A method for in-situ remediation of heavy metals in soil using environmental functional materials.

Abstract

This invention relates to the field of soil environmental remediation technology and discloses a method for in-situ remediation of heavy metals in soil using environmental functional materials. The method includes obtaining the dynamic injection impedance and impedance time-series change rate during the injection of the main remediation fluid; combining this with the variance of the three-dimensional charge rate spatial distribution of the soil layer detected by a combined electrode module; performing intersection determination to extract the three-dimensional spatial coordinates of the dominant flow channel; subsequently injecting an electrically responsive phase change fluid according to these coordinates; using DC voltage to trigger local electrolytic solidification to cut off the channel; switching back to the main remediation fluid based on impedance recovery feedback and driving high-frequency pulsed infiltration; and reversing the electrical polarity after reaching a preset remediation volume to trigger reverse depolymerization of the phase change material, restoring it to a sol state. This invention achieves precise targeted blocking of the dominant flow and uniform distribution of the agent, expanding the affected volume, and restoring the original hydrodynamic environment of the underground aquifer without damage after remediation.

Description

Technical Field

[0001] This invention relates to the field of soil environmental remediation technology, specifically to a method for in-situ application of environmental functional materials for the remediation of heavy metals in soil. Background Technology

[0002] In-situ chemical application is a common method for treating heavy metal pollution in heterogeneous underground soils. It mainly involves continuously pumping liquid remediation materials into the target soil layer through injection pipelines. However, the geological structure in nature usually has significant spatial heterogeneity, with widespread high-permeability areas formed by gravel pores or natural fissures. During pressurized injection, the remediation agent will migrate along these areas with low permeability resistance, thus forming dominant flow channels. The generation of dominant flow not only leads to a large amount of agent being lost to non-target areas, but also causes the surrounding low-permeability clay matrix to fail to receive effective agent coverage due to leakage of injection pressure, seriously reducing the overall remediation uniformity of the site.

[0003] To address the issue of reagent loss caused by dominant flows, traditional methods typically employ chemical grouting to block high-permeability channels. However, in practical engineering applications, conventional grouting materials undergo irreversible physical or chemical solidification and cross-linking after sealing. This permanent network structure alters the original pore connectivity of the aquifer, thereby disrupting the natural permeability coefficient and hydrodynamic environment of the formation. Furthermore, due to the concealed nature of underground spaces, existing methods for monitoring fluid pressure or flow rate exhibit significant lag, making it difficult to accurately invert and locate the spatial coordinates of dominant flows within the complex three-dimensional space. This often results in anomalies only being detected after large-scale ineffective reagent escape, hindering early prediction and precise targeted interception of localized dominant channels. Therefore, how to achieve precise spatial positioning and temporary sealing of dominant flow channels without damaging the original hydrogeological structure of the formation, thereby encouraging reagent permeation into the low-permeability matrix, is a practical problem that needs to be solved in current in-situ remediation projects. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for in-situ remediation of heavy metals in soil using environmental functional materials. This method solves the problems of existing in-situ remediation methods where the agents are lost along the dominant flow, and conventional sealing processes are difficult to accurately locate spatially and can permanently damage the original permeability structure of underground aquifers.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of this invention provides a method for in-situ remediation of heavy metals in soil using environmental functional materials, comprising: continuously injecting a main remediation fluid into a target underground soil layer and obtaining dynamic injection impedance and impedance time-series change rate; generating a three-dimensional charge rate matrix of the soil layer using a combined electrode module and calculating the spatial distribution variance of the charge rate; comparing the impedance time-series change rate with a first threshold and comparing the spatial distribution variance with a second threshold to determine the formation of a dominant flow channel in the underground soil layer and extracting the three-dimensional spatial coordinates corresponding to the dominant flow node; pushing an electroresponsive phase change fluid to the position corresponding to the three-dimensional spatial coordinates and triggering a local electrolysis reaction by outputting a DC voltage from the combined electrode module to cut off the channel; when the dynamic injection impedance rises back to a preset range, switching back to the main remediation fluid and adjusting to a high-frequency micro-pulse mode to promote radial extrusion of the main remediation fluid; and after extrusion is completed, reversing the electrical polarity of the combined electrode module to trigger reverse depolymerization of the material at the cut-off channel to restore it to a sol state.

[0006] The technical principle of this invention lies in combining dynamic injection impedance monitoring of hydrodynamic indicators with three-dimensional charge rate inversion of electromagnetic indicators. Based on the changes in fluid seepage pressure and flow characteristics, and in conjunction with the polarization distribution variance caused by the conductivity differences in porous media, the three-dimensional spatial coordinates of the dominant flow channel are obtained. After locating the channel, an environmentally responsive material is injected, and an in-situ electrolysis environment is constructed using combined electrodes. The strongly alkaline field generated directionally in the cathode region promotes the solidification of the phase change liquid, thereby cutting off the agent leakage channel. After the sealing is completed, the system superimposes high-frequency micropulses to induce the main agent to enter the surrounding low-permeability medium, thereby expanding the swept volume. After the injection operation is completed, by reversing the electrode polarity, oxygen evolution and acid production in the original cathode region are caused to neutralize the local alkaline field, and the solid network structure undergoes reverse depolymerization due to deprotonation. This method achieves temporary targeted sealing of the dominant flow channel while restoring the original hydrodynamic environment of the formation without damage.

[0007] Furthermore, the steps for obtaining the dynamic injection impedance and the impedance time-series change rate include: setting the variable frequency pump to start operation at a preset displacement based on the initial permeability parameters of the target soil layer and obtaining pressure and flow data located at the injection pipeline orifice; using a moving average filtering algorithm to smooth and reduce noise in the pressure and flow data over time; using fluid dynamics impedance conversion rules to calculate the dynamic injection impedance of the current underground porous medium; establishing a first-order backward difference model based on a preset time window and comparing the impedance values ​​at the current time with those at historical times to calculate the impedance time-series change rate.

[0008] Further steps for calculating the spatial distribution variance of the charge rate include: acquiring the surface secondary potential difference data sequence generated by the joint electrode module under alternating weak current detection mode; using a three-dimensional frequency domain inversion algorithm to perform forward simulation and inverse iterative solution to generate a three-dimensional charge rate matrix of the soil layer characterizing the degree of polarization distribution; extracting the charge rate feature values ​​of each discrete grid node, performing scalar dimensionality reduction processing, and accumulating and averaging to obtain the arithmetic mean of the charge rate; and constructing a statistical spatial variance model to calculate the spatial distribution variance of the soil layer charge rate.

