t.
6. According to claim 5, the intelligent simulation method of groundwater solute transport under strong heterogeneity conditions based on BiLSTM, It is characterized in that The output of the backward LSTM unit is as follows: In the formula, Represents the output of the backward input gate at time t, which is used to control the state of the candidate unit at the current time How much information is updated to the cell state Middle; W xi1 , W hi1 is the weight matrix of the backward input gate, is the backward input gate bias term, is the input at the current moment, is the groundwater pollutant concentration of observation wells Obs1, Obs2, Obs3, Obs4, Obs5, Obs6, and Obs7 at the next moment; σ represents the sigmoid function, whose output value is between 0 and 1, and is used to determine the degree of information retention; In the formula, Represents the output of the backward forget gate at time t, which is used to control the unit state at the next moment How much information has been forgotten? xf1 , W hf1 is the weight matrix of the backward forget gate, is the backward forget gate bias; In the formula, Represents the output of the backward output gate at time t, which is used to control the cell state How much information is output to the outside, W xo1 , W ho1 is the weight matrix of the backward output gate, is the backward output gate bias term; In the formula, Represents the state of the backward candidate unit at time t, which is used to update the backward unit state W xg1 , W hg1 is the weight matrix of the backward candidate unit state, It is the backward candidate unit state bias term, tanh hyperbolic tangent function, whose output value is between -1 and 1, used to generate the value of the backward candidate unit state; In the formula, Represents the output of the backward LSTM unit at time t, that is, the groundwater pollutant concentrations of observation wells Obs1, Obs2, Obs3, Obs4, Obs5, Obs6, and Obs7 at time t.
7. According to claim 6, the intelligent simulation method of groundwater solute transport under strong heterogeneity conditions based on BiLSTM, It is characterized in that Hidden states of the forward LSTM unit and the backward LSTM unit at each time step t and Combine them to get the groundwater pollutant concentration h corresponding to the observation wells Obs1, Obs2, Obs3, Obs4, Obs5, Obs6, and Obs7 t :
8. Intelligent simulation system of groundwater solute transport under strong heterogeneity conditions based on BiLSTM, It is characterized in that include: The conceptual model building module generalizes the hydrogeological conditions and preliminarily builds a hydrogeological conceptual model of groundwater solute transport to reflect the law of groundwater solute transport; The working condition acquisition module determines the pollution source characteristics and aquifer parameters of pollutants in the hydrogeological conceptual model of groundwater solute transport; according to the value range of the pollution source characteristics, several groups of working conditions are randomly sampled; Benchmark model building module, building a dual-domain benchmark model of groundwater solute transport under strong heterogeneity conditions; The sample set acquisition module is used to input several groups of working conditions into the groundwater solute transport-dual domain benchmark model one by one, and obtain the model response of each group of pollution source emission intensity samples, which corresponds to the pollutant concentration at the monitoring well location, to form a training sample set and a test sample set consisting of "model input-model response" sample pairs; The alternative model building module uses the deep learning-BiLSTM algorithm to build an alternative model of the groundwater solute transport-dual domain benchmark model based on the training sample set, and uses the alternative model to predict the pollutant concentration in the observation well.
9. An electronic device comprising a memory, a processor and a computer program stored and executed on the memory, It is characterized in that When the processor executes the program, the BiLSTM-based intelligent simulation method for groundwater solute migration under strong heterogeneity conditions is implemented.
10. A computer-readable storage medium, It is characterized in that A computer program is stored thereon, and when the program is executed by a processor, the intelligent simulation method for groundwater solute migration under strong heterogeneity conditions based on BiLSTM is implemented.