[0009] Furthermore, the steps of determining the dominant flow channel generated in the underground soil layer and extracting the three-dimensional spatial coordinates corresponding to the dominant flow node include: generating a determination signal when the impedance time-series change rate is lower than a first threshold and the spatial distribution variance is higher than a second threshold; selecting grid nodes in the three-dimensional charge rate matrix of the soil layer whose charge rate feature value is greater than a preset polarization anomaly threshold as dominant flow nodes; and extracting the three-dimensional spatial coordinates corresponding to the dominant flow node as target points.

[0010] Furthermore, the step of pushing the electroresponsive phase change fluid to the position corresponding to the three-dimensional spatial coordinates includes: instructing the electromagnetic switching valve group to cut off the main repair fluid and select the electroresponsive phase change fluid supply circuit; calculating the arithmetic mean of the coordinate components of the three-dimensional spatial coordinates of all dominant flow nodes in the depth direction as the target depth; converting the physical length of the pipeline and combining the geometric parameters of the inner diameter of the injection pipeline with the instantaneous output displacement of the variable frequency pump to calculate the fluid injection time delay required for the electroresponsive phase change fluid front to reach the target coordinate node.

[0011] Furthermore, the steps of triggering a local electrolysis reaction to cut off the channel include: after the fluid injection time delay, instructing the programmable power supply to switch the working state to DC strong field excitation mode; calculating the projection azimuth angle of the geometric center point relative to the central deep hole electrode and selecting the array electrode of the corresponding azimuth as the anode; simultaneously setting the central deep hole electrode as the cathode and outputting high-voltage DC power between the cathode and the anode; and using the strongly alkaline environment generated on the cathode surface to trigger the electroresponsive phase change liquid to undergo an in-situ physical phase change and transform into a solid hydrogel to seal the pores.

[0012] Furthermore, the steps for monitoring the dynamic injection impedance to rise back to the preset range include: continuously monitoring the numerical evolution of the dynamic injection impedance on the time axis; extracting the background soil reference impedance when no dominant flow escape occurs and superimposing the system tolerance coefficient to construct a judgment range characterizing the completion of flow field resistance reconstruction; and outputting a local cutoff compliance signal after monitoring the dynamic injection impedance value to rise back and stably fall into the judgment range to trigger the liquid supply circuit switchback action.

[0013] Furthermore, the step of adjusting to a high-frequency micro-pulse mode to induce radial extrusion of the main repair fluid includes: generating micro-pulse control commands and sending them to the variable frequency pump; forcing the variable frequency pump to perform mechanical reciprocating work according to a physical waveform superimposed with a reference constant discharge rate and a sinusoidal high-frequency disturbance; and driving the agent to radially extrude into the surrounding low-permeability soil matrix.

[0014] Furthermore, the step of reversing the electrical polarity of the combined electrode module to trigger the material depolymerization at the cutoff channel includes: after the repair operation reaches the preset reagent injection volume, controlling the programmable power supply to switch from forward DC to reverse DC mode to forcibly reverse the electrical polarity of the central deep hole electrode to the anode; using the oxygen evolution and acid production reaction occurring on the anode surface to reduce the local pH to trigger the deprotonation and depolymerization of the solid hydrogel network and restore it to a flowing sol state.

[0015] In a preferred embodiment of the present invention, the drug application method further includes a step of multidimensional physical field time-series prediction and precursor identification through a neural network model, and triggering the process of cutting off the channel in advance before determining that irreversible dominant flow escape and destruction has occurred.

[0016] Furthermore, the steps for multidimensional physical field time series prediction using a neural network model include: collecting historical monitoring sequences in parallel for the hydrodynamic and electromagnetic dimensions; extracting the dynamic injection impedance, impedance time series change rate, and charge rate spatial distribution variance within the historical time window to form a three-dimensional input tensor; and performing maximum and minimum value normalization on the three-dimensional input tensor to eliminate dimensional differences.

[0017] Furthermore, the steps for multidimensional physical field time series prediction using neural network models also include: continuously feeding the normalized three-dimensional input tensor into the long short-term memory neural network model to perform spatiotemporal correlation calculation of deep multidimensional features; and outputting the predicted spatial distribution variance of the target land parcel within a set future time span.

[0018] Furthermore, the steps for triggering the cutoff channel in advance include: comparing the variance of the predicted charge rate spatial distribution with the system's preset tolerance warning threshold; when the predicted value exceeds the limit, calling historical data to perform node differential calculation to obtain the charge rate increment of each grid node; selecting grid node combinations with charge rate increments greater than the preset precursor increment threshold to form a potential dominant flow channel; extracting the three-dimensional spatial coordinates of each node in the potential dominant flow channel and triggering the electromagnetic switching valve group to perform the action of cutting off the main repair fluid.

[0019] A second aspect of the present invention provides a system for implementing the above-described drug delivery method, comprising a control unit and a fluid injection module and a combined electrode module connected to the control unit, wherein the control unit is equipped with an impedance analysis module, a spatial inversion module, a cross-conversion determination module, a phase transition excitation module, and a percolation desealing module.

[0020] Furthermore, the combined electrode module includes a central deep hole electrode located at the center of the target plot and an array of electrodes arranged in a ring array around the central deep hole electrode on the ground surface. Both the central deep hole electrode and the array electrodes are connected to a dual-mode programmable power supply controlled by a control unit.

[0021] Furthermore, the dual-mode programmable power supply has an alternating weak current detection mode and a DC strong field excitation mode, and performs automatic electrical commutation and polarity reversal functions when it receives instructions from the control unit.

[0022] This invention provides a method for in-situ application of environmental functional materials for the remediation of heavy metals in soil. It has the following beneficial effects: 1. This invention obtains the dynamic injection impedance and impedance time-series change rate during continuous injection, and combines the variance of the three-dimensional charge rate spatial distribution of the soil layer generated by the joint electrode module to perform a joint determination of fluid mechanics and electromagnetics; it can accurately identify and extract the three-dimensional spatial coordinates of the dominant flow channel in complex heterogeneous soil layers, solve the problem of lag in traditional one-sided monitoring methods, and avoid the blind injection and loss of remediation agents.

[0023] 2. This invention injects an electro-responsive phase change liquid according to the extracted spatial coordinates and uses a combined electrode to output a DC voltage. The alkaline environment generated in the cathode region triggers the in-situ solidification of the phase change material to cut off the high-permeability channels. After the cut-off, the system switches back to the main remediation liquid based on impedance feedback and superimposes a high-frequency micro-pulse to drive the fluid to radially squeeze into the low-permeability soil layer, overcoming the sweeping dead zone caused by the dominant flow and improving the sweeping volume and remediation uniformity of the main agent in the overall contaminated soil layer.

[0024] 3. After reaching the preset repair volume, this invention reverses the electrical polarity of the combined electrode module, turning the original cathode into the anode and causing an oxygen evolution and acid production reaction. The generated hydrogen ions trigger the reverse depolymerization of the solid hydrogel network and restore it to a liquid sol. This process avoids the irreversible solidification caused by conventional chemical grouting and sealing processes, and achieves the non-destructive restoration of the original hydrodynamic permeability characteristics of the underground aquifer after the completion of temporary targeted sealing. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the system architecture of the present invention; Figure 2 This is a flowchart of the application method for in-situ remediation of heavy metals in soil using the environmental functional materials of the present invention; Figure 3 This is a detailed diagram of the fluid dynamics injection and impedance feature extraction process of the present invention; Figure 4 This is a detailed diagram of the well-to-surface joint detection and spatial distribution heterogeneity quantification process of the present invention; Figure 5 This is a detailed flowchart of the adaptive switching and directional electrochemical cutoff process of the present invention; Figure 6 This is a detailed diagram of the resistance reconstruction feedback and environmental reverse desealing process of the present invention; Figure 7 This is a detailed flowchart of the multidimensional physical field time-series prediction and precursor identification process of the present invention; Figure 8 This is a comparison chart of the dynamic injection impedance timing sequence and three-dimensional charge rate scatter plot data of each group in this invention. Detailed Implementation

[0026] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Please see the appendix Figure 1 , Figure 1 This is a schematic diagram of a system architecture according to an embodiment of the present invention. To implement the in-situ application method for remediating heavy metals in soil using environmental functional materials provided by the present invention, the present invention provides a hardware environment for an in-situ application system supporting the operation of this method. This in-situ application system may include: a control unit, a fluid injection module, and a combined electrode module.

[0028] The control unit establishes a data transmission channel with the fluid injection module and the combined electrode module through the underlying communication bus, enabling the in-situ drug delivery system to operate in a multi-physics field collaborative control environment. The fluid injection module is used to store and deliver the main repair fluid and the electro-responsive phase change fluid. Its supply end is equipped with a multi-channel variable frequency plunger pump and an electromagnetic switching valve group. After reading the fluid dynamics data from the high-frequency pressure transmitter and electromagnetic flowmeter at the injection pipeline orifice, the control unit forms an impedance change sequence array according to the time-series differential rule for subsequent dynamic calculation and physical field intersection determination.

[0029] The combined electrode module is equipped with a central deep-hole electrode, an array electrode, and a dual-mode programmable power supply. The dual-mode programmable power supply switches between alternating weak current detection mode and DC strong field excitation mode according to the communication commands issued by the control unit. The control unit reads the surface secondary potential difference sequence and spatial iterative cache data collected by the combined electrode module according to the communication commands.

[0030] The control unit is internally configured with various logic modules for executing the drug application method. These logic modules may include: an impedance analysis module, a spatial inversion module, a convergence determination module, a phase change activation module, a squeezing and infiltration unsealing module, and a precursor prediction module.

[0031] The impedance analysis module is used to receive pressure and flow data from the external injection pipeline, calculate the dynamic injection impedance and impedance time-series change rate, and extract transient characteristic parameters of the fluid penetration resistance in the underground medium. The spatial inversion module is used to generate a three-dimensional charge rate matrix of the soil layer based on the feedback of alternating weak current detection between the central deep hole electrode and the array electrode, and to calculate the spatial distribution variance of the charge rate using a three-dimensional frequency domain inversion algorithm. The intersection determination module is used to compare the calculation results of impedance time-series change rate and spatial distribution variance. The first threshold is set to 80% to 90% of the lower limit of normal seepage impedance fluctuation, and the second threshold is set to 1.5 to 2.0 times the background distribution variance. When the impedance time-series change rate is lower than the first threshold and the spatial distribution variance is higher than the second threshold, it is determined that the underground soil layer has a dominant flow channel. The grid nodes in the three-dimensional charge rate matrix of the soil layer with charge rate characteristic values ​​greater than the preset polarization anomaly threshold are selected as dominant flow nodes. The value range of the preset polarization anomaly threshold is defined as 2.0 to 3.0 times the average charge rate characteristic value of the background soil layer, and the three-dimensional spatial coordinates corresponding to the dominant flow node are extracted accordingly. The phase change excitation module is used to drive the electromagnetic switching valve group to select the energized phase change liquid based on the three-dimensional spatial coordinates of the dominant or potential dominant flow channel, and to control the dual-mode programmable power supply to output DC voltage to drive the cathode to generate a strongly alkaline environment, triggering the energized phase change liquid to turn into a hydrogel to cut off pores or potential pores. The squeezing and desealing module is used to switch the multi-channel variable frequency plunger pump to high frequency micro-pulse mode to drive the squeezing and desaturation of the main repair fluid after the dynamic injection impedance rises, and to control the dual-mode programmable power supply to switch the electrode polarity and output reverse voltage to trigger hydrogel depolymerization after squeezing and desaturation is completed. The precursor prediction module is used to collect historical monitoring sequences in both fluid dynamics and electromagnetic dimensions in parallel and feed them into a long short-term memory neural network model to predict the spatial distribution variance of charge rate. When it is determined that a dominant flow escape failure is about to occur, the module calls the background soil three-dimensional charge rate matrix cached at the initial detection time and the soil three-dimensional charge rate matrix at the current time to perform node difference calculation to obtain the charge rate increment of each grid node. Grid nodes with charge rate increments greater than the preset precursor increment threshold are selected as potential dominant flow nodes to form potential dominant flow channels. Then, the three-dimensional spatial coordinates of each node in the potential dominant flow channel are extracted to trigger the replacement and cutoff process of the electro-response phase change fluid in advance.

[0032] See attached document Figure 2 , Figure 2 This is a flowchart of a method for in-situ remediation of heavy metals in soil using environmental functional materials according to an embodiment of the present invention. The present invention provides a method for in-situ remediation of heavy metals in soil using environmental functional materials, comprising the following steps: The impedance analysis module instructs the fluid injection module to select the main repair fluid supply circuit and continuously inject the main repair fluid into the underground target soil layer. The combined electrode module is controlled to be in alternating weak current detection mode. The pressure and flow data of the injection pipeline are received to calculate the dynamic injection impedance and impedance time-series change rate. The surface secondary potential difference data between the central deep hole electrode and the array electrode is obtained through the spatial inversion module. The three-dimensional frequency domain inversion algorithm is used to generate the three-dimensional charge rate matrix of the soil layer, and the nodes of the three-dimensional charge rate matrix of the soil layer are extracted to calculate the spatial distribution variance of the charge rate. The impedance time-series change rate and spatial distribution variance are received by the intersection determination module. The impedance time-series change rate is compared with a first threshold and the spatial distribution variance is compared with a second threshold. When the impedance time-series change rate is lower than the first threshold and the spatial distribution variance is higher than the second threshold, it is determined that a dominant flow channel is generated in the underground soil layer. The grid nodes with charge rate eigenvalues ​​greater than the preset polarization anomaly threshold in the three-dimensional charge rate matrix of the soil layer are selected as dominant flow nodes, and the three-dimensional spatial coordinates corresponding to the dominant flow node are extracted accordingly. The phase change excitation module instructs the fluid injection module to select the electro-responsive phase change liquid supply circuit and push the electro-responsive phase change liquid to the position corresponding to the three-dimensional spatial coordinates of the dominant flow channel or potential dominant flow channel. The combined electrode module is then instructed to switch to the DC strong field excitation mode, setting the array electrode corresponding to the dominant flow channel or potential dominant flow channel as the anode and the central deep hole electrode as the cathode. A DC voltage is output between the cathode and the anode to trigger a local electrolysis reaction. The strongly alkaline environment generated by the cathode triggers the electro-responsive phase change liquid filled in the dominant flow channel or potential dominant flow channel to undergo a physical phase change and transform into a hydrogel to cut off the dominant flow channel or potential dominant flow channel. When the dynamic injection impedance rises back to the preset range by monitoring the squeezing and unsealing module, the fluid injection module is instructed to switch back to the main repair fluid supply circuit and adjust to the high-frequency micro-pulse mode, so as to cause the main repair fluid to radially squeeze and seep into the surrounding low-permeability soil layer. After the squeezing and seepage repair operation is completed, the combined electrode module is controlled to switch the electrical polarity of the electrode and output a reverse voltage, so as to cause the central deep hole electrode to turn into the anode and produce oxygen evolution and acid production reaction, reduce the alkalinity of the local environment and trigger the reverse depolymerization of the hydrogel to restore it to the sol state. The precursor prediction module continuously collects historical monitoring sequences in both hydrodynamic and electromagnetic dimensions and performs normalization processing. The input tensor is fed into a long short-term memory neural network model to predict the spatial distribution variance of charge rate within a set time span. When the predicted variance exceeds the system's preset tolerance and warning threshold, the system calls the background soil layer three-dimensional charge rate matrix cached at the initial detection time and the soil layer three-dimensional charge rate matrix at the current time to perform node difference calculation to obtain the charge rate increment of each grid node. Grid nodes with charge rate increments greater than the preset precursor increment threshold are selected as potential dominant flow nodes to form potential dominant flow channels. The three-dimensional spatial coordinates of each node in the potential dominant flow channel are sent to the phase change excitation module to trigger the replacement and electrochemical cutoff process of the electro-response phase change fluid in advance.

[0033] The following will elaborate on the technical implementation details of the above steps of the present invention in conjunction with specific physical calculation rules and execution logic.

[0034] See attached document Figure 3 , Figure 3 This is a detailed flowchart of the fluid dynamic injection and impedance feature extraction process according to an embodiment of the present invention. In this embodiment, the impedance analysis module built into the control unit sends an initialization command to the fluid injection module to drive the valve group to select the supply circuit of the main repair fluid. The impedance analysis module sets the variable frequency pump to start operation at a preset constant discharge rate based on the initial permeability parameters of the target soil layer. The preset constant discharge rate is set to a range of 10 to 30 liters per minute to match the seepage acceptance capacity of conventional porous media. The above-mentioned discharge rate driving logic causes the main repair fluid to enter the inner cavity of the metal drill pipe through the injection pipeline and be continuously injected into the underground target soil layer from the bottom injection hole section. As a preferred method, the impedance analysis module synchronously generates an electromagnetic control signal and sends it to the programmable power supply of the combined electrode module, instructing the programmable power supply to switch the working state to the alternating weak current detection mode. In the alternating weak current detection mode, the combined electrode module emits low-frequency alternating current into the ground to construct a background electric field characterizing the initial electrical distribution characteristics of the underground space medium.

[0035] During the continuous injection of the main repair fluid into the target soil layer, pressure transmitters and flow meters located at the injection pipe orifice synchronously acquire the original pressure and flow data during the dynamic fluid transport process at a fixed sampling frequency. The variable frequency pump generates high-frequency fluid pulsation noise during the mechanical reciprocating work phase. After receiving the original data, the impedance analysis module uses a moving average filtering algorithm to smooth and reduce noise in the original signal of the time series. The processing logic outputs effective pressure and flow data after eliminating high-frequency mechanical interference. For the specific mathematical calculation method of using the moving average filtering algorithm to process one-dimensional time series signals, the technicians can refer to conventional filtering methods in the field of digital signal processing in this embodiment.

[0036] The impedance analysis module extracts smoothed effective pressure and flow data and uses fluid dynamics impedance conversion rules to calculate the dynamic injection impedance of the current underground porous media. Combined with the physical mechanism of seepage in porous media, it is known that when fluid advances radially in a dense matrix, it must overcome solid frictional damping on the surface of soil particles and capillary damping encountered when flowing through complex pore throats. The dynamic injection impedance, in a macroscopic physical characterization, rigorously quantifies the level of mechanical energy dissipation during this radial migration process of a unit flow rate of the bulk remediation fluid. The specific formula for calculating the dynamic injection impedance is given below: ; In the formula: The dynamic injection impedance calculated for the current time point; The current continuous-time variable used for data sampling by the system; This is the valid pressure data output for the current time point; This outputs the valid traffic data corresponding to the current time point.

[0037] The pore structure of underground soil layers exhibits heterogeneous distribution characteristics in three-dimensional space. When the leading edge of the main remediation fluid advances in the porous matrix and enters the highly permeable fractures, the fluid penetration resistance decreases. The impedance analysis module establishes a first-order backward difference model based on a preset sliding time window and calculates the impedance time-series change rate by comparing the impedance values ​​at the current moment with those at historical moments. The length of the sliding time window is preset based on the empirical value of the permeability coefficient of the target soil layer, and its calculation formula is given below: ; In the formula: The impedance time-series change rate calculated for the current time point; The impedance at a historical moment is defined by a time interval equal to the length of a sliding time window. The value of the sliding time window length is a preset constant, ranging from 10 to 30 system data sampling periods.

[0038] See attached document Figure 4 , Figure 4This is a detailed flowchart of a well-ground joint detection and spatial distribution heterogeneity quantification process according to an embodiment of the present invention. In this embodiment, the spatial inversion module built into the control unit continuously receives monitoring data generated by the joint electrode module in alternating weak current detection mode. Specifically, the programmable power supply continuously emits multi-band alternating weak current between the central deep hole electrode and the surface array electrode. Since inorganic conductive tracer salt is pre-mixed in the main remediation fluid, the main remediation fluid objectively changes the original complex resistivity and excitation polarization response characteristics of the local soil when it migrates and seeps in the porous media network. The spatial inversion module synchronously collects a series of surface secondary potential difference data sequences induced on the surface of the array electrode. The surface secondary potential difference data sequence maps the dynamic change process of electrical distribution under the multi-field coupling effect in underground space at the macroscopic level.

[0039] After extracting the surface secondary potential difference data sequence containing stratigraphic polarization characteristics, the spatial inversion module uses a built-in three-dimensional frequency domain inversion algorithm to perform forward simulation and inverse iterative solution of the underground physical field. As a preferred method, the algorithm processing logic divides the geological space of the target block into a certain number of discrete three-dimensional grid nodes according to the preset physical field detection resolution and establishes the mapping relationship between node coordinates and electrical parameters. The value range of the above-mentioned physical field detection resolution is preferably set to 0.5 meters to 2.0 meters to balance the inversion detection accuracy and system computing power. After multiple iterations and convergence, the spatial inversion module generates a three-dimensional charge rate matrix of the soil layer that characterizes the degree of three-dimensional polarization distribution of the soil layer. For the specific mathematical calculation logic of solving the partial differential equation system and performing matrix iterative optimization using the three-dimensional frequency domain inversion algorithm, those skilled in the art can refer to the conventional inversion modeling methods in the field of geophysical exploration. The electromagnetic field forward and inverse modeling mechanism and code implementation process are well-known technologies in this field and will not be elaborated here.

[0040] The three-dimensional charge rate matrix of the soil layer contains spatial polarization node information. The spatial inversion module extracts the charge rate feature values ​​of each discrete grid node at the current moment from the constructed three-dimensional charge rate matrix of the soil layer and performs scalar dimensionality reduction on them. The spatial inversion module performs arithmetic summation and averaging of the charge rate feature values ​​of all grid nodes within the target monitoring range to obtain the overall average reagent enrichment level of the target plot. The total number of discrete grid nodes is set based on the volume of the target plot and combined with the detection resolution as a constant input to avoid grid divergence that could lead to algorithm dead zones. The specific formula for calculating the arithmetic mean of node charge rate is given below: ; In the formula: This is the arithmetic mean of the charge rate within the target grid space obtained at the current time point; The total number of discrete grid nodes within the target monitoring area; The positive integer index number for traversing each discrete grid node; For the current time node The inversion charge rate eigenvalues ​​corresponding to each discrete grid node.

[0041] Based on the aforementioned arithmetic mean, the spatial inversion module further constructs a statistical spatial variance model to quantify the heterogeneous morphology of agent diffusion within the soil. Combined with the physical diffusion laws of multiphase flow, it can be seen that when the main remediation fluid containing conductive salts exhibits uniform radial permeation in a dense matrix, the increase in charge rate at each grid node remains essentially consistent. When fluid forms dominant channels locally, leading to concentrated loss, this corresponds to an increase in the charge rate at local channel nodes, while low-permeability areas maintain a low background value, thus exhibiting a polarization deviation. The spatial inversion module quantifies the degree of this polarization deviation by calculating the spatial distribution variance of the charge rate. The specific formula for calculating the spatial distribution variance is given below: ; In the formula: The spatial distribution variance of soil charge rate calculated for the current time point; The total number of discrete grid nodes within the target monitoring area; The positive integer index number for traversing each discrete grid node; For the current time node Next, the The inversion charge rate characteristic value corresponding to each discrete grid node; For the current time node The arithmetic mean of the charge rates of all discrete grid nodes in the target grid space is obtained below. This refers to the current continuous-time variable used for data sampling by the system. The variance of the spatial distribution of soil charge rate serves as a core electromagnetic spatial dimension scaling parameter in subsequent control logic for determining whether dominant escape failure has occurred in the physical framework of formation seepage.

[0042] See attached document Figure 5 , Figure 5This is a detailed flowchart of the adaptive switching and directional electrochemical cutoff process according to an embodiment of the present invention. In this embodiment, the phase change excitation module built into the control unit issues a fluid path reconstruction command to the fluid injection module based on the three-dimensional spatial coordinates of the dominant or potential dominant flow channels extracted by the intersection determination module or the precursor prediction module. The phase change excitation module instructs the electromagnetic switching valve group to operate to cut off the main repair fluid and select the phase change fluid supply circuit to respond to the energization. Considering the significant physical pipeline space distance between the ground valve group and the underground dominant flow pores, the phase change excitation module targets the dominant or potential flow channels containing multiple nodes. The dominant flow channel calculates the arithmetic mean of the Z-axis coordinate components of the three-dimensional spatial coordinates of all nodes as the target dominant flow depth or potential dominant flow depth. Based on this, the physical length of the pipe from the ground orifice to this dominant flow depth or potential dominant flow depth is calculated. Then, combining the inner diameter geometric parameters of the injection pipe with the instantaneous output displacement of the variable frequency pump, the time delay required for the electro-response phase change fluid front to reach the target coordinate node is calculated. This time delay calculation logic objectively avoids the spatiotemporal misalignment problem caused by premature triggering of the electric field, leading to ineffective electrolysis of the fluid. The specific formula for calculating the injection time delay is given below: ; In the formula: For the calculated fluid injection time delay; The physical length of the pipe from the orifice to the target dominant flow depth; Pi is a mathematical constant. The internal geometric radius of the injection pipe; This represents the current instantaneous output displacement of the variable frequency pump.

[0043] After the system time has passed and the injection time delay has been exceeded, the phase change excitation module sends a strong field excitation command to the combined electrode module. The phase change excitation module instructs the dual-mode programmable power supply to switch its working state from alternating weak current detection mode to DC strong field excitation mode. As a preferred method, the phase change excitation module extracts the geometric center point of the three-dimensional spatial coordinates of all nodes of the dominant or potential dominant flow channels projected onto the horizontal plane, and calculates the projection azimuth angle of the geometric center point relative to the central deep hole electrode. Based on the azimuth angle, the corresponding array electrode is selected as the anode in the ring-shaped surface array. The phase change excitation module simultaneously sets the central deep hole electrode, which is lowered into the hole, as the cathode and outputs high-voltage DC current between the cathode and the anode to construct a low-resistance DC return field that penetrates the strata. Under the strong drive of the DC electric field, the system electrons accumulate towards the central deep hole electrode, which serves as the cathode, and trigger a strong water electrolysis reduction reaction. This electrochemical mechanism continuously generates hydrogen and a large number of hydroxide ions on the cathode surface, thereby causing the local underground fluid environment in the deep hole to rapidly evolve into a strongly alkaline state.

[0044] The sharp increase in pH around the deep-hole electrode directly constitutes the core chemical boundary condition that triggers the physical transformation of polymer materials. The electro-response phase change liquid filling and flowing through the dominant or potential dominant flow channels is rich in specific polymer materials that are extremely sensitive to hydroxide ions. These polymer materials maintain a low-viscosity sol state under the conventional weakly acidic to neutral environment of the soil to maintain good pore permeability. Once the fluid enters the strongly alkaline region around the cathode generated by water electrolysis, its molecular chain segments will instantly undergo deprotonation and trigger three-dimensional cross-linking between molecules. The rapid evolution of this polymer network structure causes the liquid sol to undergo an in-situ physical phase transition in a very short time and transform into a solid hydrogel with high yield stress. The solidified hydrogel plug, with its high structural strength, directly blocks the high-conductivity dominant or potential dominant flow pore network, thereby cutting off the seepage escape path.

[0045] See attached document Figure 6 , Figure 6 This is a detailed diagram of the resistance reconstruction feedback and environmental reverse unsealing process according to an embodiment of the present invention. In this embodiment, the squeezing and unsealing module built into the control unit continuously monitors the numerical evolution of the dynamic injection impedance on the time axis and uses this to assess the physical cutoff state of the dominant flow channel or potential dominant flow channel. Combined with the multiphase flow seepage mechanics characteristics, it can be seen that when the hydrogel plug forms in the dominant flow pore or potential dominant flow pore, the fluid front encounters a solid barrier and is forced to find a new micropore seepage path. The microscopic forced change of the seepage path is directly manifested macroscopically as the dynamic injection impedance of the system. Based on the significant rebound, as a preferred method, the seepage release module pre-extracts the background soil reference impedance when no dominant flow escape occurs and superimposes it with the system tolerance coefficient to construct a judgment interval characterizing the completion of flow field resistance reconstruction. The system tolerance coefficient ranges from 0.85 to 1.15. The above judgment interval is specifically defined as 85% to 115% of the background soil reference impedance value. After monitoring the dynamic injection impedance value to rebound and stably fall into the above judgment interval, the seepage release module outputs a partial cutoff compliance signal and instructs the fluid injection module to switch the liquid supply circuit back to the main repair liquid.

[0046] After confirming the fluid loop switching is complete, the seepage release module generates and sends a high-frequency micro-pulse control command to the variable frequency pump. Conventional constant flow injection can cause cracking damage to the soil skeleton when facing dense pores, thus reducing the uniformity of repair. The seepage release module outputs a periodically oscillating drive signal based on the control logic of the superposition of the benchmark constant flow and sinusoidal high-frequency disturbance, and forces the variable frequency pump to perform mechanical reciprocating work according to this physical waveform. This specific mode excites continuous high-frequency water hammer waves inside the fluid medium and forces the main repair fluid to radially seep into the low-permeability soil layer by relying on pulsating kinetic energy. The specific high-frequency pulse instantaneous flow control formula is given below: ; In the formula: This represents the instantaneous pulse flow rate actually output by the variable frequency pump at the current time point. The baseline constant displacement set for the system ranges from 5 to 15 liters per minute; The disturbance amplitude of the high-frequency micro-pulse flow field is set to a value range of 10% to 20% of the baseline constant displacement. The mechanical disturbance frequency is set to match the acoustic characteristics of the formation, and the value of the mechanical disturbance frequency ranges from 10 Hz to 50 Hz.

[0047] After the seepage remediation operation covers the target formation and reaches the preset agent injection volume, the seepage unsealing module sends a polarity reversal command to the combined electrode module and triggers the environmental reverse recovery program. The preset agent injection volume is precisely calculated by multiplying the spatial geometric volume of the target plot with the effective porosity of the soil layer. The seepage unsealing module controls the dual-mode programmable power supply to switch from forward DC to reverse DC mode, thereby forcibly reversing the electrical polarity of the central deep hole electrode from the original cathode to the anode. Under the drive of electric field reversal, a strong oxygen evolution reaction occurs on the surface of the metal anode, accompanied by the generation of a large number of hydrogen ions. The hydrogen ions diffuse outward and neutralize with the residual hydroxide ions, causing the local high pH environment to drop back to the original environmental background state. The hydrogel network with cut-off pores loses its cross-linking stability under the protonation of hydrogen ions and undergoes polymer chain segment breakage and depolymerization. The above mechanism causes the gel plug to liquefy back into a sol and dissipate with the groundwater, thereby completely restoring the formation hydrological permeability skeleton at the end of the project.

[0048] See attached document Figure 7 , Figure 7 This is a detailed flowchart of a multidimensional physical field time series prediction and precursor identification process according to an embodiment of the present invention. In this embodiment, the precursor prediction module built into the control unit continuously and in parallel collects historical monitoring sequences in the fluid dynamics and electromagnetic dimensions to perform data preprocessing operations. Combining the nonlinear evolution characteristics of underground medium polarization and seepage resistance on the time axis under multi-field coupling, it can be seen that conventionally set threshold judgments often have a certain physical response lag. As a preferred method, the precursor prediction module extracts the dynamic injection impedance, impedance time series change rate, and charge rate spatial distribution variance within a fixed time window in the past to form a three-dimensional input tensor. The system performs maximum and minimum value normalization processing on the above three-dimensional time series feature data to eliminate the numerical span differences caused by different physical dimensions. This preprocessing logic strictly constrains all input features to the standard zero to one interval, thereby objectively ensuring the gradient propagation stability during subsequent matrix operations.

[0049] The precursor prediction module continuously feeds the normalized three-dimensional input tensor into the built-in long short-term memory neural network model to perform spatiotemporal correlation calculations of deep multidimensional features. The network model structure includes an input layer that receives three-dimensional time-series data and two cascaded hidden layers containing forget gates and input gates. The hidden layers continuously step through the internal cell state vectors to transmit historical time-series state information to extract the nonlinear precursor evolution law of the main repair fluid seepage path deviating from uniform radial diffusion. The high-dimensional feature vector output by the hidden layers is then passed to the fully connected output layer containing a linear activation function for dimensionality reduction mapping. The specific physical state meaning of the final output result of the network model is the predicted spatial distribution variance of the charge rate of the target plot within a set time span in the future.

[0050] The precursor prediction module determines in advance whether irreversible dominant flow escape failure is imminent in the underground soil layer by comparing the predicted variance value with the system's preset tolerance warning threshold. This tolerance warning threshold is dynamically assigned based on 1.2 to 1.5 times the variance of the initial background charge rate distribution. When it determines that irreversible dominant flow escape failure is imminent, it calls the cached background soil layer 3D charge rate matrix at the initial detection time and the soil layer 3D charge rate matrix at the current time to perform node difference calculation to obtain the charge rate increment of each grid node, and filters out precursors with charge rate increments greater than the preset threshold. The grid nodes with the incremental threshold are combined as potential dominant flow nodes to form a potential dominant flow channel. The value range of the preset precursor incremental threshold is set to 30% to 50% of the initial characteristic value of the charge rate of the background grid nodes. The system uses this to eliminate the small polarization fluctuations caused by conventional infiltration and sends the three-dimensional spatial coordinates of each node in the potential dominant flow channel to the phase change excitation module. This allows the system to send commands to the fluid injection module and the phase change excitation module in advance to trigger the replacement and electrochemical cutoff process of the electro-response phase change liquid, thereby blocking the dominant flow at the bud stage.

[0051] Before being formally deployed to the field system, the Long Short-Term Memory (LSTM) neural network model performs offline supervised learning based on its built-in historical engineering database to optimize the weights of internal network nodes. The precursor prediction module uses measured multi-physics time-series data segments from historical engineering projects as training samples and extracts the variance of the spatial distribution of the true charge rate of the corresponding time-series segments by extending a specific time window forward as a supervision label. During the training iteration phase, the system employs a time backpropagation algorithm and uses mean squared error (MSE) as the core loss function to quantify the numerical dispersion deviation between the network model's predicted values ​​and the true physical labels. The specific formula for calculating the MSE loss function is given below: ; In the formula: This is the mean squared error loss value calculated iteratively for the current training batch; This represents the total number of time-series feature samples input into the model within the current training batch. The positive integer index number for traversing each temporal feature sample in the current training batch; The current output of the network model is the first... Variance of the spatial distribution of predicted charge rate for each time series sample; For the first in the historical engineering database The network model minimizes the variance label of the spatial distribution of the real objective charge rate corresponding to each time series sample until the internal parameters converge, thereby ensuring that those skilled in the art can fully reproduce the construction and training process of the prediction model in accordance with this constraint specification.

[0052] Please see the appendix Figure 8 : To aid in understanding the technical solution of this invention, the following is an application example of in-situ chemical remediation verification work performed on a deep soil contaminated with lead and cadmium in a decommissioned electroplating plant area.

[0053] The target site has a silty clay layer with a high-permeability sand and gravel permeable layer distributed in the area with a depth of 5 to 10 meters. The control unit controls a multi-channel variable frequency plunger pump to continuously inject the main remediation solution mixed with conductive tracer salt into the ground at a preset constant discharge rate of 20 liters per minute. The combined electrode module synchronously establishes a detection electric field and extracts the characteristic value of the underground background average charge rate as 2%. The system monitors that when the application operation reaches the 40th minute, the dynamic injection impedance drops from the initial 0.8 per megapascals per second per liter to 0.3. The impedance analysis module calculates that the impedance time sequence change rate at the current time point is lower than the lower limit of normal seepage impedance fluctuation.

[0054] The spatial inversion module calculates the spatial distribution variance of the charge rate of the underground soil layer through three-dimensional frequency domain inversion, which is 2.1 times the background variance. The intersection judgment module compares the multi-dimensional physical field data of fluid mechanics and electromagnetics to determine that the main repair fluid forms a dominant flow channel in the sand and gravel permeable layer. The control unit commands the electromagnetic switching valve group to select the liquid supply circuit of the energized phase change fluid according to the extracted three-dimensional spatial coordinates of the dominant flow node and pushes it in. The phase change excitation module calculates the ten-minute injection time delay in combination with the inner diameter of the pipeline and commands the dual-mode programmable power supply to set the central deep hole electrode as the cathode and output a 300-volt high-voltage DC current. The electrode interface undergoes a water electrolysis reaction to generate a strongly alkaline environment, triggering the cross-linking and solidification of the energized phase change fluid that enters the pores of the sand and gravel layer into a hydrogel, which cuts off the dominant flow channel.

[0055] The infiltration and unsealing module monitors the pipeline pressure and obtains the dynamic injection impedance value after the channel is blocked. The value is 0.85 MPa / second / liter and falls within the judgment range. The system switches back to the main repair fluid and drives the variable frequency pump to drive the agent to radially infiltrate into the surrounding low-permeability clay matrix by superimposing a mechanical disturbance frequency of 30 Hz and a disturbance amplitude of 15% of the base displacement in a high-frequency micro-pulse mode. After the cumulative injection volume of the repair agent reaches the preset volume, the infiltration and unsealing module controls the programmable power supply to switch the electrical polarity. The central deep hole electrode turns into an anode, triggering an oxygen evolution and acid production reaction, which in turn triggers the deprotonation and depolymerization of the solid hydrogel network structure and restores it to a sol state.

[0056] In a field control experiment, three identical test grid areas with the same geological pore structure and hydrological environment were divided in the same decommissioned electroplating contaminated site, and different processes were used for drug application. Control group 1 used a fixed-discharge in-situ drug injection process to push the drug at a fixed pumping pressure. Control group 2 used an irreversible chemical gel process to seal the formation of leakage channels. The experimental group used the multi-dimensional physical field synergistic electro-response reversible hydrogel targeted cut-off and high-frequency micropulse squeezing infiltration combined process provided by this invention to perform the drug application operation.

[0057] The system control center generates a measured multidimensional comparison parameter matrix based on the sampling and testing data of different drug application process groups, and renders multidimensional comparison charts using the dynamic injection impedance time series and three-dimensional charge rate scatter data of each group. Figure 8 The experimental data were extracted from the soil heavy metal leaching toxicity analysis report and the in-situ pumping test permeability test results. The verification and evaluation indicators included the effective swept volume ratio of the agent, the average removal rate of heavy metals lead and cadmium, and the recovery rate of the formation hydrodynamic permeability coefficient after the completion of the remediation project.

[0058] Table 1: Comparison of Field Application Effects of Different Remediation Application Techniques Experimental group Application control process category Effective volume percentage of the drug Average removal rate of heavy metals lead and cadmium Recovery rate of formation permeability after repair Control Group 1 Fixed displacement in-situ drug injection process 35.6% 42.1% 98.5% Control Group 2 Irreversible chemical gel plugging process 82.4% 85.3% 21.3% experimental group Multi-field synergistic truncation and reversible depolymerization process 86.7% 89.2% 97.8% From Table 1, we can obtain: Based on the field experimental verification data in Table 1, the experimental group of this invention obtained the highest numerical index within the test range. The main repair fluid of control group 1 was lost along the channel, resulting in an effective volume fraction of 35.6% and an average removal rate of heavy metals lead and cadmium of 42.1%. The irreversible chemical gel was injected into control group 2, resulting in an effective volume fraction of 82.4% and a formation permeability coefficient recovery rate of 21.3%. The experimental group implemented multi-physics field monitoring and electrochemical reversible phase change depolymerization, resulting in an effective volume fraction of 86.7% and a formation permeability coefficient recovery rate of 97.8%.

[0059] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for in-situ application of environmental functional materials for the remediation of heavy metals in soil, wherein the method is configured and executed in an in-situ application system, characterized in that, The system includes a control unit, a fluid injection module, and a combined electrode module. The control unit establishes data transmission channels with each module through a low-level communication bus, using methods including: Acquire fluid data during the injection of the main repair fluid into the fluid injection module, and extract dynamic injection impedance and impedance time-series change rate. Based on the potential difference data sequence of the combined electrode module, a three-dimensional charge rate matrix of the soil layer is generated and the spatial distribution variance is calculated. The intersection determination is performed by combining the impedance time-series change rate and the spatial distribution variance. When a dominant flow channel is determined to be generated, the three-dimensional spatial coordinates of the dominant flow channel are extracted. According to the three-dimensional spatial coordinates, inject the electroresponsive phase change liquid and control the combined electrode module to output DC voltage to trigger local electrolysis, thereby cutting off the dominant flow channel; The feedback based on the dynamic injection impedance switches back to the main repair fluid to drive pulsed extrusion, and after extrusion, the electrical polarity is reversed to trigger reverse depolymerization.

2. The application method according to claim 1, characterized in that, The steps for acquiring fluid data during the injection of the main repair fluid into the fluid injection module and extracting the dynamic injection impedance and impedance time-series change rate include: The pressure and flow data of the fluid injection module located at the orifice are acquired, and the pressure and flow data are smoothed and denoised using a moving average filtering algorithm. The dynamic injection impedance is calculated based on the processed pressure and flow data using fluid dynamics impedance conversion rules. A first-order backward difference model is established based on a preset time window, and the dynamic injection impedance at the current time is compared with that at a historical time to calculate the impedance time-series change rate.

3. The application method according to claim 1, characterized in that, The steps for generating a three-dimensional charge rate matrix of the soil layer and calculating the spatial distribution variance based on the potential difference data sequence of the combined electrode module include: The data sequence of secondary potential difference on the ground surface generated by the combined electrode module transmitting alternating weak current into the ground under alternating weak current detection mode is collected. The potential difference data sequence is solved iteratively using a three-dimensional frequency domain inversion algorithm to generate the three-dimensional charge rate matrix of the soil layer. The charge rate feature values ​​of each discrete grid node in the three-dimensional charge rate matrix of the soil layer are extracted, scalar dimensionality reduction is performed, and the arithmetic mean of the charge rate is obtained by accumulating and averaging. A statistical spatial variance model is constructed to calculate the spatial distribution variance.

4. The application method according to claim 1, characterized in that, Combining the impedance time-series change rate with the spatial distribution variance to perform intersection determination, when determining the occurrence of a dominant flow channel, the step of extracting the three-dimensional spatial coordinates of the dominant flow channel includes: Compare the impedance time-series change rate with a first threshold, and compare the spatial distribution variance with a second threshold; When the impedance time-series change rate is lower than the first threshold and the spatial distribution variance is higher than the second threshold, it is determined that the dominant flow channel is generated. Grid nodes with charge rate eigenvalues ​​greater than a preset polarization anomaly threshold in the three-dimensional charge rate matrix of the soil layer are selected as dominant flow nodes, and the three-dimensional spatial coordinate set corresponding to all the dominant flow nodes is extracted to define the spatial location of the dominant flow channel.

5. The application method according to claim 1, characterized in that, The steps of injecting the electrically responsive phase change fluid according to the three-dimensional spatial coordinates include: The fluid injection module is controlled to cut off the main repair fluid and select the supply circuit of the electro-responsive phase change fluid; Calculate the arithmetic mean of the coordinate components of all the three-dimensional spatial coordinates in the depth direction as the target dominant flow depth, and convert the physical length of the injection pipeline according to the target dominant flow depth; By combining the pipeline inner diameter geometric parameters and the instantaneous output displacement of the variable frequency pump, the fluid injection time delay required for the electroresponsive phase change fluid to reach the target dominant flow depth is calculated.

6. The application method according to claim 5, characterized in that, The step of controlling the output DC voltage of the combined electrode module to trigger local electrolysis to cut off the dominant flow channel includes: After the system has passed the fluid injection time delay, the combined electrode module is instructed to switch to DC high field excitation mode; Calculate the projection azimuth angle of the geometric center point of the three-dimensional spatial coordinates relative to the central deep hole electrode, select the array electrode with the corresponding azimuth as the anode, and simultaneously set the central deep hole electrode as the cathode. Direct current is output between the cathode and the anode. An alkaline environment is generated by the water electrolysis reaction on the surface of the cathode, which triggers the physical phase change of the electro-responsive phase change liquid filled in the dominant flow channel to transform into a solid hydrogel to cut off the pores.

7. The application method according to claim 1, characterized in that, The step of switching back to the bulk repair fluid based on the feedback of the dynamic injection impedance to drive pulsed exudation includes: Extract the background soil reference impedance when no seepage escape failure occurs, and superimpose the system tolerance coefficient to construct the judgment interval characterizing the completion of flow field resistance reconstruction; After detecting that the dynamic injection impedance rises and stabilizes within the determination range, the fluid injection module is controlled to switch back to the main repair fluid supply circuit. A control command combining a constant baseline discharge rate and a sinusoidal high-frequency disturbance is sent to the fluid injection module to drive the main repair fluid to radially squeeze and infiltrate into the surrounding low-permeability soil layer.

8. The application method according to claim 6, characterized in that, The steps of reversing the electrical polarity after infiltration to trigger reverse depolymerization include: After the repair operation reaches the preset injection volume, the combined electrode module is controlled to switch from positive DC to reverse DC mode. The electrical polarity of the central deep hole electrode is reversed to that of the anode to induce an oxygen evolution and acid production reaction, thereby reducing the alkalinity of the local environment. The generated hydrogen ions trigger the deprotonation and depolymerization of the solid hydrogel network, restoring it to a flowing sol state.

9. The application method according to claim 1, characterized in that, The drug application method also includes a precursor prediction step: The dynamic injection impedance, impedance temporal change rate, and spatial distribution variance within the historical time window are collected in parallel to form a three-dimensional input tensor. The three-dimensional input tensor is subjected to maximum and minimum value normalization to eliminate dimensional differences; The normalized three-dimensional input tensor is fed into a long short-term memory neural network model for spatiotemporal correlation calculation, and the predicted spatial distribution variance of charge rate within a future set time span is output.

10. The method of applying the drug according to claim 9, characterized in that, The precursor prediction step also includes: Compare the predicted charge rate spatial distribution variance with the tolerance warning threshold; When it is determined that the variance of the predicted charge rate spatial distribution exceeds the standard, historical data is called to perform node difference calculation to obtain the charge rate increment of each grid node; The grid node combinations whose charge rate increment is greater than a preset precursor increment threshold are selected to form potential dominant flow channels; The three-dimensional spatial coordinates of the potential dominant flow channels are extracted to trigger the injection and local electrolytic cutoff process of the electroresponsive phase change fluid in advance.

Images